The A.I.

Cruise Line
by

ANDREW W. PEARSON

i
Copyright © 2020 Andrew Pearson
All rights reserved
ISBN- 13: 9798647285706

ii
 CONTENTS
ACKNOWLEDGMENT..............................................................................vii
PREFACE ..................................................................................................... ix
CHAPTER ONE: THE SET UP .................................................................. 1
Overview.............................................................................................................. 1
Why AI? ................................................................................................................ 8
Behavioral + Experiential Marketing .................................................... 13
BI + CRM + CX + IoT ..................................................................................... 20
Personalization .............................................................................................. 26
Going Social ..................................................................................................... 30
Capacity Management ................................................................................. 34
The Future ....................................................................................................... 37
CHAPTER TWO: PERSONALIZATION............................................... 45
Overview........................................................................................................... 45
Customer Relationship Management ................................................... 58
Segmentation ............................................................................................................ 64
Customer Lifecycle ....................................................................................... 74
Customer Loyalty .......................................................................................... 82
Cruise Line Engagement and Loyalty Platform ......................................... 86
Natural Language Processing .................................................................. 92
Creating a market for feedback......................................................................... 94
Reducing Search Friction with A.I. .................................................................. 96
Conclusion........................................................................................................ 98
CHAPTER THREE: ANALYTICS......................................................... 103
Overview......................................................................................................... 103
Data Mining ................................................................................................... 113
Artificial Intelligence & Machine Learning ....................................... 117
7 Patterns of AI ....................................................................................................... 121
Cruise Line Analytics ................................................................................. 125
Analytical Methods ............................................................................................... 130
Decision Trees ................................................................................................... 130
k-Means Cluster ................................................................................................ 132
k-Nearest Neighbors ....................................................................................... 134
Logistic Regression.......................................................................................... 136
A/B Testing ......................................................................................................... 138
Time Series Model............................................................................................ 139
Neural Networks .............................................................................................. 140
Discriminant Analysis .................................................................................... 143
Survival or Duration Analysis ..................................................................... 144
Cruise line Analytical Models ................................................................. 145
Customer Segmentation ..................................................................................... 145
iii
 ANDREW W. PEARSON

Customer Acquisition Model ........................................................................... 146

Recency-Frequency-Monetary (RFM) Models ......................................... 147
Propensity to Respond Model ......................................................................... 148
Customer Conversion Model ........................................................................... 149
Identify When a Patron is Likely to Return ............................................... 150
Patron Worth Model ............................................................................................ 150
Customer Churn Model ...................................................................................... 152
Optimizing Offers .................................................................................................. 163
Chronological View of a Cruise Line Analytics Implementation ...... 164
Clickstream Analysis ................................................................................. 166
Location Analytics ...................................................................................... 169
Conclusion ..................................................................................................... 171
CHAPTER FOUR: MARKETING......................................................... 177
Overview ........................................................................................................ 177
Data & Marketing........................................................................................ 181
Analytics ................................................................................................................... 183
Mobile Marketing ....................................................................................... 186
Digital Interactive Marketing: The Five Paradigms ............................... 189
Website Morphing...................................................................................... 192
Psychology of Personalization .............................................................. 198
Social Proof .............................................................................................................. 207
Psychometrics.............................................................................................. 212
Facial Recognition ...................................................................................... 217
Affective Computing .................................................................................. 220
SEO ................................................................................................................... 227
Content Intelligence .................................................................................. 235
Measurement ............................................................................................... 240
Geofencing Applications .......................................................................... 242
Going Viral ..................................................................................................... 248
Conclusion ..................................................................................................... 256
CHAPTER FIVE: SOCIAL A.I. .............................................................. 269
Overview ........................................................................................................ 269
Social Media Analytics .............................................................................. 272
Sentiment Analysis .................................................................................... 280
Sentiment Analysis Tools .................................................................................. 286
Social Media Monitoring .......................................................................... 291
Social Media Marketing ............................................................................ 295
Conclusion ..................................................................................................... 301
CHAPTER SIX: THE SMART CRUISE LINE..................................... 305
Overview ........................................................................................................ 305
Data Governance......................................................................................... 308
Capacity and Demand Management ................................................... 315

iv
 THE A.I. CRUISE LINE

Cost of Outages ....................................................................................................... 336

Unused Capacity .................................................................................................... 346
Streaming Analytics ................................................................................... 349
Augmented and Virtual Reality ............................................................. 354
Internet of Things (IoT)............................................................................ 358
Wearables ................................................................................................................. 362
Robotics .......................................................................................................... 363
CHAPTER SEVEN: THE A.I. CRUISE LINE....................................... 377
Overview......................................................................................................... 377
Text ................................................................................................................... 379
Chatbots .................................................................................................................... 381
Chatfuel ................................................................................................................. 381
Botsify ................................................................................................................... 382
Flow XO ................................................................................................................. 382
Motion.ai .............................................................................................................. 382
Chatty People ..................................................................................................... 382
QnA bot ................................................................................................................. 383
Recast.ai................................................................................................................ 383
BotKit ..................................................................................................................... 383
ChatterOn ............................................................................................................. 383
Octane AI .............................................................................................................. 384
Converse AI ......................................................................................................... 384
GupShup ............................................................................................................... 384
Augmented Search ................................................................................................ 385
Image................................................................................................................ 386
Facial Recognition ................................................................................................. 386
Image Search ........................................................................................................... 389
Video ................................................................................................................ 392
Audio ................................................................................................................ 395
Voice Activated Internet .................................................................................... 396
Voice Search ............................................................................................................ 397
Programmatic Advertising ...................................................................... 400
The Customer Journey .............................................................................. 403
Listening .................................................................................................................... 404
Rules Engine ............................................................................................................ 412
Automation .............................................................................................................. 413
Moderation............................................................................................................... 413
Messaging ................................................................................................................. 417
Data & Analytics..................................................................................................... 421
The Future is Now ...................................................................................... 423
ABOUT THE AUTHOR ......................................................................... 435
INDEX ....................................................................................................... 437

v
ANDREW W. PEARSON

vi
 THE A.I. CRUISE LINE

ACKNOWLEDGMENT

Shinley Uy – for still keeping the faith

vii
ANDREW W. PEARSON

viii
 THE A.I. CRUISE LINE

PREFACE
Under a different name, i.e., The A.I. Cruise Line, this book could have been the
second edition in that predictive series, but I decided to start an entirely new
series centered around A.I. because this technology is currently taking the
business and retail world by storm. I felt the topic was so in the zeitgeist that it
deserved a book all its own. I thoroughly believe a book entirely devoted to how
A.I., ML, and deep learning affects the retail and ecommerce industry would be
instructive and extremely useful for anyone working in the retail industry today,
as well as many who are just interested in how it functions.
I envision The A.I. Cruise Line to be a series of books. Since this book is mostly
about technology and technology, literally, changes by the second, there will
surely be another soon. Subjects such as facial recognition, emotional
recognition, the psychology of personalization, content intelligence, capacity
management, Robotic Process Automation, and programmatic advertising,
amongst others, would be a nice addition to some of the topics I wrote about in
The Predictive Cruise line. I’ve even added a section on marketing virality, which
should be helpful for advertisers who want to tap into this type of exponential
marketing. It’s not easy, but there are steps to increase one’s chances of going
viral.
Parts of this book have been written on the Macau-Taipa bus, the Macau-Hong
Kong ferry, 30,000 feet above both the Pacific and the Dark Continent of Africa,
on the Guangzhou-Macau high-speed rail line, on Jeju island off the Korean
Peninsula, in a café in Siem Reap, on the Hong Kong-Zhuhai-Macau bridge, within
a taxi stuck in the smog-choked streets of Manila, in a Dubai Denny’s, as well as
on the London Underground, to say nothing of all the hotel and motel rooms I’ve
been scribing in; planes, trains, automobiles, tuk tuks, and ferries, too.
If you find this book instructive, please keep up to date with my latest work on
social media. A list of sites can be found on my author page at the back of the
book. In 2020, I will take a more active social media role and will be tweeting,
Facebooking, Instagramming, YouTubing, Slidesharing, Prezing (if that’s a word)
the latest AI, BI, CX, omnichannel, analytics, IoT, social media, etc., developments
worthy of discussion, so feel free to reach out to me on any of those channels.

ix
ANDREW W. PEARSON

x
 CHAPTER ONE: THE SET UP

Overview
Today, it can safely be said, we are living in the age of AI. Everywhere you look,
companies are touting their most recent AI, machine learning (ML), and deep
learning breakthroughs, even when they are far short of anything that could be
deemed a “breakthrough.” “AI” has probably eclipsed “Blockchain”, “Crypto”,
and/or “ICO” as the buzzword of the day. Indeed, one of the best ways to raise
VC funding these days is to stick ‘AI’ or ‘ML’ at the front of your prospectus and
a “.ai” at the end of your website.
Separating AI fact from fiction is one of the main goals of this book; the other is
to help retail and etail executives understand AI so that they can utilize this
groundbreaking technology in ways that are simple and complex and, hopefully,
rather ingenious. Most importantly, in ways that will be recognized by their
customers as something that improves the overall customer, employee, and
vendor experience.
According to IBM1, 2.5 quintillion bytes of data are created each day. That is 10
to the power of 18 and it is a number that is growing exponentially each year;
90% of the world’s data was created over the past two years and data creation
is certainly not going to slow down any time soon. This data, which has been
dubbed “Big Data”, comes from everywhere; our daily financial transactions; our
personal online shopping history; our social media uploads; our mobile
downloads, our advertising data, even, in some cases, IoT sensor data coming off
machines and people.
The social nature of sharing personal content with family, friends and associates
may be the driver behind this growth and it is a growth that several studies2 3
suggest will soon outpace revenue generated by commercial media, such as
music downloads, video clips, and games. This is the kind of growth that a cruise
line operator ignores at its own peril, but when a cruise line delves into this Big
Data world, it needs to ensure that what it’s opening up is a treasure trove of
useful information, not a Pandora’s Box full of data pain.
Major tech companies have embraced AI and machine learning as if it was one
of the most important discoveries ever invented; Google, whose CEO compares
AI to the discovery of fire and electricity, is now an “AI-first” company; Amazon’s
entire business is shaped by AI, from its customer personalization, to its
warehousing, robotics and logistics capabilities, to its voice-activated smart
speakers; IBM has Watson; Facebook has AI and ML algorithms that test out

1
 ANDREW W. PEARSON

which of its AI and ML algorithms are most effective and should be rolled-out
company-wide; Adobe, a big player in the multi-channel marketing space, runs
much of its Experience Cloud marketing platform through its Sensei AI product;
even the analytics powerhouse SAS has recently announced4 that it will spend
US $1B over the next three years on AI software and initiatives. Even some of
the smaller vendors we are partnered with at Intelligencia have embraced AI for
fear of missing out on this exploding market.
In its report Sizing the prize. What’s the real value of AI for your business and
how you can capitalise5, PWC believes that, “AI could contribute up to $15.7
trillion to the global economy in 2030, more than the current output of China
and India combined. Of this, $6.6 trillion is likely to come from increased
productivity and $9.1 trillion is likely to come from consumption-side effects.”
Because AI is still in its infancy, PWC believes that there are opportunities for
emerging markets to leapfrog more developed counterparts with AI as the
engine.5 Although this is a possibility, the inherent requirements of AI – a highly
educated workforce, strong backing from higher learning institutes, a strong
legal and regulatory framework, and access to huge sources of data — might
limit emerging market successes.5 However, this shouldn’t make companies in
the industrial world too comfy. PWC is probably onto something when it claims
that, “within your business sector, one of today’s start-ups or a business that
hasn’t even been founded yet could be the market leader in ten years’ time.”5 AI
threatens on both the micro and macro front, which is rare in a technology.
According to PWC’s analysis, “global GDP will be up to 14% higher in 2030 as a
result of the accelerating development and take-up of AI–the equivalent of an
additional $15.7 trillion.”5 For PWC, the economic impact of AI will be driven by5:
1. Productivity gains from businesses automating processes (including the
use of robots and autonomous vehicles).
2. Productivity gains from businesses augmenting their existing labour
force with AI technologies (assisted and augmented intelligence).
3. Increased consumer demand resulting from the availability of
personalised and/or higher-quality AI-enhanced products and services.5
In her article Gartner Top 10 Strategic Technology Trends for 2018 6, Kasey
Panetta argues that AI was the number one trend for 2018. Panetta states that,
“The ability to use AI to enhance decision making, reinvent business models and
ecosystems, and remake the customer experience will drive the payoff for digital
initiatives through 2025.”6 The other nine technological trends were: intelligent
apps and analytics, intelligent things, digital twins, cloud to the edge,
conversational platforms, immersive experience, blockchain, event-driven, and
continuous adaptive risk and trust.6
“Given the steady increase in inquiry calls, it’s clear that interest is growing. A
recent Gartner survey showed that 59% of organizations are still gathering

2
 THE A.I. CRUISE LINE

information to build their AI strategies, while the remainder have already made
progress in piloting or adopting AI solutions,” states Panetta.6 If your competitor
isn’t already looking into AI, they probably are about to and, if you aren’t yet
looking into AI, then it might already be too late.
“Although using AI correctly will result in a big digital business payoff, the
promise (and pitfalls) of general AI where systems magically perform any
intellectual task that a human can do and dynamically learn much as humans do
is speculative at best,” contends Panetta.6 “Enterprises should focus on business
results enabled by applications that exploit narrow AI technologies and leave
general AI to the researchers and science fiction writers,” says David Cearley, a
Gartner vice president and Gartner Fellow.6
Panetta claims that, “Over the next few years every app, application and service
will incorporate AI at some level. AI will run unobtrusively in the background of
many familiar application categories while giving rise to entirely new ones.”6
Cearley recommends companies challenge their packaged software and service
providers “to outline how they’ll be using AI to add business value in new
versions in the form of advanced analytics, intelligent processes and advanced
user experiences.”6 This is very sound advice.
Panetta says that, “Intelligent apps also create a new intelligent intermediary
layer between people and systems and have the potential to transform the
nature of work and the structure of the workplace, as seen in virtual customer
assistants and enterprise advisors and assistants.” 6
Cearley recommends that businesses “Explore intelligent apps as a way of
augmenting human activity, and not simply as a way of replacing people”6
“Augmented analytics is a particularly strategic growing area that uses machine
learning for automating data preparation, insight discovery and insight sharing
for a broad range of business users, operational workers and citizen data
scientists,” adds Panetta.6
“Intelligent things use AI and machine learning to interact in a more intelligent
way with people and surroundings,”6 says Panetta, adding, “Some intelligent
things wouldn’t exist without AI, but others are existing things (i.e., a camera)
that AI makes intelligent (i.e., a smart camera).”6 “These things operate semi-
autonomously or autonomously in an unsupervised environment for a set
amount of time to complete a particular task,” explains Panetta.6 Panetta sees a
shift from “stand-alone intelligent things to a swarm of collaborative intelligent
things.”6 In this model, she explains, multiple devices collaborate together,
either with or without human input.
“A digital twin is a digital representation of a real-world entity or system,”
explains Panetta.6 “In the context of IoT, digital twins are linked to real-world
objects and offer information on the state of the counterparts, respond to
changes, improve operations and add value.”6 With an estimated 21 billion

3
 ANDREW W. PEARSON

connected sensors and endpoints in operation by 2020, “digital twins will exist
for billions of things in the near future.” 6 Billions of dollars in cost savings from
maintenance repair and operation as well as optimized IoT asset performance
are on the table, claims Cearley.6
Digital twins offer immediate help with asset management. 6 As these systems
collect enormous amounts of data, they will “offer value in operational efficiency
and insights into how products are used and how they can be improved,” 6 which
should prove beneficial not just to the operators but also the companies that
build the products that the data is being collected upon.
“Over time, digital representations of virtually every aspect of our world will be
connected dynamically with their real-world counterparts and with one another
and infused with AI-based capabilities to enable advanced simulation, operation
and analysis,” says Cearley.6 “City planners, digital marketers, healthcare
professionals and industrial planners will all benefit from this long-term shift to
the integrated digital twin world,” he adds.6
Gartner’s fifth trend–Edge computing–describes “a computing topology in which
information processing and content collection and delivery are placed closer to
the sources of this information.”6 Panette argues that enterprises should begin
using edge design patterns in their infrastructure architectures because
“Connectivity and latency challenges, bandwidth constraints and greater
functionality embedded at the edge favors distributed models.” 6
Cloud and edge computing should not be seen as competing approaches, argues
Panette.6 According to her, “Edge computing speaks to a computing topology
that places content, computing and processing closer to the user/things or ‘edge’
of the network. Cloud is a system where technology services are delivered using
internet technologies, but it does not dictate centralized or decentralized service
delivering services.”6 “When implemented together, cloud is used to create the
service-oriented model and edge computing offers a delivery style that allows
for executions of disconnected aspects of cloud service,” concludes Panette.6
According to Panette, conversational platforms are currently capable of
answering simple questions like “How’s the weather?”, or handling more
complicated interactions, like booking a reservation at the Italian restaurant on
Parker Avenue, but their capacity is severely limited.6 That will soon change.
“These platforms will continue to evolve to even more complex actions, such as
collecting oral testimony from crime witnesses and acting on that information
by creating a sketch of the suspect’s face based on the testimony,” argues
Panette.6
The challenge that conversational platforms face today “is that users must
communicate in a very structured way, and this is often a frustrating
experience,”6 as anyone who has dealt with the limited range of a Facebook bot
understands. “A primary differentiator among conversational platforms will be

4
 THE A.I. CRUISE LINE

the robustness of their conversational models and the API and event models
used to access, invoke and orchestrate third-party services to deliver complex
outcomes,” contends Panette.6 In chapter seven, I explain many of the chatbot
platforms that are currently available to cruise lines that contain some highly
sophisticated AI capabilities.
Augmented reality (AR), virtual reality (VR) and mixed reality (MR) will combine
with conversational platforms to create a user experience reliant on an invisible
and immersive experience layer provided by application vendors, system
software vendors and development platform vendors, who will all compete to
deliver unique and groundbreaking solutions.6
Panette argues that, “Over the next five years the focus will be on mixed reality,
which is emerging as the immersive experience of choice, where the user
interacts with digital and real-world objects while maintaining a presence in the
physical world.”6 Mixed reality exists along a spectrum and includes head-
mounted displays (HMD) for AR or VR, as well as smartphone- and tablet-based
AR. “Given the ubiquity of mobile devices, Apple’s release of ARkit and iPhone X,
Google’s Tango and ARCore, and the availability of cross-platform AR software
development kits such as Wikitude,”6 Gartner saw 2018 as a time when the
battle for smartphone-based AR and MR really caught fire.6
Today’s cruise lines have to be conscious of the complex security environment
that their business exists in. The number of hacks and corporate IT hostage
taking that is already occurring will only increase as technology increases in
sophistication.6
In their MIT Sloan Management Review article Winning with AI7, Sam
Ramsbotham et al. surveyed more than 2,500 executives and had 17 interviews
with leading experts in the hope of providing “a data-driven view of what
organizations that succeed with AI are doing and what real success with AI looks
like.”7 The writers found that, “Companies that capture value from their AI
activities exhibit a distinct set of organizational behaviors.”7 These businesses:
• Integrate their AI strategies with their overall business strategy.
• Take on large, often risky, AI efforts that prioritize revenue growth over
cost reduction.
• “Align the production of AI with the consumption of AI, through
thoughtful alignment of business owners, process owners, and AI
expertise to ensure that they adopt AI solutions effectively and
pervasively.”7
• Unify their AI initiatives with their larger business transformation
efforts.
• “Invest in AI talent, data, and process change in addition to (and often
more so than) AI technology. They recognize AI is not all about
technology.”7

5
 ANDREW W. PEARSON

“The net effects of these behaviors, and their underlying commitments, are to
address difficulties generating value with AI, manage unavoidable competitive
and implementation risks from AI, and effectively exploit AI-related
opportunities,” argue Ramsbotham et al.7
Ramsbotham et al. discovered that, “To a large extent, difficulties with
generating value from AI show up in the data as organizational rather than
technological,”7 which isn’t all that surprising. “Companies that focus solely on
the production of AI (data, technology, tools) are less likely to derive value than
those companies that actively align business owners, process owners, and AI
owners. Leaders enable their organizations to consume AI as much as to
produce AI,” contend Ramsbotham et al.7 “Companies that treat AI as a
‘technology thing’ struggle to deliver value: An IT focus on AI tends to generate
less value than a broad strategic focus,” note Ramsbotham et al.7 Merely
developing a strategy for AI is not enough, Ramsbotham et al.’s research found.7
Tying a strategy for AI to the company’s overall corporate strategy was
essential.7
“Aligning AI and strategy requires organizations to look backward from strategy,
not forward from AI,” maintain Ramsbotham et al.7 “Jeroen Tas, Royal Philips’
chief innovation and strategy officer, explains that AI is integrated into corporate
strategy by working ‘our way backward’ from the company’s overall strategy for
customer health,” explain Ramsbotham et al.7 Philips “then identifies how AI can
support this. Philips isn’t starting with AI and looking forward to where it can
support the strategy; rather, it finds areas in which the strategy needs support
and looks for the best way to provide it.”7 “Philips focuses specifically on how AI
can provide better consumer experiences, better health outcomes, improved
care provider experience, and lower costs of care,” note Ramsbotham et al.7
“Those companies that obtain business value from AI build internal teams and
rely less on outside vendors; they selectively import experienced AI talent for
technical leadership roles; and they upskill their existing workforce to enable AI
literacy and understanding of how to manage with AI,” say Ramsbotham et al.7
“Despite talent scarcity, companies of all sizes across industries report similarly
positive outcomes when they make these three talent investments,” the writers
state.7 Although acquiring talent is a big issue, AI can assist in automating certain
repetitive tasks, especially for creatives. Services like Kaggle can be useful to
outsource model building, and AI can be particularly helpful for a company’s
CRM and marketing departments. I will provide more details about these
opportunities in the ensuing chapters.
Ramsbotham et al.’s research found two broad ways that companies were
managing risks that emerged either directly or indirectly from theirs and others’
AI deployments.7 “First, companies that have obtained value from AI are more
likely to manage proactively: They make bigger, sometimes riskier, investments.
These aren’t gambles, however, but rather, calculated strategy,” explain

6
 THE A.I. CRUISE LINE

Ramsbotham et al.7 They also add, “Second, in fast-moving market

environments, strategic alignment becomes more challenging and more critical
to get right. Misalignment, accordingly, becomes a greater and more common
risk.”7
“Companies that derive value from AI are more likely to integrate their AI
strategy with their overall corporate strategy,” argue Ramsbotham et al.7
“Organizations that are most effective at obtaining value from AI more likely
generate value from AI-driven revenue, rather than from cost savings alone,”
note Ramsbotham et al.7 They add: “Most executives believe that the highest
future value from AI will be on the revenue and growth side rather than on the
cost side.”7
“Genuine success with AI — over time — depends on generating revenue,
reimagining organizational alignment, and investing in the organization’s ability
to actually use AI across the enterprise,” conclude Ramsbotham et al.7
Shivaji Dasgupta, managing director of Deutsche Bank, “notes a growing concern
that there will no longer be ‘a level playing field’ thanks to AI, especially in a
highly regulated industry like banking.”7 Ramsbotham et al. point out that, “New
competitors from industries not bound by rules imposed on incumbents are
already creating competitive risks.”7 Ramsbotham et al. see Apple with its Apple
Pay and recently launched Apple Card and Amazon with its Amazon Cash as two
new entrants into the financial sector that could prove daunting competitors to
normal banks.7 “With massive amounts of data, the ability to apply AI and other
technologies to capitalize on that data, and their loyal customer bases, the tech
giants’ respective moves into financial services pose formidable threats to
traditional banking and financial services companies,” argue Ramsbotham et al.7
“Deutsche Bank’s Dasgupta describes an impressive revenue-side achievement:
For one consumer credit product in Germany, AI makes a real-time decision on
whether or not to extend a loan to a customer as the customer is filling out the
loan application,” explains Ramsbotham et al.7 “This has generated a lot of
interest among consumers,” Dasgupta reports, so much so that, for that specific
product, loan issuance shot up 10- to 15-fold in eight months following the
service’s launch.7 Dasgupta believes it is because, in Germany, an individual’s
credit rating is damaged by applying and not receiving a loan.7 “Deutsche Bank’s
new AI solution removes that risk for customers by telling them whether or not
they will be approved for a given amount before they hit ‘apply,’” explains
Ramsbotham et al.7 “The largest gains haven’t come from better serving those
customers who would have applied for loans anyway,” say Ramsbotham et al.7
“Rather, the benefit comes from reaching those who would not have applied in
the first place.”7 This example has a direct relevance to the retail industry, which
is often extending credit to it customers.

7
 ANDREW W. PEARSON

Why AI?
So why should a cruise line choose to go down the complex AI road? Well, in
their article Artificial intelligence Unlocks the True Power of Analytics8, Adobe
explains the vast difference between doing things in a rules-based analytics way
and an AI-powered way, including:
• Provide warnings whenever a company activity falls outside the norm.
The difference:
o Rules-based analytics: You set a threshold for activity (e.g.,
“200–275 orders per hour”) and then manually investigate
whether each alert is important.
o AI-powered analytics: The AI analytics tool automatically
determines that the event is worthy of an alert, then fires it off
unaided.
• Conduct a root cause analysis and recommend action. The difference:
o Rules-based analytics: You manually investigate why an event
may have happened and consider possible actions.
o AI-powered analytics: Your tool automatically evaluates what
factors contributed to the event and suggests a cause and an
action.
• Evaluate campaign effectiveness:
o Rules-based analytics: The business manually sets rules and
weights to attribute the value of each touch that led to a
conversion.
o AI-powered analytics: The AI analytics tool automatically
weights and reports the factors that led to each successful
outcome and attributes credit to each campaign element or
step accordingly.
• Identify customers who are at risk of defecting:
o Rules-based analytics: You manually study reports on groups
of customers that have defected and try to see patterns.
o AI-powered analytics: Your tool automatically identifies which
segments are at greatest risk of defection.
• Select segments that will be the most responsive to upcoming
campaigns:
o Rules-based analytics: You manually consider and hypothesize
about the attributes of customers that might prove to be
predictive of their response.
o AI-powered analytics: Your tool automatically creates
segments based on attributes that currently drive the desired
response.
• Find your best customers:

8
 THE A.I. CRUISE LINE

o Rules-based analytics: You manually analyze segments to

understand what makes high-quality customers different.
o AI-powered analytics: Your tool automatically identifies
statistically significant attributes that high-performing
customers have in common and creates segments that include
these customers for the business to act on.
Beyond the reasons listed above, this book will explain how AI can be used in
website morphing, customer and media recommendations, purchase
prediction, demand forecasting, programmatic advertising, as well as social
listening.

GENERAL USE CASE INDUSTRY

Sound
Voice recognition UX/UI, Automotive, Security, IoT
Voice search Handset maker, Telecoms
Sentiment analysis CRM for most industries
Flaw detection Automotive, Aviation
Fraud detection Finance, Credit cards
Time Series
Log analysis/Risk detection Data centers, Security, Finance
Enterprise resource planning Manufacturing, Auto, Supply Chain
IoT, Smart home, Hardware
Predictive analytics using sensor data
manufacturing
Business and Economic analytics Finance, Accounting, Government
Recommendation engine E-Commerce, Media, Social Networks
Text
Sentiment analysis CRM, Social Media, Reputation mgmt.
Augmented search, Theme detection Finance
Threat detection Social Media, Government
Fraud detection Insurance, Finance
Image
Facial recognition Multiple industries
Image search Social Media

9
 ANDREW W. PEARSON

GENERAL USE CASE INDUSTRY

Machine vision Automotive, Aviation
Photo clustering Telecom, Handset makers
Video
Motion detection Gaming, UX, UI
Real-time threat detection Security, Airports

Table 1: A.I. use cases

Source: skymind9

Table 1 shows the general use cases for AI broken down by industry. This is a
general list and many of these use cases can be utilized by industries other than
the ones specified, including the cruise line industry.
Terry Gou, the chairman of the Taiwanese electronics manufacturer Foxconn,
says the company plans to replace 80 percent of its workforce with robots in the
next five to 10 years.10 Richard Liu, the founder and CEO of Chinese e-commerce
company JD.com, hopes to one day have a completely automated company —
100% operated by AI and robots — and the company has invested $4.5 billion to
build an AI center in Guangdong, China, to implement such a strategy.11
Automation doesn’t just have to be about robots and factories; however, it can
also remove the day-to-day drudgery work that humans spend so much time
doing. As I will explain later in the book, there are AI tools out there that can
automate away repetitive processes, like cataloging images or video and let
human do what humans do best — create.
Google Duplex has shown that AI bots can do things like make reservations at
hair salons12 and this is one of deep learning’s futures. Cruise lines need to
develop voice and speech understanding technology or they risk being left
behind by their competitors. In particular, voice is a technology waiting for mass
use. We communicate through voice as much as any other sense and the
companies that win the battle in voice will win the battle for the 21 st century
consumer. One of the most important stats about voice is the fact that voice
searches on Google are now 30 times more likely than text searches to be action
queries.13
74% of adults use their smart phones to get directions and other information
based on their current location.14 In addition, 30% of adults who have a social
media account say that at least one of those accounts will include their current
location when they post.15 Even if we ignore hand held devices, as many as 96%
of cars produced in 2013 are built with event recorders that include GPS. 16
Bonchi and Wang give the troubling example of a politician in Germany who
“went to court to find out how often and what kind of data was collected by his

10
 THE A.I. CRUISE LINE

cell phone company. The results were staggering. In just a six-month period, his
coordinates were recorded and stored over 35,000 times.”17
In a March 2017 note to clients18, RBC Capital argued that Amazon's voice
assistant Alexa “could bring the U.S. e-commerce giant $10 billion of revenues
by 2020 and be a ‘mega-hit.’” According to Arjun Kharpal, “The investment bank
has dubbed the technology ‘voice-activated internet (VAI)’ and said it represents
a ‘material opportunity’ for both Amazon and Google, which has its own
technology called Google Assistant.”18
RBC breaks down the impressive numbers as follows18:
• Alexa device sales, which could reach $60 million by 2020.
• Voice driven shopping sales, which could reach $400 per customer by
2020.
• Platform revenues: If Amazon reaches over 100 million installed Alexa
devices then it could create an app store and tap into “platform
revenue.”
• Amazon Web Services (AWS) tailwind.18
RBC Capital notes that, “As the number of skills rises, Amazon will create a
marketplace that will allow them to charge companies to appear more
prominently in its app store.”18 Paid skills on Alexa could be lucrative and
Amazon could collect revenue sharing payments accordingly.
Of course, Amazon is not alone in the voice activated internet (VAI) market. 18
Google has its own voice assistant built into Android smartphones and its own
smart speaker called Google Home.18 According to Kharpal, “RBC was surprised
by the popularity of Google Home since it was only launched in October 2016 in
the U.S.”18 Apple and Microsoft are also highly involved in this space. Siri hasn’t
been the success Apple hoped it would be, but they still believe it has an integral
part of their VAI future.
“Awareness of Google Home among 1,748 Amazon customers surveyed by RBC
was 60 percent. Whereas when RBC did a similar survey in September 2015, just
shortly after the Echo had launched widely in the U.S., only 33 percent of
respondents had heard of Alexa. Google however still only has around 80
Actions, which are like Alexa's skills, which total above 10,000,” notes Kharpal.18
All-in-all, The A.I. Cruise line should not concern itself with who might win the
VAI battle as any of these platforms can bring considerable eyeballs to the table
(or the store). However, any money spent to reach these eyeballs should
produce a healthy ROI. The app market that Amazon could build atop Alexa could
also be a hidden opportunity for creative cruise line marketers.
In chapter two, I discuss Milgrom and Tadelis’ study19 that utilizes natural
language processing (NLP) to create an environment that promotes trust,
similar to the way institutions emerged in the medieval trade fairs of Europe

11
 ANDREW W. PEARSON

that helped foster trust amongst traders.19 Milgrom and Tadelis believe that AI
can be applied to today’s marketplaces like eBay to help foster a more
trustworthy and better buying experience for their customers. 19
One of the major use cases for AI is sentiment analysis, which uses NLP to gain
insight into how a business is seen on social media. According to skymind.ai20:
“Natural language refers to language that is spoken and
written by people, and natural language processing (NLP)
attempts to extract information from the spoken and written
word using algorithms. NLP encompasses active and a [sic]
passive modes: natural language generation (NLG), or the
ability to formulate phrases that humans might emit, and
natural language understanding (NLU), or the ability to build a
comprehension of a phrase, what the words in the phrase refer
to, and its intent. In a conversational system, NLU and NLG
alternate, as algorithms parse and comprehend a natural-
language statement and formulate a satisfactory response to
it.”
Another important area for AI is text analytics. In his article Text Analytics: How
to Analyse and Mine Words and Natural Language in Businesses 21, Bernard Marr
states that, “Text analytics, also known as text mining, is a process of extracting
value from large quantities of unstructured text data.”
Marr elucidates: “While the text itself is structured to make sense to a human
being (i.e., a company report split into sensible sections) it is unstructured from
an analytics perspective because it doesn’t fit neatly into a relational database
or rows and columns of a spreadsheet. Traditionally, the only structured part of
text was the name of the document, the date it was created and who created
it.”21 “Access to huge text data sets and improved technical capability means text
can be analysed to extract high-quality information above and beyond what the
document actually says,” argues Marr.21 “Text can be assessed for commercially
relevant patterns such as an increase or decrease in positive feedback from
customers, or new insights that could lead to product tweaks, etc.” 21 This means
text analytics can help us discover things we didn’t already know but, perhaps
more importantly, had no way of previously knowing.21 These could be incredibly
important insights for a business both about itself and, potentially, about its
competitors.21
For Marr, “Text analytics is particularly useful for information retrieval, pattern
recognition, tagging and annotation, information extraction, sentiment
assessment and predictive analytics.”21 It could both reveal what customers
think about a company’s products or services, or highlight the most common
issues that instigate customer complaints.21

12
 THE A.I. CRUISE LINE

Behavioral + Experiential Marketing

The great advertising executive David Ogilvy once said, “There isn’t any
significant difference between the various brands of whiskey, or cigarettes or
beer. They are all about the same. And so are the cake mixes and the detergents,
and the margarines… The manufacturer who dedicates his advertising to building
the most sharply defined personality for his brand will get the largest share of
the market at the highest profit.”22 This is truer for the retail industry than most
other industries.
Although the great behaviorist B.F. Skinner once said, “The major difference
between rats and humans is that rats learn from experience,” human beings are,
first and foremost, creatures of habit. If a cruise line can understand these habits
on both a micro and macro level, it can both predict what its customers are going
to want, and also what they will do, so a business can, potentially, shape that
behavior. Smart businesses can utilize all this behavior in a predictive way and
optimize multiple parts of their operations, including labor management.
Marketing has always been about influencing people’s behavior and what could
be different here is The A.I. Cruise Line’s ability to understand how one
customer’s actions will affect the company’s entire operation. With this insight
extrapolated over a million customers over 365 days of the year, the brand can
take the most appropriate — and optimized — action to reap the highest profit.
In his article How Real-time Marketing Technology Can Transform Your
Business23, Dan Woods makes an amusing comparison of the differing
environments that today’s marketers face compared to what their 1980s
counterparts did. He says:
“Technology has changed marketing and market research into
something less like golf and more like a multi-player first-
person-shooter game. Crouched behind a hut, the stealthy
marketers, dressed in business-casual camouflage, assess their
weapons for sending outbound messages. Email campaigns,
events, blogging, tweeting, PR, ebooks, white papers, apps,
banner ads, Google Ad Words, social media outreach, search
engine optimization. The brave marketers rise up and blast
away, using weapons not to kill consumers but to attract them
to their sites, to their offers, to their communities. If the
weapons work, you get incoming traffic.”
As behavioral economist Susan Menke explains in her paper Humanizing
Loyalty24, “Decision fatigue and cognitive fatigue are the opposite of flow and
seamlessness. We are making too many decisions that tax our cognitive bank
account. We dole it out on important things and not on things that are already
operating well.”

13
 ANDREW W. PEARSON

In her paper, Menke touches upon the concept of psychological scripts — the
idea that the mind doesn’t have to focus on many day-to-day activities as they
can be handled without much thought. 24 The more seamless a company can
make the customer interaction process, the more likely a customer will continue
to do business with it.24
Tom Fishburne, the founder of Marketoonist, says “the best marketing doesn’t
feel like marketing,” and his words are a good motto for today’s digital marketer.
AI can help make marketing so personalized and wanted that customers enjoy
receiving it and have no qualms positively responding to it. The seamlessness of
the marketing is paramount.
We live in an instant gratification world and the companies that will thrive in this
new environment will be the ones who can both keep up with the requirements
of their discerning and demanding customers as well as predict what these
customers will be wanting throughout their customer journeys. Today,
companies need every advantage they can get so that they can provide better
service than their competitors.
Being able to accurately predict not only who a marketer’s best leads and
prospects are, as well as how and when it is best to engage those leads is nice,
but understanding how their acceptance of these marketing offers will affect the
overall bottom line is what The A.I. Cruise line is all about.
This ability will not only empower marketers and salespeople in the coming years
to be radically more productive and profitable than they are today, but also give
multiple corporate departments visibility on their micro and macro needs. Used
properly, predictive analytics and AI can transform the science of sales
forecasting from a dart-throwing exercise to a precision instrument.
The concept of sales and marketing automation has already produced some of
the highest-flying successes in high-tech. Companies like Salesforce.com have
been wildly successful in automating the sales process for salespeople and sales
managers. Big software vendors like SAP, SAS, IBM, Microsoft, and Oracle are
vying for supremacy, while smaller players like Pega Systems, SugarCRM,
Netsuite, and Sage are offering interesting products at highly affordable prices.
In their article 10 Principles of Modern Marketing25, Ann Lewnes and Kevin Lane
Keller argue that, “Technology has changed everything. Fundamentally, it allows
for new ways to create customer experiences, new mediums to connect with
customers and other constituents, and trillions of data points to understand
customer behavior and the impact of marketing programs and activities. Yet,
with all that progress, we are still only at the tip of the iceberg in terms of the
profound impact technology will have on the future of marketing.”
In their AI: Your behind-the-scenes marketing companion26, the Adobe Sensei
Team claims that, “The battle to win customer hearts and minds is no longer

14
 THE A.I. CRUISE LINE

simply about your product. It’s about the experience. Because that’s what keeps
customers coming back. To compete on experience, you need to understand
what customers want now while anticipating what they’ll do next. And because
your customers have lots of choices, you don’t have a lot of time to get it right.”
“But many times, the knowledge you need to personalize interactions and
compel customers to act is locked up in huge amounts of data,” warns the Adobe
Sensei Team.26 “This means someone has to sift through it all to recognize
patterns, trends, and profiles, so you can quickly act on insights. The problem is,
it’s too much data for humans to sort through alone. That’s where artificial
intelligence and machine learning come in,” the team says.26
“Customers will always expect a human touch in their interactions,” claims the
Adobe Sensei Team.26 “These new technologies won’t replace marketing jobs,
but they will change them,” the team adds.26 Brands should think of AI and
machine learning as their behind-the-scenes marketing assistant who helps
unlock insights in volumes of data, develops a deeper understanding of what
customers want, a forecasting tool that predicts trends, as well as monitors
unusual activity, such as spikes or drops in sales — all while giving brands more
time to make decisions that matter.26
“To fully realize the potential of technology,” Lewnes and Keller argue that, “it
takes transformation across people, processes, and technology. Only by
recognizing all three forces will modern marketers reap the full benefits that
technology can have on marketing transformation.”25
“To thrive in this new era, it is imperative that marketers embrace developments
in technology and test and adopt new advancements that fit their business —
whether AI, or voice, or augmented reality — before they lose a competitive
edge,” claim Lewnes and Keller.25 “At the same time, mastering technology is not
the only criterion for success in the modern marketing era — the right people
and processes must also be put in place to properly develop, manage, and
nurture the benefits of that technology,” they add. 25
“In terms of people, today’s marketers must possess many traits. They must be
curious, flexible, agile, and nimble. They must be willing to be change agents,
always looking around the corner and helping to scale transformation as
champions for change,” say Lewnes and Keller. 25 The status quo no longer works
— continuous development of new skills for all marketers is critically
important.25
Today’s marketing organization needs people with diverse skill sets and
expertise in key areas.25 “Managers should ensure their marketing teams include
members who bring creative and analytical capabilities, as well as individuals
who can play newly evolved roles on a team — whether that’s someone skilled
in web development, data analytics, e-commerce, or new media,” argue Lewnes
and Keller.25 Marketing organizations almost have an impossible task as many of

15
 ANDREW W. PEARSON

the needed skills for these new jobs “didn’t exist four or five years ago, and even
if they did, they have changed dramatically in recent years.” 25
Furthermore, while these new, specialized jobs have emerged, marketers must
keep in mind the broadening marketing ecosystem.25 “The dynamic cross-
channel nature of marketing today requires that campaigns be integrated and
connected across every channel,” explain Lewnes and Keller.25
“Processes must also change for technology organizations. Today, the customer-
decision process is becoming more complex and varied. As the customer journey
becomes increasingly nonlinear, the organization must change to reflect that,”
warn Lewnes and Keller.25 “In a more complex marketplace, internal
organizational lines need to be redrawn. Silos must be broken down and cross-
functional relationships established so that marketing works seamlessly across
other groups in the organization such as IT, finance, sales, and product
management,” say Lewnes and Keller. 25
“Marketing can benefit from the output of these other groups and also
contribute to the groups’ effectiveness and success at the same time,” claim
Lewnes and Keller.25 “For example, to improve the reliability of financial
forecasting, marketing can share early-warning lead indicators that have been
shown to affect bottom-of-the-funnel behaviors and ultimately revenue (for
example, the number of customer visits to company-controlled websites).”25 The
marketing department can show its growing worth and value by demonstrating
“its impact on the business, validating the ROI of every dollar to peer groups in
the organization and becoming a strategic driver of the business.”25
All these changes, however, require that organizations adapt to this new
marketing and technological environment. Lewnes and Keller argue that
marketers “must learn to be agile, take risks, fail fast, and apply lessons.”25 They
must also learn how to get the most out of a data-rich world by testing,
optimizing, and activating.25
Lewnes and Keller claim “experience is the new brand.” 25 They are right,
experience will be one of the big differentiators for companies going forward. 25
“With traditional marketing, the customer-decision and company-selling process
was comparatively simple with customers entering into a company’s sales and
marketing funnel and making various choices along the way to becoming loyal,
repeat customers,” explain Lewnes and Keller.25 Today, every “customer touch
point online and offline — as wide-ranging as a tweet, product download, in-
store purchase, the company’s social purpose, its executives’ behavior, and the
corporate culture — can shape experiences that define a brand for customers,”
contend Lewnes and Keller.25
“Marketers operate at the intersection of many of these customer experiences
and are uniquely positioned to help steer the future directions for brands,” claim
Lewnes and Keller.25 “In doing so, marketers of technology products cannot just

16
 THE A.I. CRUISE LINE

worship the product alone and be transactional in their customer interactions.

They must create full-on, immersive experiences for customers that build strong
ties to the company and the brand as a whole. Experiences are the new
competitive battlefield and a means to create powerful differentiation from
competitors,” argue Lewnes and Keller.25
A new acronym that is making the software rounds these days is CXM —
Customer Experience Management — and it helps businesses collect and
process real-time data from across an organization. A CXM platform activates
content based on customer profiles, allowing personalized experiences to be
delivered in real time. This is the future of both AI and retail brand marketing.
In chapter two, I lay the foundation for how a cruise line can utilize both AI and
ML, while in chapter seven I detail specific examples of how these technologies
can be utilized on several different fronts. Certainly, many cruise lines are
already using these technologies for things like customer segmentation, market
forecasting, recommendation engines, as well as in target marketing, but they
can also be used to weed out problem or addicted gamblers, as well as increase
employee productivity.
The A.I. Cruise Line is a cruise line that takes into account all kinds of data that
can be created at a retail store or on an ecommerce website. Data created by its
employees, vendors, patrons, and customers (we’ll consider these the people
who haven’t signed up for a loyalty card yet and aren’t, therefore, as trackable
as patrons who are in the cruise line’s database) are also quantified and
optimized.
The A.I. Cruise Line utilizes all the data associated with the business to make
better business decisions for the company as a whole. The A.I. Cruise Line is
viewed holistically and the proverbial butterfly’s wing that flaps somewhere
inside the company can set off a chain of events that either helps or hurts the
company’s bottom line, potentially weeks or months down the line; captured
and analyzed the data will be, so that surprises and negative impacts can be
mitigated, if not reversed.
Descriptive analytics, diagnostic analytics, predictive analytics, prescriptive
analytics and the newest field of analytics — edge analytics — are exploited
throughout The A.I. Cruise Line to try to reach as real-time an IT environment as
possible. The data I will be focusing on throughout this book will be culled from
mostly the following sources:
• Operational systems.
• Customer Relationship Management (CRM) or CX software.
• Business Intelligence (BI) systems.
• Transaction data from Point-of-Sales (POS) systems.
• Clickstreams from the cruise line’s website.

17
 ANDREW W. PEARSON

• Call center systems.

• Surveillance and security systems, including facial recognition datasets.
• Emotional recognition datasets.
• Geo-location data from in-house Wi-Fi systems.
• Social media data from WeChat, Facebook, Weibo, Twitter, Jiepang,
Instagram, YouTube, Twitch, and other mobile and social media apps.
• Customer management systems.
• Social media listening hubs.
• Google analytics and web tracking information.
• Programmatic advertising information.
• HR and ERP systems.
All of this information can be fed into an Enterprise Data Warehouse (EDW),
where it can be utilized by a multitude of departments, including call
center/customer service, marketing, social media marketing, patron
management, affiliate marketing, all the way up to the top executive branches,
including individuals in the C-level suite.
In recent years, businesses in general and cruise lines in particular have come to
the realization that data warehouses, while perfectly able to handle the BI and
analytics needs of yesterday, don’t always work in today’s complex IT
environments, which contain structured, unstructured, and semi-structured
data.
Normal relational databases worked fine when business users were restricted to
proprietary databases and the scope of work was confined to canned reports
and modest dashboards that included limited drill down functionality. Today,
however, with the inclusion of so much unstructured data coming from mobile,
social, web logs, etc., and semi-structured data originating from a multitude of
sources, limitations abound. Standard data warehouses require built-in,
understandable schemas, but unstructured data, by definition, doesn’t have a
definable schema that is accessible and understandable in every case.
Today’s IT environment is nothing like the IT environment of even three years
ago. Real-time data management capabilities have brought a whole new level of
data available to customer intelligence, customer interaction, patron
management and social media systems.
One of the biggest challenges for IT departments today is scalability. With the
multitude of data warehouse and cloud offerings available today, cruise lines can
dynamically scale up or down according to their storage needs. Over the past
few years, the cost of storage has plummeted, and virtual servers can be spun
up almost instantaneously, as well as quite inexpensively (relative to the outright
purchase of hardware). With this instant access to data, a whole new world of
real-time interactions can flourish.

18
 THE A.I. CRUISE LINE

The concept of “Edge Analytics” — i.e., the processing of analytics at the point
or very close to the point of data collection — exponentially increases the ability
to use predictive analytics where it can be utilized best — at the point of
interaction between the business and the consumer. In short, edge analytics
brings analytics to the data rather than vice-versa, which, understandably, can
reduce cost and increase its usage as the data is analyzed close to where it can
best be utilized. This also reduces latency, which could be the difference
between useful and useless analytics.
Today, the analytics space is more crowded than ever before; standard ETL-
solution providers are adding analytics to their multitude of offerings. Many new
players in the Master Data Management (MDM) field have BI platforms that
combine integration, preparation, analytics and visualization capabilities with
governance and security features. Such standard analytics processes as column
dependencies, clustering, decision trees, and a recommendation engine are all
included in many of these new software packages.
Instead of forcing clients to frustratingly purchase module on top of module on
top of module, new software companies are creating packages that contain
many pre-built analytical functions. Open source products like R, Python, and the
WEKA collection in Vantara can easily be added to many of these software
solutions as well, thereby reducing the need for expensive analytics layers.
The fact that many of these analytical packages are open source is a further
advantage because, since they are free to download and use, they have a robust
user base and consultants are sometimes easier to find than analysts with highly
developed analytics skills.
Before going any further, I believe one of the first questions that needs to be
answered here is, “What exactly is analytics?” The standard answer is that there
are four types of analytics and they are:
• Descriptive analytics — What happened?
• Diagnostic analytics — Why did it happen?
• Predictive analytics — What will happen?
• Prescriptive analytics — How can we make it happen again?
For a cruise line, descriptive analytics could include pattern discovery methods
such as customer segmentation, i.e., culling through a patron database to
understand a patron’s preferred game of choice. Simple cluster segmentation
models could divide customers into their preferred sizes and styles.
Market basket analysis, which utilizes association rules, would also be
considered a descriptive analytics procedure. Cruise lines can use market basket
analysis to bundle and offer promotions as well as gain insight into a customer’s
buying habits. Detailed patron shopping and purchasing behavior could also be
used to develop future products.

19
 ANDREW W. PEARSON

In her article How Much ROI Can Data Analytics Deliver? 27, Annie Eissler points
out that, according to Nucleus Research “analytics and business intelligence
solutions deliver, on average, $13.01 for every dollar spent” and leading
companies have been achieving double-digit return on investment (ROI) from
their analytics investments for several years now.” In chapter three, I will delve
deeper into how a cruise line can utilize analytics to both reduce costs and, by
delivering personalized marketing, increase customer satisfaction.
In his article Will ‘Analytics on The Edge’ Be The Future Of Big Data? 28, Bernard
Marr ponders the question: “Rather than designing centralized systems where
all the data is sent back to your data warehouse in a raw state, where it has to
be cleaned and analyzed before being of any value, why not do everything at the
‘edge’ of the system?”
Marr uses the example of a massive scale CCTV security system that is capturing
real-time video feeds from tens of thousands of cameras.28 “It’s likely that 99.9%
of the footage captured by the cameras will be of no use for the job it’s supposed
to be doing — i.e., detecting intruders. Hours and hours of still footage is likely
to be captured for every second of useful video. So what’s the point of all of that
data being streamed in real-time across your network, generating expense as
well as possible compliance burdens?”28
The solution to this problem, Marr argues, is for the images themselves to be
analyzed within the cameras at the moment the video is captured. 28 Anything
deemed out-of-the-ordinary will trigger alerts, while everything considered to
be unimportant will either be discarded or marked as low priority, thereby
freeing up centralized resources to work on data of actual value. 28
For a cruise line, the CCTV security systems could be set up to alert a clerk that a
VIP has stepped onto the property.
Using edge analytics and real-time stream processing engines, cruise lines could
“analyze point-of-sales data as it is captured, and enable cross selling or up-
selling on-the-fly, while reducing bandwidth overheads of sending all sales data
to a centralized analytics server in real time.”28
Edge analytics, of course, goes hand-in-hand with the Internet of Things–“the
network of physical objects that contain embedded technology to communicate
and sense or interact with their internal states or the external environment.”29

BI + CRM + CX + IoT
The term “Big Data” has become a way-too-common and enormously prevalent
term and it is being bandied about a lot in the world of IT these days because it
has become a kind of catch-all for analytics, IoT, social media data, etc., etc.
Although not a comprehensive list, Big Data analytics techniques can include

20
 THE A.I. CRUISE LINE

association, classification, cluster analysis, crowdsourcing, data fusion, data

mining, machine learning (ML), modeling, network analysis, optimization,
predictive, regression, rule learning, special analysis, text analytics, time series
analysis, amongst many, many others. Which techniques should a cruise line
use? Well, that all depends on what type of data is being analyzed, the
technology available to it, the skills of the business users, and the business
problems it is trying to solve.
In chapter three, I break down how these analytical techniques would work in
the concept of the customer journey, and I will specifically explain in what
circumstances decision trees, time series, discriminant analysis, K-means
clustering, and K-Nearest Neighbor processes, amongst others, should be
utilized.
None of these techniques, however, will amount to anything if the underlying
data environment isn’t robust and properly cleansed; “junk in, junk out”, as most
analysts warn. The quality and quantity of the data gathered will be directly
proportional to the accuracy of the predictive model. Enormous attention must
be paid to ensure the data is prepped and cleansed, otherwise nothing of value
will be achieved, no matter how fast and/or robust your analytics software might
be able to crunch the underlying numbers.
In his seminal 2009 article for the RFID Journal, That 'Internet o Things' Thing30,
Kevin Ashton made the following assessment:
Today computers — and, therefore, the Internet — are almost
wholly dependent on human beings for information. Nearly all
of the roughly 50 petabytes (a petabyte is 1,024 terabytes) of
data available on the Internet were first captured and created
by human beings — by typing, pressing a record button, taking
a digital picture, or scanning a bar code. Conventional diagrams
of the Internet include servers and routers and so on, but leave
out the most numerous and important routers of all — people.
The problem is, people have limited time, attention and
accuracy — all of which means they are not very good at
capturing data about things in the real world. And that's a big
deal. We're physical, and so is our environment. Our economy,
society and survival aren't based on ideas or information —
they're based on things. You can't eat bits, burn them to stay
warm or put them in your gas tank. Ideas and information are
important, but things matter much more. Yet today's
information technology is so dependent on data originated by
people that our computers know more about ideas than things.
If we had computers that knew everything there was to know
about things — using data they gathered without any help from
us — we would be able to track and count everything, and

21
 ANDREW W. PEARSON

greatly reduce waste, loss and cost. We would know when

things needed replacing, repairing or recalling, and whether
they were fresh or past their best. The Internet of Things has the
potential to change the world, just as the Internet did. Maybe
even more so.
IoT technology costs are coming down, broadband’s price has dropped, while its
availability has increased, and there is a proliferation of devices with Wi-Fi
capabilities and censors built into them. Smart phone penetration is also
exploding. 5G is rolling out worldwide. All of these individual technological
advances were good for the IoT environment; together, however, they have
created a perfect storm for it.31 With less than 0.1% of all the devices that could
be connected to the Internet currently connected31, there is tremendous growth
potential here and those who embrace it now should have the first mover
advantage that could prove enormously valuable in terms of ROI over the long
term.
For this book, I will consider CRM as a two-part process that allows a cruise line
to track and organize its current and prospective customers, as well as to manage
the endpoints of customer relationships through its marketing promotions.
When done right, CRM systems enable data to be converted into information
that provides insight into customer behavior and, from these insights, some form
of behavioral influencing can occur.
Although widely recognized as an important element of most business’ customer
experience platform, there is no universally accepted definition of CRM. In his
paper Accelerating customer relationships: Using CRM and relationship
technology32, R. Swift defines CRM as an “enterprise approach to understanding
and influencing meaningful communications in order to improve customer
acquisition, customer retention, customer loyalty, and customer profitability.” J.
Kincaid adds that CRM is “the strategic use of information, processes, technology
and people to manage the customer’s relationship with your company
(Marketing, Sales, Services, and Support) across the whole customer lifecycle.” 33
In their paper Customer Relationship Management: Emerging Practice, Process,
and Discipline34 , Parvatiyar and Sheth claim CRM is, “a comprehensive strategy
and process of acquiring, retaining, and partnering with selective customers to
create superior value for the company and the customer. It involved the
integration of marketing, sales, customer service, and the supply chain functions
of the organization to achieve greater efficiencies and effectiveness in delivering
customer value.”
In their comprehensive article on the subject, Application of Data Mining
Techniques in Customer Relationship Management: a Literature Review and
Classification35, Ngai et al. argue that these varying definitions emphasize the
importance of “viewing CRM as a comprehensive process of acquiring and

22
 THE A.I. CRUISE LINE

retaining customers, with the help of business intelligence, to maximize the

customer value to the organization.”
In this book, we will consider CRM as a two-part process that allows a cruise line
to track and organize its current and prospective customers, as well as to manage
the endpoints of customer relationships through its marketing promotions.
When done right, CRM systems enable data to be converted into information
that provides insight into customer behavior.
The process of segmenting a market is deceptively simple; seven basic steps
describe the entire process, including segmentation, targeting, and positioning.
In practice, however, the task can be very laborious since it involves poring over
enormous amounts of data, and it requires a great deal of skill in analysis,
interpretation and some personal judgment.
Today, personalized web pages can be rendered during a web page load and
elements of the page can take into account past purchase history, clickstream
information, as well as a whole host of other personal customer details. In
chapter four, I will delve into the how and why of these systems in deep detail,
explaining both how website morphing works, as well as how advertisers should
utilize the psychology of personalization in their marketing.
Data coming from mobile and social media sources like WeChat, Weibo,
Facebook, YouTube, Twitter, YouKu, etc., tend to be highly unstructured, while
data coming from CSVs, XML and JSON feeds are considered semi-structured.
NoSQL databases are also considered semi-structured, while text within
documents, logs, survey results, and e-mails also fall into the unstructured
category. Structured data coming in from the plethora of retail source systems,
undoubtedly, can feed into a data warehouse, where it can be merged with
unstructured data and then utilized in ways that are almost impossible for a
normal relational DW to handle.
Highly structured customer data could be combined with unstructured data
coming in from social media to reveal deep customer insights. Setting up JSON
feeds for Twitter user accounts is a simple process and many other social media
companies offer APIs that allow access to customer accounts. These are two-way
systems as well, and the cruise line’s marketing department could include social
media as a channel to connect with customers and potential customers.
How does a cruise line get a player’s WeChat, RenRen, Facebook, Twitter, Weibo,
YouTube, or even Twitch account? Easy, just make the customer an offer he or
she can’t refuse, and, in most cases, that offer probably wouldn’t have to be
much more than a coupon for a percentage off their next purchase.
With quick and easy accessibility to a cruise line’s data, customer conversion
rates can be improved, revenue can be increased, and customer churn can be
predicted and, hopefully, mitigated as much as possible. Customer acquisition

23
 ANDREW W. PEARSON

costs can also be lowered. By utilizing the complex web of customer data coming
in from several different channels — mobile, social media, customer loyalty
programs, transaction data, e-commerce weblogs, sensors, amongst others — a
cruise line can also work more productively.
As Kai Wähner explains in his article Real-Time Stream Processing as Game
Changer in a Big Data World with Hadoop and Data Warehouse36, “Stream
processing is required when data has to be processed fast and/or continuously,
i.e. reactions have to be computed and initiated in real time.” Wähner
continues36:
“’Streaming processing’ is the ideal platform to process data
streams or sensor data (usually a high ratio of event throughput
versus numbers of queries), whereas “complex event processing”
(CEP) utilizes event-by-event processing and aggregation (e.g. on
potentially out-of-order events from a variety of sources — often
with large numbers of rules or business logic). CEP engines are
optimized to process discreet ‘business events’ for example, to
compare out-of-order or out-of-stream events, applying decisions
and reactions to event patterns, and so on. For this reason multiple
types of event processing have evolved, described as queries, rules
and procedural approaches (to event pattern detection).”
Stream processing acts on real-time streaming data feeds, using “continuous
queries” (i.e., SQL-type queries that operate over time and buffer windows). 36
With its ability to continuously calculate mathematical or statistical analytics on
the fly within the stream, streaming analytics is an essential part of stream
processing.36 “Stream processing solutions are designed to handle high volume
in real time with a scalable, highly available and fault tolerant architecture,” adds
Wähner36
“In contrast to the traditional database model where data is first stored and
indexed and then subsequently processed by queries, stream processing takes
the inbound data while it is in flight, as it streams through the server,” explains
Wähner.36 Stream processing can also connect to an external data source,
thereby adding a whole new dimension to analytical processes. Think social
media, geo-location, facial recognition, shelf sensor data, RFID inputs, or a whole
host of other data streams.
Real-time stream processing is an integral part of this rapidly changing marketing
environment and if cruise lines don’t join the real-time marketing world, they
will certainly be left behind, I have no doubt.
For cruise lines, real-time streaming can help in the following ways:
• Customer Service:

24
 THE A.I. CRUISE LINE

o Geo-locating a customer when he or she signs onto the cruise

line’s Wi-Fi.
o Video analytics with facial recognition technology can spot
and/or confirm a customer’s identity and perhaps cross-
reference him or her against a VIP list or known shoplifters.
o Social media customer service can cut down on normal
customer service expenses, as well as connect with customers
on the channels that they prefer.
• E-Commerce:
o Clickstream analysis could allow personalized offers to
potentially returning customer when they are browsing a
cruise line’s website.
• Human Capital Management:
o Employee schedules can be adjusted in real time according to
labor management’s needs, as well as its predictive and
anticipatory needs.
o Cruise lines can take the guesswork out of hiring employees by
building templates that show what a model employee should
possess in terms of skills and experience.
• Patron Management:
o The ecommerce department can get more accurate attribution
analysis — “the process of identifying a set of user actions
(‘events’) that contribute in some manner to a desired
outcome, and then assigning a value to each of these events” 37
— so that it understands which advertising is associated with
which user, making it more quantifiable and, therefore, more
actionable.
o Customer Relationship Management (CRM) systems can add
social media as a channel to feed targeted messages to only
those customer who are most likely to respond to a promotion.
o The amount of promotions available and channels through
which to market through increases considerably as campaign
lift can be assessed in terms of hours rather than in days or
weeks.
o Customer acquisition is accelerated because business users
throughout can quickly be questioned on the following:
▪ Which combinations of campaigns accelerate
conversion?
▪ What behavior signals churn?
▪ Do web search key words influence deal size?
▪ Which product features do users struggle with?
▪ Which product features drive product adoption and
renewal?

25
 ANDREW W. PEARSON

▪ What drives customers to use costly sales channels?

o Customer interaction data can quickly be turned into business
opportunities.
o Powerful recommendation engines can ingest data from a
multitude of sources and then be made available to frontline
staff, who can react in near real time.
• Security:
o Uncover AML activity.
o Spot shoplifters.
o Spot VIPs.
How can a cruise line utilize IoT technology? IoT sensors can be used for smart
parking or smart lighting. They can also be used for silo stock calculation —
measuring the emptiness level and weight of goods — as well as waste
management, and perimeter access control tools.
Liquid presence detection in places like data centers can help ensure the
integrity of the company’s IT backbone. For the cruise line, IoT can help with
supply chain control, NFC payment systems, inventory shrinkage, as well as
smart product management. For the cruise line’s logistics department, IoT aids
quality of shipment conditions, item location, storage incompatibility detection,
and fleet tracking.
Combining IoT data together with other structured and unstructured data isn’t
easy, though. Previous attempts at broad-based data integration has forced
users to build data sets around common predetermined schema, or a unifying
data model, but this becomes impossible when unstructured and semi-
structured data are included in the mix.

Personalization
Successful mobile advertising requires three things — reach, purity and analytics;
reach can be fostered by accessing accounts through multiple platforms like
blogs, geofencing applications, OTT services, mobile apps, QR codes, push and
pull services, RSS feeds, search, social media sites, and video-casting, amongst
others.38 “Purity” refers to the message and its cleanliness; if the data is
unstructured and untrustworthy it is, basically, useless and data governance is
paramount for real-time advertising to work properly.38 The third ingredient,
analytics, “involves matching users’ interests – implicit and explicit, context,
preferences, network and handset conditions – to ads and promotions in real
time.”38
Knowing what might interest a consumer is only half the battle to making the
sale and this is where customer analytics shines. Customer analytics have
evolved from simply reporting customer behavior to segmenting a customer

26
 THE A.I. CRUISE LINE

based on his or her profitability, to predicting that profitability, to improving

those predictions (because of the inclusion of new data), to actually
manipulating customer behavior with target-specific promotional offers and
context-aware marketing campaigns. These are the channels that real-time
thrives in and this is where a cruise line can gain a powerful competitive
advantage. Think about how much more powerful a marketing offer would be if
it was sent to a customer as he was entering a retail store.
Composing the marketing message, however, is probably the easiest part of the
process. In its Delivering New Levels of Personalization In Consumer
Engagement39, Forrester Research found that survey participants believed that
personalization had the potential to increase traffic, raise customer conversion
rates, and increase average order value. Surveyed marketers felt that
personalization capabilities could improve a variety of business metrics,
including customer retention (75%), lifetime customer value (75%), and
customer conversion rates (71%).39
Today, “Personalization” is becoming the optimum word in a radically different
business environment and even though this personalization comes at a price —
privacy — it is a price most consumers seem more than willing to pay if a
recognized value is received in return. For the cruise line, “personalization”
requires an investment in software analytics, but cruise lines should recognize
that this price must be paid because highly sophisticated consumers will soon
need an exceptional shopping and entertainment experience to keep them from
switching to a competitor. This kind of personalization also gives the cruise line
powerful information to build optimization models that can reduce cost and
increase productivity down the line.
These survey participants see email, call centers, corporate websites, mobile
websites and physical locations (such as stadiums, sporting venues and
hospitality sites) as today’s key customer interaction channels, but any future
marketing efforts to reach them should be “focused on mobile websites,
applications, and social media channels.”39 Cruise lines should keep these
channels in mind while devising their customer experience (CX) campaigns.
Understanding customer-specified preferences is imperative for personalization;
“80% of marketing executives currently use them in some or all interaction
channels. In addition, 68% of marketers personalize current customer
interactions based on past customer interaction history. Other commonly used
personalization methods used by nearly 60% of firms in some or all of their
interaction channels are based on the time of day or day of the week of customer
interactions.”39 Forrester Research states that the difficulties of personalization
include39:
1. Continuously optimizing campaigns in response to a customer’s most
recent interactions.

27
 ANDREW W. PEARSON

2. Optimizing content or offers for each person by matching identities to

available products, promotions, messages, etc.
3. Creating a single repository containing structured and unstructured
data about a consumer.
4. Delivering content or offers to a customer’s chosen channel in real time
for purposes of conversion.
5. Analyzing all available data in real time to create a comprehensive,
contextually sensitive consumer profile.
The executives pooled by Forrester Research expected there to be a “huge rise
in personalization using consumer’s emotional state, social media sentiment,
and context”39 as well. “Only 29% of respondents claim today to use inferences
about the consumer’s emotional state in some or all channels. But 53% expect
to do this in two to three years’ time.”39 Forrester’s report adds that, “Only 52%
of marketers currently use sentiments that consumers express in social media to
personalize interactions today, but fully 79% expect to do this in two to three
years. In addition, only 54% capitalize on the consumer’s current contextual
behavior, but 77% expect to do so in two to three years’ time.” 39
Today, mobile apps, mobile commerce, mobile chat, and mobile gaming have
revolutionized the way people do business, seek entertainment, and gamble.
Mobile commerce has evolved into what has become known as “omni-
commerce”, a seamless approach to selling that puts the shopper’s experience
front and center, giving that shopper access to what he or she wants through
these multiple channels.
Mobile marketing via Bluetooth, OTT, SMS, MMS, CSC and/or QR codes has
become some of the most effective marketing around, while social media has
turned the normal customer channels — and even the idea of marketing — on
its head.
By accessing the web through a wireless connection, mobile users can surf the
Internet as seamlessly as if they were using a PC at home. At the touch of a
button, photos and videos can be uploaded seconds after they are taken, then
shared with the most intimate of friends or the most distant of peoples. Live
streaming channels allow cheap video streams that can be viewed almost
anywhere on the planet as well.
The mobile platform is so robust and it has so much capability that if a marketing
executive had been asked to dream up the perfect device to connect to, market
to, and sell its company's products and/or services to its customers and potential
customers, he could hardly have come up with something more superior to it.
One of mobile’s best features is its ability to cross-pollinate the marketing
message through several mediums, which include social media — and I will
expound upon this throughout the book.
In its paper 5 E-Commerce Marketing Prediction For the Next 5 Years 40, the B2C

28
 THE A.I. CRUISE LINE

marketing cloud company Emarsys argues that, “Smart marketers need real-time
insights into mobile marketing performance in order to understand how end
users are (or aren’t) engaging with their mobile marketing programs or
applications.”
Emarsys adds that40:
“We will move from a world focused on designing for mobile as a
secondary approach, to designing for mobile first. E-commerce
organizations will finally fully alter the online shopping experience
from responsive to completely mobile experiences. This mobile-
only approach will be different, as it won’t just be a smaller design
but will also include more responsive websites and shopping
experiences. The mobile-only experience will lead to fully tailored
shopping experiences primarily designed for engagement on a
mobile device.”
Emarsys adds that, “Within the next five years, consumers will be able to swipe
right, up, and down to make their selections, all via their mobile devices. And
when the consumer is ready to complete the transaction? Easy. It just takes one
click; the purchase is complete, and the items arrive at the consumer’s house.” 40
The stakes couldn’t be any higher. In its paper, Emarsys concludes that, “In an
effort to remain competitive and innovative in today’s digital and always-
connected world, marketers should continually be piloting and testing mobile
strategies with a small subset of their users or target audience.” 40 However, “If
a brand slows mobile innovation, or pauses testing and optimization for mobile
devices, the brand is risking the loyalty of current users as well as jeopardizing
new user acquisition,” warns Emarsys.40
Much more than a wireless transmitter optimized for voice input and output, a
mobile phone, a tablet, or a phablet is an always-on, anytime, anywhere
marketing and sales tool that follows a mobile user throughout his or her digital
day.40 It is also an entertainment, CRM, and social networking tool, which makes
it, potentially, the most powerful device in the history of marketing and
customer relations.40 The mobile device is, literally, a marketing tool that can —
and usually is — personalized by its owner, and it is within reach of that owner
almost every hour of every single day; in essence, a marketer's dream.40
Push technology even puts the power of communication into the hands of the
marketer, allowing cruise lines to both initiate contact with an opted-in customer
and then sending him or her a wide variety of products and content. Many cruise
lines now have mobile apps in which they can connect to their patrons. As long
as a customer is opted into a CRM system, a cruise line can foster a two-way
dialogue with that customer and this dialogue can grow more sophisticated over
time as more is learned about the customer’s wants, desires, habits, and needs.

29
 ANDREW W. PEARSON

Push technology has moved from clumsy blanket SMS blasts to the sophisticated
use of mobile apps that allow customers to interact with their personal patron
points balance information.

Going Social
One of the most important elements of social media is its inter-connectedness.
An upload to YouTube can go viral through Twitter, Facebook, LinkedIn, WeChat,
WhatsApp, Youku, as well as a whole host of other social media and mobile
media platforms instantly.
Within seconds, something uploaded onto a social media website in the US can
end up on a mobile application in China or Japan or Korea, or almost anywhere
else in the world that has mobile or Wi-Fi access. We are truly living in an
interconnected world and this interconnectedness is creating a whole host of
ways to market a product, a service, or even a cruise line.
Social media will also be explored in depth throughout this book. It is quite ironic
that, in one sense, engaging in social media can be one of the most anti-social
behaviors one can do; sitting alone in a room, typing away on a computer was
once the realm of solitary computer geeks, but it has now become an activity
that most people engage in almost every single day. Perhaps this is because
human beings are, first and foremost, social beings and we crave a
connectedness that social media offers, even if it is only a virtual connection. The
use of social media on mobile has expanded its reach exponentially as well,
making it the perfect place to market a cruise line’s products.
It should be of no surprise that one of the greatest inventions of the twentieth
century – the internet – would become the watering hole of the twenty-first
century; a place where human beings can quickly gather to socialize and connect
with friends, family members, and acquaintances in a way that was almost
inconceivable only 20 years ago. Smart retail marketers can tap into this
interconnectedness to get out their marketing message far and wide.
Almost a decade ago, “most consumers logged onto the internet to access e-
mail, search the web, and do some online shopping. Company websites
functioned as vehicles for corporate communication, product promotion,
customer service, and, in some cases, e-commerce. Relatively few people were
members of online communities”41 and “Liking” something had no social context
at all. How times have changed.
In his article Understanding social media in China 41, C.I. Chui argues that the
secret to social media’s growth is right there in its name — “Social” — as in the
fundamental human behavior of seeking “identity and ‘connectedness’ through
affiliations with other individuals and groups that share their characteristics,
interests, or beliefs.”41 For Chui, “Social media taps into well known, basic

30
 THE A.I. CRUISE LINE

sociological patterns and behaviors, sharing information with members of the

family or community, telling stories, comparing experiences and social status
with others, embracing stories by people with whom we desire to build relations,
forming groups, and defining relationships to others.”41
“Today, more than 1.5 billion people around the globe have an account on a
social networking site, and almost one in five online hours is spent on social
networks — increasingly via mobile devices.”41 In little more than a decade,
social technology has become a cultural, social, political and economic
phenomenon.41 Most importantly, “hundreds of millions of people have adopted
new behaviors using social media — conducting social activities on the Internet,
creating and joining virtual communities, organizing political activities” 41, even,
as with the case of Egypt’s “Twitter Revolution”, toppling corrupt governments.
Social technologies allow individuals to interact with large groups of people at
almost any location in the world, at any time of the day, at marginal, if not no
cost at all.41 With advantages like these, it is not surprising that social media has
become so widespread that almost one-in-four people worldwide uses it. It is
actually surprising that the figure is so low, although with mobile and 5G
technology rolling out in some of the most remote locations on earth, that figure
is sure to rapidly climb over the next few years.
Businesses are also quickly recognizing the power of social media. In his article
The social economy: Unlocking the value and productivity through social
technologies42, Chui et al. argue that, “Thousands of companies have found that
social technologies can generate rich new forms of consumer insights — at lower
cost and faster than conventional methods.”42 In addition to this, businesses can
watch what “consumers do and say to one another on social platforms, which
provide unfiltered feedback and behavioral data (i.e., do people who “like” this
movie also “like” this brand of vodka?).”42 This can be a treasure trove of
company competitive analysis and I believe cruise lines would profit from
spending more on these types of social media listening efforts. In chapter five
and seven, I provide information on platforms and methodologies in which to
utilize social media as a powerful marketing channel.
Social technologies also “have enormous potential to raise the productivity of
knowledge workers,”42 a very significant development in a world where
knowledge workers are becoming highly sought-after assets. “Social
technologies promise to extend the capabilities of such high-skill workers (who
are increasingly in short supply) by streamlining communication and
collaboration, lowering barriers between functional silos, and even redrawing
the boundaries of the enterprise to bring in additional knowledge and expertise
in ‘extended networked enterprises.’”42
In this book, I will use Chui et al.’s definition of “social technologies” as the
“products and services that enable social interactions in the digital realm, and

31
 ANDREW W. PEARSON

thus allow people to connect and interact virtually.”42 These include42:

“A message to be communicated (a tweet or a blog), adding
content to what is already online, or adding information about
content (‘liking’ a piece of content). Content creation also
includes performing an action that an individual knows will be
automatically shared (e.g., listening to a piece of music when
you know your music choice will be displayed to others). Social
technologies allow anyone within a group to access and
consume content or information. They include technologies that
also have been described as ‘social media,’ ‘Web 2.0’ and
‘Collaboration tools’.”42
In China, users spend more than 40 percent of their time online on social media
websites, a figure that is expected to continue its rapid rise over the next few
years.41 “This appetite for all things social has spawned a dizzying array of
companies, many with tools that are more advanced than those in the West: for
example, Chinese users were able to embed multimedia content in social media
more than 18 months before Twitter users could do so in the United States.”41
Companies like WeChat are revolutionizing social networks, adding malls as part
of their platforms, while Taobao has teamed up with Weibo to allow instant
commentary and blogging on purchased items. yy.com inverted the concept of
reality TV, by taking a singing competition and broadcasting it over the Internet,
while allowing viewers to directly remunerate the contestants.
China is also making big strides in AI. The big four as they are called – Alibaba,
Baidu, WeChat, Ten Cent – are working on some very cutting-edge AI technology
that will probably be replicated in the West shortly. The Chinese characters for
“Artificial Intelligence” are actually “人工智能”, the first two of which,
amazingly, look like the acronymed version of the English AI; perhaps if there’s
some kind of international human cosmic convergence coming, AI will be leading
the way…
Facebook might be the biggest social network in the world, but its penetration
in China is minimal and it will probably remain so for a long time to come, not
only because China sensors Facebook, but also because Facebook’s Chinese
competitors are actually creating some very technologically savvy products.
WeChat, in particular, has proven to be highly successful, and it is growing
rapidly, both in China and throughout the rest of the world, but companies like
QQ, Weibo, Hexun, Youku, Huya, Jiepang, Qieke, Ushi and Ku6 have all
experienced exponential growth. With a base of 1.3 billion people, it isn’t too
hard for services that catch on in China to rapidly get to tens of millions of users
within a year or even sooner.
In August 2017, Steven Millward noted in his article Facebook’s new China app

32
 THE A.I. CRUISE LINE

is reliant on WeChat43 that Facebook launched a photo-sharing app to test the

waters in a market that Facebook has been locked out of since 2009. As Millward
writes, “Facebook’s new app, a China-segregated photo-sharing service called
Colorful Balloons, is reliant on WeChat for its viral growth.43 The app is not
connected to Facebook and it cannot tap into a user’s Facebook contacts. 43 “Hit
the WeChat button, and you will send your friend an invite to your photo album
within Facebook’s new app. It’s made up of an album preview plus a QR code,
sent via a WeChat message, which takes your buddy either to the Colorful
Balloons app or to the iOS App Store,” notes Millward. 43
There is an old adage in social media marketing that says, “Content is king” and,
with social media, that adage has never been truer. Those destined to succeed
in the social media sphere won’t be the ones with the most content; they will be
the ones with the best and most searchable content. And that content will drive
eyeballs unlike any other form of marketing in the history of advertising. To
succeed in this new environment, cruise lines should think of themselves first
and foremost as creators and syndicators of content.44 But “If content is king,
then ‘conversion is queen,’” argues John Munsel, CEO of Bizzuka, and he has an
important point there; social media is about converting customers, driving
eyeballs, and building channels to reach them.
Starbucks has even also developed a metric it believes quantifies the value of its
social media marketing in terms of media spend — the “company’s 6.5 million
Facebook fans are worth the equivalent of a US $23.4 million annual spend,
according to calculations by social media specialists Virtue, reported in
Adweek.”45
Virtue claims that, on average, “a fan base of 1 million translates to at least $3.6
million in equivalent media over a year, or $3.6 per fan. Virtue arrived at its $3.6
million figure by working off a $5 CPM, meaning a brand’s 1 million fans generate
about $300,000 in media value each month.”45 That’s quite a significant amount
of money and, if a coffee company can find success in social media, retail
companies should be able to get a similar, if not higher, ROI.
Chapter five delves into the world of social media, including a discussion of
Kaplan and Haenlein’s six types of social media.46 In their influential article Users
of the world, unite! The challenge and opportunities of Social Media 46, Kaplan
and Haenlein show how all social media websites can be broken down into one
of six different types; collaborative projects; blogs and micro-blogs; content
communities; social networking sites; virtual game worlds; and virtual social
worlds.46 Anyone devising a social media marketing plan for a cruise line should
find this chapter particularly helpful in understanding how to use each separate
channel and platform, both singularly and, even better, combined together.
Google Analytics defines six sources of traffic – direct, referral, paid search,
display advertising, social, and email and I will detail how AI can utilize some of

33
 ANDREW W. PEARSON

these channels.47
Private messaging platforms such like Messenger, WeChat, WhatsApp, Snapchat
in Western markets and Line, Viber, and Telegraph in Asian countries create
click-through traffic from links shared in those networks, which go under the
Google Analytics radar. Today, a company’s social media marketing plan must
include the dark social channels like WhatsApp, Facebook Messenger, WeChat,
LINE, Viber, and Telegraph, et al.
In his Dark Social: We have the whole history of the web wrong48, Alexis Madrigal
of The Atlantic coined the term “Dark Social” to describe these channels.
Madrigal claims that the “vast trove of social traffic is essentially invisible to most
analytics programs.” “It shows up variously in programs as ‘direct’ or
‘typed/bookmarked’ traffic, which implies to many site owners that you actually
have a bookmark or typed in www.theatlantic.com into your browser. But that's
not actually what's happening a lot of the time. Most of the time, someone
Gchatted someone a link, or it came in on a big email distribution list, or your
dad sent it to you,” claims Madrigal.48
Gilliland calls dark social today’s “biggest missed opportunity in marketing”49 and
these are platforms that are growing bigger and more important by the day.
Cruise lines need to tap into this opportunity quickly and I will provide examples
of how to do that in chapters four and five.
I have also added a section that discusses marketing virality and lays out Kaplan
and Haenlein’s feeling that “viral marketing is as much an art as it is a science.”50
Companies wishing to go viral should firmly understand that, even with the best
laid plans, there’s a fair amount of luck in it.

Capacity Management
In chapter six, capacity management and planning take center stage. Today, a
single amusement park ride can cost more than the production cost of a
Hollywood blockbuster movie.51 A cruise ship in the Caribbean Princess class
recently cost approximately US$500 million, with a berth cost of US$200,000. 52
The value of a 150-room hotel can be as much as £13.6 million, with a cost per
square meter of £1,300.53 However, as Pullman and Rodgers lament in their
Capacity Management for hospitality and tourism: a review of current
approaches54, “enterprises often make important capital investment decisions
without taking advantage of the full range of sophisticated quantitative tools
developed in the field of operations management.” There are several capacity
models that represent a valuable and often untapped resource for firms in the
tourism and hospitality industries to use and I will detail them throughout
chapter six.54

34
 THE A.I. CRUISE LINE

In her article Connectivity helps hotels manage energy use55, Esther Hertzfeld
writes that, “Connectivity is at the forefront of many technologies these days
and with good reason. Artificial intelligence can greatly help hotels manage their
energy costs by automating tasks that traditionally were performed by property
staff.” For example, the obvious one: “energy-management thermostats
automatically set temperatures back in unoccupied rooms to save energy.”55
Machine learning is also used “to understand how quickly a guestroom can be
heated or cooled at any given time in order to recover to the guest’s preferred
temperature in a predetermined amount of time.”55
“Integrated connectivity will continue to evolve and will allow hotels to manage
energy costs more precisely with greater ease and convenience than ever,” said
TJ Wheeler, VP of marketing and product management at Friedrich Air
Conditioning.55 “However, guest expectations may well play into this as well,”
said Wheeler.55 “As guests become accustomed to smart devices in their own
homes, they may expect to be able to control room temperature, TVs, lights and
more through a smart device or app,” added Wheeler.55
“Energy-management technologies can reduce hotel energy consumption 25
percent to 35 percent by automatically responding to guestroom occupancy
patterns and adjusting the thermostat to conserve energy for heating and
cooling needs when a guest is not in the room, which, according to industry
statistics, is about 50 percent of the time in most hotels,” explains Hertzfeld.55
“Companies are making significant efforts to improve building automation and
control systems to optimize performance as well as increase guest comfort, said
Ryan Gardner, product marketing manager for Honeywell/Inncom.”55
“However, it is important to keep in mind that as Internet of Things systems
proliferate in hospitality and the number of property sensor networks increase,
the complexity of managing these systems will also increase,” argues Hertzfeld.55
“Gone are the days when a straightforward mechanical fix was all that was
needed to solve an efficiency or comfort issue,” said Gardner.55 “For this reason,
Honeywell is working on making the management of these complex systems
easier for property staff using IoT. For too long, technology vendors have focused
on providing advanced technology offerings, but have not made these systems
easy to manage, or even to be self-managed,” he added.55
“Gardner said utilizing systems that provide predicative analytics will help
hoteliers identify when the system is not performing optimally.”55 “The best
systems monitor real-time savings and offer insights on how an inefficient
system can get back on track,” contends Hertzfeld.55
“The rise of Amazon Echo, Google Home and other devices can greatly help
hoteliers because they eventually will allow guests to control all the devices in
their room via voice command,” explains John Attala, marketing director at
Verdant Environmental Technologies.55 “Verdant is currently testing technology

35
 ANDREW W. PEARSON

to allow guests to adjust the temperature in the room without getting out of
bed. Guests will also eventually be able to order room service without having to
pick up the phone,” he said.55
“The ability to include a developed application programming interface into the
guest loyalty app gives the hotelier the ability to integrate mobile key,
temperature control, lighting control and other control features tied to the guest
preferences or patterns,” said Chad Burrow, director of sales at Telkonet.55
“In other words, when the room recognizes my mobile key, the room can
automatically set preferred temperature, lights on or shades open, and the TV
on with a welcome message to me personally,” said Burrow.55 “Combine this
with a geofencing component and the room will not only revert to an unoccupied
status with lights off, temperature set-back, but start reverting back to occupied
when I am back in range.”55
“Sometimes the pursuit of efficiency gives hoteliers and technology vendors a
black eye if they give the illusion that these solutions sacrifice guest comfort for
profits,” argues Hertzfeld.55 “Nothing could be further from the truth when
hoteliers deploy systems and technologies,” Gardner said. 55 “It can be easy to
forget that the focus of hospitality is on the guest,” he adds.55
“Many guests are concerned with energy usage and the impact that energy
consumption has on the environment so being at the forefront in terms of
reducing energy usage is always a good thing,” said Wheeler.55 “Part of that is
perception — that the hotel property is a good corporate citizen but also that it
shows responsibility in the use of resources,” he said. 55 “The best solutions will
balance reduced energy consumption with optimum comfort and convenience
for the guest.”55
More and more guests are sustainability focused consumers, agrees Burow.55
“Leadership in Energy and Environmental Design certification is one of many
ways to market a sustainable property,” argues Burow.55 “A hotel’s loyalty app
is also a great place to put the power of sustainability into the guests' hands with
proper marketing,” said Burow.55
“Common areas are often forgotten as a source of potential energy savings,”55
contends Hertzfeld, adding “Hallways, ballrooms, gyms, spas and pools are all
areas where hotels can reduce energy consumption.”55 “Meeting rooms are
prime candidates for energy savings because they generally have larger surface
areas than guestrooms and are used intermittently,” says Attala.55 “Burow
believes common spaces like ballrooms can benefit by utilizing intelligent plug
loads and switches. The same can be said for office spaces, kitchens and other
back-of-house spaces.”55
“Zone cooling and heating systems are ideal for gyms and common areas that
are extremely busy in the morning and evening hours but see a reduced number

36
 THE A.I. CRUISE LINE

and frequency of guests during the daytime,” says Gardner.55 “They operate
quietly and can be part of a multizone system where each unit can be
independently controlled. They are also very efficient in both cooling and heating
mode.”55
Hotel kitchens can be big energy-wasting areas as well.55 “With the big
commercial appliances working continuously to prepare breakfast, brunch,
lunch and dinner for hundreds of guests on a daily basis, the final electricity bill
can be quite discouraging,” Hertzfeld.55

The Future
My hope is this book can be a blueprint for a cruise line to step into the AI, Big
Data, CX, IoT, ML and predictive analytics world, so that it can both understand
its customers on a truly intimate level and also shape the experiences of those
customers so that a healthy ROI can be extracted from any tech investment
made.
Every cruise line must ask the question: “When will we reach the point of
diminishing returns?”, and “Where does the incremental cost of improved
performance equal or exceed the incremental value created?” Having a data
driven organization is imperative in today’s world of demanding customers, who
expect highly sophisticated technology to be available to them 24 hours a day,
seven days a week, 365 days a year.
When it comes to analytics and Big Data, Caesars was the first casino company
to collect and analyze data for Customer Intelligence (CI) purposes and, since the
inception of it Total Rewards programme, the company has grown from “being
able to trace the journey of 58% of the money spent in their cruise lines to
85%.”56 Caesars also credits the widespread adoption of Big Data analytics as the
driving force behind its rise from an “also ran” chain to one of the largest casino
groups in the U.S.57
In chapter seven, I attempt to create a holistic view of how the cruise line of the
future — The A.I. Cruise Line — would operate. I have added some real-world
examples of how a cruise line’s IT department would build an A.I. system that
could surface information from facial recognition cameras, angel eye and POS
systems, geo-locating devices, on-floor patron card swipes, etc., etc. This data
would quickly become actionable intelligence once it is put into the hands of the
front-line staff.
In their MIT Sloan Management Review article Winning with AI7, Ramsbotham
et al. provide the following caution:
“Most AI success stories focus on improving existing business
processes, whether in sales, marketing, pricing, servicing,

37
 ANDREW W. PEARSON

forecasting, manufacturing, or the like. But these

improvements are comparable to improving the gas mileage of
combustion engine vehicles in an era of new transportation
possibilities. Business executives need to consider how they can
reinvent and reimagine many of those processes in the context
of what AI enables. This is where AI’s true potential will
emerge: not in doing the same thing better, faster, and cheaper
but by doing new things altogether. This is where AI will disrupt
industries the most.”
“As business leaders look to the future, they must also carefully consider how AI
may affect their talent strategy,” warn Ramsbotham et al.7 “In most companies,
the skill sets and success profiles of the workforce (and the talent pools from
which they will come) will be materially different in the next decade or two than
they are today; the effect of this change on a company’s long-term HR strategy
will be nothing short of massive,” the writers note.7
One thing is certain, Ramsbotham et al. argue, “If AI initiatives are not core to a
company’s business strategy, they are unlikely to create meaningful value and
scale.”7 On an ending note, the writers argue that, “if a company’s current
business strategy ignores AI as a risk or as an opportunity, it probably needs
revisiting.”7
Big Data can either provide a useful treasure trove of information or it can
release a Pandora’s Box full of pain. It can certainly be either one but going down
the Big Data road requires a commitment that is all encompassing and difficult
to implement. State of the art technology is required and that always means the
potential for severe bumps in the road along the way. However, it is a road that
must be traversed as today’s consumer have become highly sophisticated and
they demand a level of personalization for their continuing patronage; if a
company’s marketing efforts aren’t personal enough, these consumers will find
another company that will provide the level of service they insist upon.
Throughout the book, I will try to avoid what has become known as
“wishcasting”, a useful term that the field of meteorology has recently given us.
As Rob Tracinski explains in his article How Not to Predict the Future58, “It started
with the observation that weathermen disproportionately predict sunny
weather on the 4th of July and snow on Christmas Day. Their forecasts are
influenced not just by the evidence, but by what they (or their audience) want
to hear.” The writer of any book that delves into current and future technology
will, obviously, be susceptible to wishcasting, but I will try to temper my
enthusiasm and add a dash of skepticism to all I write here.
I will also try to avoid “Zeerust” — “The particular kind of datedness which afflicts
things that were originally designed to look futuristic.”59 Taken from The
Meaning of Liff by Douglas Adams, TV Tropes explains it this way: “datedness

38
 THE A.I. CRUISE LINE

behind zeerusty designs lies in the attempt of the past designers to get an
advantage over the technology of their time, only to find out that more mundane
designs are actually far more efficient if advanced engineering and
craftsmanship are used on them.”59
Although this book was written for retail executives mostly in the IT and analytics
space, they might view the following story from The Economist as instructive.
Bartleby’s article What businesses can learn from the arts60 describes a unique
MBA class visit at Oxford’s Saïd Business School. As Bartleby explains: “Some of
the students had to try conducting the choir. The first to take the challenge was
a rather self-confident young man from America. It didn’t take long for him to go
wrong. His most obvious mistake was to start conducting without asking the
singers how they would like to be directed, though they had the expertise and
he was a complete tyro.”60
“The session, organised by Pegram Harrison, a senior fellow in entrepreneurship,
cleverly allowed the students to absorb some important leadership lessons. For
example, leaders should listen to their teams, especially when their colleagues
have specialist knowledge. All they may need to do, as conductors, is set the pace
and then step back and let the group govern itself,” says Bartleby.60
“It was noticeable, too, that the choir managed fairly well even if the conductors
were just waving their batons in an indeterminate fashion. The lesson there, Mr.
Harrison said, was that leaders can only do so much damage — provided they do
not attempt to control every step of the process. The whole exercise illustrated
it is possible for a lesson to be instructive and entertaining at once,” concluded
Bartleby.60 This is a good lesson for executives in a business as complicated as IT
and analytics, which requires employees highly skilled in extremely unique
subjects, whose work often needs to be augmented by outside consultants who
are skilled in even more unique aptitudes.
A modern full-scale symphony orchestra consists of approximately one hundred
permanent musicians, including 1st violins, 2nd violins, violas, cellos, double
basses, flutes, oboes, clarinets, bassoons. In theory, all woodwind players are
expected to be able to play all auxiliary instruments in addition to their main
instrument, something many an IT person can relate to. Furthermore, there is a
horn section that includes trumpets, trombones, a tuba, a kettledrum player,
several percussionists, a harp or two and a keyboard player.
Saxophonists, guitarists, bass oboists, and synthesizer players are brought for
special projects, kind of like how outside consultants can be brought in to add to
an IT team to add something the company lacks and, hopefully to create a fuller
harmonic whole. An orchestra is an excellent analogy for an IT department as it
contains so many singular pieces that must work together – in concert, if you will
– to produce something that is, at times, almost magical.
In French, “ROI” (or “Roi” more precisely) literally means “King” and in this book,

39
 ANDREW W. PEARSON

ROI is king; every piece of technology I discuss will be looked at through the lens
of ROI. As I detail in this book, positive ROI can be created with AI and machine
learning, all forms of analytics, marketing, and social media, amongst other
things. Detailed examples will be provided throughout.

1 http://www-01.ibm.com/software/data/bigdata/ (accessed: December 6, 2016)

2 Anderson, C. (2004). Wired. The Long Tail, pp. 171-177.
3 Berman, S. J. (2007). Executive Brief: Navigating the media divide: Innovating and
enabling new business models. IBM Institute for Business Value.
4 SAS. (2019). SAS announces $1 Billion investment in Artificial Intelligence (AI). March
17, 2019. https://www.sas.com/en_us/news/press-releases/2019/march/artificial-
intelligence-investment.html (Accessed 28 March 2019).
5 PWC. (2017). Sizing the prize. What’s the real value of AI for your business and how you
can capitalise. https://www.pwc.com/gx/en/issues/analytics/assets/pwc-ai-analysis-
sizing-the-prize-report.pdf (Accessed 19 February 2019).
6 Panetta, Kasey. (2017). Gartner Top 10 Strategic Technology Trends for 2018.
Gartner.com. https://www.gartner.com/smarterwithgartner/gartner-top-10-strategic-
technology-trends-for-2018/ (Accessed 8 January 2018).
7 Ransbotham, Sam, Khodabandeh, Shervin, Fehling, Ronny, Lafountain, Burt and Kiron,
David. 2019. MIT Sloan Management Review. Winning with AI. 15 October 2019.
(Accessed 16 October 2019).
8 Adobe. Artificial intelligence Unlocks the True Power of Analytics.
https://www.adobe.com/au/insights/ai-unlocks-the-true-power-of-analytics.html
(Accessed 14 January 2019).
9 https://skymind.ai/wiki/use-cases (accessed 22 September 2019).
10 Tang, Ziyi, and Lahiri, Tripti. (2018). Quartz. Here’s how the plan to replace the humans
who make iPhones with bots is going. June 22, 2018. https://qz.com/1312079/iphone-
maker-foxconn-is-churning-out-foxbots-to-replace-its-human-workers/ (Accessed 27
January 2019).
11 Bird, Jon. (2018). Forbes. Chilling or Thrilling? JD.com Founder Envisions a ‘100%’ Robot
Workforce. April 27, 2018. https://www.forbes.com/sites/jonbird1/2018/04/27/chilling-
or-thrilling-jd-coms-robotic-retail-future/#520fb59f7fcf (Accessed 27 January 2019).
12 https://www.youtube.com/watch?v=D5VN56jQMWM (Accessed 12 November 2019).
13 https://www.youtube.com/watch?v=co3fdFNUaFA&feature=youtu.be&t=2m59s
(Accessed 12 March 2019).
14 Summary, E. (2014). ‘Cisco Visual Networking Index: Global Mobile Data Traffic
Forecast Update, 2013-2018,’
http://www.cisco.com/c/en/us/solutions/collateral/service-provider/visual-networking-
index-vni/white\_paper\_c11-520862.html. (Accessed 17 November 2018).
15 Zixkhur, K., ‘Location-Based Services,’ http://www.pewinternet.org/2013/09/12/
location-based-services, 2013, (Accessed 17 November 2019).
16 Ward, Katrina. "Privacy preservation in social media environments using big data"
(2019). Doctoral Dissertations. 2797.
https://scholarsmine.mst.edu/doctoral_dissertations/2797 (Accessed 17 November
2019).

40
 THE A.I. CRUISE LINE

17 Bonchi, F. and Wang, H. W. (2011). ‘Trajectory Anonymity in Publishing Personal

Mobility Data,’ SIGKDD. Vol 13, pp. 30–42.
18 Kharpal, Arjun. (2017). CNBC. Amazon’s Alexa voice assistant could be a 10 billion
‘mega hit’ by 2020: Research. https://www.cnbc.com/2017/03/10/amazon-alexa-voice-
assistan-could-be-a-10-billion-mega-hit-by-2020-research.html (Accessed 12 January
2019).
19 Milgrom, Paul R. and Tadelis, Steve. (2018). How Artificial Intelligence and Machine
Learning Can Impact Market Design. 6 January 2018.
https://www.nber.org/chapters/c14008.pdf (Accessed 20 January 2019).
20 Skymind. A beginning’s guide to natural language processing.
https://skymind.ai/wiki/natural-language-processing-nlp (Accessed 2 January 2019).
21 Marr, Bernard. Bernardmarr.com. Text Analytics: How to Analyse and Mine Words And
Natural Language in Businesses. https://bernardmarr.com/default.asp?contentID=1754
(Accessed 8 January 2018).
22 Aaker, David A., Batra, Rajeev, Myers, John G. (2009). Advertising Management, 5Th
Edition. 9788177588507, Pearson Education India.
23 Woods, D. (2011, May 6). How Real-time Marketing Technology Can Transform Your
Business. Retrieved from Forbes.com:
http://www.forbes.com/sites/ciocentral/2011/05/06/how-real-time-marketing-
technology-can-transform-your-business/
24 Humanizing Loyalty A road map to establishing genuine emotional loyalty at scale.
Olson1to1.com. https://go.icf.com/rs/072-WJX-782/images/Humanizing%20Loyalty%20-
%20Olson%201to1.pdf (Accessed 6 August 2018).
25 Lewnes, Ann and Keller, Kevin Lane. (2019). MIT Sloan Management Review. 10
Principles of Modern Marketing. April 3, 2019. https://sloanreview.mit.edu/article/10-
principles-of-modern-marketing/ (Accessed 5 April 2019).
26 Adobe Sensei Team. Adobe. AI: Your behind-the-scenes marketing companion.
https://www.adobe.com/insights/sensei-ai-for-marketers.html (Accessed 7 April 2019).
27 Eissler, Annie. 25 April 2017. How Much ROI Can Data Analytics Deliver? MITS Blog.
https://www.mits.com/blog/how-much-roi-can-data-analytics-deliver (accessed 22
August 2017)
28 Marr, Bernard. August 23, 2016. Will ‘Analytics On The Edge’ Be the Future of Big Data?
Online: http://www.forbes.com/sites/bernardmarr/2016/08/23/will-analytics-on-the-
edge-be-the-future-of-big-data/#124af7ea2b09
29 Gartner. (2013, December 12). Gartner Says the Internet of Things Installed Base Will
Grow to 26 Billion Units By 2020. Retrieved from Gartner.com:
http://www.gartner.com/newsroom/id/2636073
30 Ashton, K. (2009, June 22). That 'Internet of Things' Thing. Retrieved from RFID Journal:
http://www.rfidjournal.com/articles/view?4986
31 Morgan, J. (2014, May 13). A Simple Explanation of 'the Internet of Things'. Retrieved
from Forbes.com: http://www.forbes.com/sites/jacobmorgan/2014/05/13/simple-
explanation-internet-things-that-anyone-can-understand/
32 Swift, R. (2001). Accelerating customer relationships: Using CRM and relationship
technologies. Upper Saddle River: Prentice Hall.
33 Kincaid, J. (2003). Customer relationship management: Getting it right. Upper Saddle
River, NJ: Prentice Hall.
34 Parvatiyar, A. &. Sheth, JN (2001). Customer relationship management: Emerging
practice, process, and discipline. Journal of Economic & Social Research, 3, 1 - 34.
41
 ANDREW W. PEARSON

35 Ngai, N. X. (2009). Application of data mining techniques in customer relationship

management: a literature review and classification. Expert systems with applications,
2592-2602.
36 Wähner, Kai (2014, September 10). Real-Time Stream Processing as Game Changer in
a Big Data World with Hadoop and Data Warehouse. InfoQ.
https://www.infoq.com/articles/stream-processing-hadoop/
37 Interactive Advertising Bureau. (2012). Attribution Primer. Retrieved from iab.net:
http://www.iab.net/media/file/AttributionPrimer.pdf
38 Sharma, C. H. (2008). Mobile Advertising: Supercharge Your Brand in the Exploding
Wireless Market. John Wiley & Sons, Inc.
39 Forrester Research. (2013, November). Delivering New Levels Of Personalization In
Consumer Engagement. Retrieved from sap.com:
https://www.sap.com/bin/sapcom/he_il/downloadasset.2013-11-nov-21-22.delivering-
new-levels-of-personalization-in-consumer-engagement-pdf.html
40 Emarsys. 5 E-Commerce Marketing Predictions for the Next 5 Years.
https://www.emarsys.com/en/resources/whitepapers/5-e-commerce-marketing-
predictions-for-the-next-5-years/
41 Chui, C. I. (2012, April). Understanding social media in China. Retrieved from
www.mckinsey.com:
http://www.mckinsey.com/insights/marketing_sales/understanding_social_media_in_c
hina
42 Chui, Michael Manyika, James, Bughin, Jaques, Dobbs, Richard, Roxburgh, Charles,
Sarrazin, Hugo, Sands, Geoffrey, Westergren. (2012). The social economy: Unlocking
value and productivity through social technologies. McKinsey Global Institute.
43 Millward, Steven. August 12, 2017. Facebook’s new China app is reliant on WeChat.
Techinasia.com. https://www.techinasia.com/facebook-china-app-reliant-on-wechat
(accessed August 18, 2017).
44 Black, L. M. (2012, November 11). 7 social media marketing tips for artists and galleries.
Retrieved from Mashable: http://mashable.com/2012/11/10/social-media-marketing-
tips-artists-galleries/
45 Woodcock, N. G. (2011). Social CRM as a business strategy. Database Marketing &
Customer Strategy Management, Vol. 18, 1, 50-64.
46 Kaplan, A. H. & Haenlein, M. (2010). Users of the world unite! The challenges and
opportunities of social media. Business Horizons, Vol. 53, Issue 1.
47 Google. Google Analytics Acquisition Channels Explained.
https://dashthis.com/blog/google-analytics-display-traffic/ (Accessed 15 December
2019).
48 Madrigal, A. (2012). Dark Social: We have the whole history of the web wrong. The
Atlantic, October 12, 2012.
https://www.theatlantic.com/technology/archive/2012/10/dark-social-we-have-the-
whole-history-of-the-web-wrong/263523/ (Accessed 16 December 2019).
49 Gilliland, N. (2018). Only 4% of marketers are taking Dark Social seriously. E-
consultancy, 27 February 2018. https://econsultancy.com/only-4-of-marketers-are-
taking-dark-social-seriously/ (Accessed 15 December 2019).
50 Kaplan, Andreas, Haenlein, Michael. (2011). Business Horizons. Two Hearts in three-
quarter time: How to waltz the social media/viral marketing dance. Business Horizons.
2011; 54 (3) : pages 253—263.
https://www.researchgate.net/publication/227413606_Two_hearts_in_three-
42
 THE A.I. CRUISE LINE

quarter_time_How_to_waltz_the_social_mediaviral_marketing_dance (accessed 22
September 2019).
51 Fitzsimmons, J., Fitzsimmons, M., 2004. Service Management, second ed.
McGrawHill/Irwin, New York.
52 Jarossy, A., 2006. Bigger, better, more: What’s in the works for cruise shipping. Marine
Log, Thursday, June 1.
53 Ransley, J., Ingram, H., 2001. What is ‘‘good’’ hotel design? Facilities 19 (1/2), 79–87.
54 Pullman, Madeleine E., Rodgers, Svetlana. (2010). International Journal of Hospitality
Management. Capacity management for hospitality and tourism: A review of current
approaches. March 2010. Volume 29(1): 177-187.
https://www.researchgate.net/publication/247123990_Capacity_management_for_hos
pitality_and_tourism_A_review_of_current_approaches (Accessed 8 November 2019).
55 Hertzfeld, Esther. (2018). Connectivity helps hotels manage energy use. Hotel
Management. January 11, 2018. https://www.hotelmanagement.net/tech/why-
connectivity-helps-energy-management (Accessed 14 November 2019).
56 Britt, P. (2013) Big Data Means Big Benefits for Entertainment: Caesars
Exec, http://loyalty360.org/resources/article/big-data-means-big-benefits-for-
entertainment-caesers-exec, accessed 5 January 2016.
57 Marr, Bernard. May 2, 2016. Big Data in Practice. John Wiley & Sons.
58 Tracinski, Rob. October 13, 2016. How Not to Predict the Future. Real Clear Future.
http://www.realclearfuture.com/articles/2016/10/13/how_not_to_predict_the_future_
111945.html
59 http://tvtropes.org/pmwiki/pmwiki.php/Main/Zeerust
60 Bartleby. (2019). The Economist. What businesses can learn from the arts. December
8, 2019. https://www.economist.com/business/2019/12/08/what-businesses-can-learn-
from-the-
arts?cid1=cust/dailypicks1/n/bl/n/2019129n/owned/n/n/dailypicks1/n/n/AP/357850/n
(Accessed 10 December 2019).

43
44
 THE A.I. CRUISE LINE

CHAPTER TWO: PERSONALIZATION

“Personalization is table stakes for today’s cruise lines, who are
increasingly competing to be relevant in the hearts and minds
of shoppers.”
~Giselle Abramovich, Adobe

Overview
Today, “Personalization” — the process of utilizing geo-location, mobile app, Wi-
Fi, and OTT technology to tailor messages or experiences to an individual
interacting with them — is becoming the optimum word in a radically new
customer intelligence environment. Even though this personalization comes at a
price — privacy — it is a price most consumers seem more than willing to pay if
a recognized value is received in return.
For a cruise line, “personalization” requires an investment in software analytics,
but retail companies should recognize that this price must be paid because highly
sophisticated consumers will soon need an exceptional shopping experience to
keep them from visiting a rival’s stores (that will, undoubtedly, offer such
services).
To get ahead in this highly competitive industry, cruise lines are recognizing the
importance of personalization when it comes to customer interactions. Most
cruise lines today have customer loyalty programs that are a part of a CRM
and/or a SCRM initiative that provides their guests with an intimate experience
that will make them want to return to the cruise line’s store again and again and
again.
Currently, however, there is a big disconnect between what companies think
they are delivering in terms of personalization and what consumers are actually
experiencing. In his article Study finds marketers are prioritizing
personalization… but are further behind then they realize 61, Andrew Jones argues
that, “Although two-thirds of the marketers surveyed rate their personalization
efforts as ‘very good’ or ‘excellent,’ just 31 percent of consumers reported that
companies are consistently delivering personalized experiences.” 61 Although
these numbers might look bad, they offer a lot of potential and opportunity for
companies that get personalization right.
“Aside from this disparity, the report finds that personalization strategies today
are immature. It shows that 91 percent of the marketers surveyed are prioritizing
personalization over the coming year, yet many still rely on basic segmentation
strategies,” notes Jones.61 This isn’t that surprising as many companies are

45
 ANDREW W. PEARSON

struggling with the ability to not just capture the information necessary for
personalization, but also creating data warehouses that can silo the data
properly, then deliver it to highly complex analytical programs that can make
sense of all that data. It’s like finding a needle in a haystack for each and every
customer in a massive database; an herculean task, no doubt.61

WHICH THREE DIGITAL-RELATED AREAS ARE THE TOP PRIORITIES FOR

YOUR ORGANIZATION IN 2018? (REGIONAL COMPARISON)

28%
Targeting and personalization 34%
34%
24%
Customer journey management 24%
28%
15%
Conversion rate optimization 25%
25%
16%
Mobile optimization 24%
24%
19%
Content marketing 22%
19%
36%
Social media engagement 18%
18%
13%
Video content 14%
18%
18%
Multichannel campaign management 24%
16%
31%
Brand building / viral marketing 19%
15%
13%
Marketing automation 18%
15%
9%
Search engine marketing 12%
13%
6%
Customer scoring and predictive marketing 8%
12%
6%
Joining up online and offilne data 14%
12%
13%
Content management 13%
9%
12%
Mobile app engagement 7%
9%
5%
Progammatic buying / optimization 5%
7%
6%
Real-time marketing 5%
6%
16%
Social media analytics 4%
4%
1%
voice interfaces 1%
1%

0% 5% 10% 15% 20% 25% 30% 35% 40%

Asia Pacific Europe North America

Figure 1: Organizations top priorities in 2018?

Source: Adobe 2018 Digital Trends in Retail62

46
 THE A.I. CRUISE LINE

Figure 1 “compares the top digital-related 2018 priorities for cruise lines across
regions, with targeting and personalization leading the way in both North
America and Europe. While not explicitly mentioned, the theme of data again
looms large for cruise lines seeking to personalize and target effectively.”62
It is obvious that creating a consolidated customer view is a necessary
component of personalization, but, unfortunately, “most marketers today are
working with customer data that is decentralized, spread across the organization
in multiple databases that are updated in batch processes. To find success,
marketers must prioritize consolidating data into a single database,” states
Jones.61
In its 2018 Digital Trends in Retail, Adobe revealed that “the most exciting
prospect through the lens of the cruise line is delivering personalized
experiences in real time, cited by 37% of retail respondents compared to 36% for
client-side respondents in other sectors.”62 This is an interesting survey to keep
in mind as cruise lines are often on the cutting edge of technology. Once they
start receiving personalization marketing from the cruise lines they buy from,
customers will expect similar service from just about every other business they
deal with from that moment on.
Adobe’s 2018 Digital Trends in Retail62 reveals that “cruise lines in Asia are much
more focused on social media engagement and brand-building/viral marketing
than their counterparts in the West, suggesting that the social and viral
marketing opportunity is disproportionally higher in Asia where social uptake has
not hit the same kind of plateau as it has in Western markets.” 62
One big cultural difference between the Asian and North American markets is
the impact of messaging apps.62 “Prompted by the launch of brand-friendly
Official Accounts on WeChat in 2013, the potential of messaging apps in retail
has been embraced more quickly by brands and consumers in China than in the
United States, where conversational commerce has been relatively slow to get
off the ground.”62
As Adobe’s 2018 Digital Trends in Retail discovered, “Retailers recognize that the
quality of the customer experience will increasingly depend on being able to
serve up the most relevant content and messaging at the right moment, with
companies embracing predictive analytics to help them anticipate the most
effective way of converting prospects into customers, and then meeting their
needs on an ongoing basis.”62
Adobe’s 2018 Digital Trends in Retail reveals that, “The appeal of real-time
personalization suggests a focus on providing the most engaging and relevant
experiences, a trend that cuts across numerous digital marketing techniques,
including analytics, marketing automation, programmatic ad buying and

47
 ANDREW W. PEARSON

dynamic content.”62

WHICH ONE AREA IS THE SINGLE MOST EXCITING OPPORTUNITY FOR YOUR
ORGANIZATION IN 2018 (RETAIL VS. OTHER SECTORS)

Delivering personalized experiences in real []

time []
Utilizing AI / bots to drive campaigns and 16%
experiences 18%
Engaging audiences through virtual or 14%
augmented reality 15%
Internet of Things / connected devices, e.g., 13%
wearables, audience tracking 13%
Enhanced payment technologies, e.g. mobile 9%
wallets, e-receipts 9%
Voice interfaces, e.g. Amazon Echo, Google 6%
Home 6%
1%
Other
3%

Agency respondents Company respondents

Figure 2: Most Exciting Prospects in Three Years’ Time?

Source: Adobe 2018 Digital Trends in Retail62

“While a range of potential game-changing technological trends in Figure 2 will

undoubtedly have a powerful impact, from the Internet of Things and connected
devices, to voice interfaces and augmented reality, cruise lines are
predominantly focused on creating a relevant, timely and engaging experience
to each of their users, to maximize sales and efficiency.” 62
AI is an example of an emerging technology that can itself help to make the
experience more relevant and personalized.62 “AI-powered machine learning can
increasingly help cruise lines comb through vast quantities of data to provide the
best possible content and recommendations to consumers as they progress
through the shipping journey from awareness and discovery to conversion.” 62
According to Michael Klein in his article Machine Learning and AI: If Only My
Computer Had a Brain Wired for Business 63, “AI helps cruise lines by serving as
an adaptive, automated interface for customer interaction. Similar to a human
interaction, this interface can work with customers to resolve issues, route
deeper concerns to the right people, and offer personalized recommendations.”
“This is because AI can act on real-time insights supplied from databases that

48
 THE A.I. CRUISE LINE

house a user’s browsing history, past purchases, and demographics,” explains

Klein.63 “Understanding this data opens opportunities for more personalized
targeting, and AI can adapt automated approaches in real time to turn shoppers
into buyers,” says Klein.63 This is going to be a big deal as, according to Tractica64,
global revenue resulting from AI technologies just in the retail sector alone is
expected to top $36.8 billion by 2025.
One thing that was surprising to the researchers was the lack of interest in voice
technology, with only 6% of respondents pointing to voice interfaces as the most
exciting opportunity.62 “The popularity of voice assistants offered by the likes of
Amazon, Google, Microsoft and Apple give retail brands the chance to increase
their presence, including in homes and cars, provided that they can find the right
kind of utility to consumers at the right time.”62
In her 2019 article 8 Things to Expect from CES, Consumer Tech’s Big Shindig 65,
Lauren Goode points out that, “There are now over 20,000 smart devices
compatible with Alexa, and over 10,000 that work with Google Assistant. CES
2019 will undoubtedly be a noisy cacophony of voice-controlled devices, ranging
from refrigerators to sound systems to smart lights in the home, to wearables
and cars outside of the home.”
Goode rightfully claims that “if you add another voice assistant to an existing
product, you can call it ‘new.’”65 Snarkiness aside, this is the wave of the future.
People are getting comfortable talking and giving instructions to smart devices.
Goode concludes the article by pointing out a common problem: “The question
around voice technology, though, isn’t so much whether it will have a presence;
the question is whether it will grow more seamless and less awkward this
year.”65
Bright Local, an SEO platform used by thousands of businesses, produced an
interesting study66 on the potential of voice search, which found:
• 58% of consumers have used voice search to find local business
information in the last 12 months.
• 46% of voice search users look for a local business on a daily basis.
• What consumers want most: to be able to use voice search to make
reservations, to hear business prices, and to find out which products
businesses have.
• 27% visit the website of a local business after making a voice search.
• 25% of consumers say they haven’t yet tried local voice search but
would consider it.
• 76% of smart speaker users perform local searches at least weekly —
with 53% searching using these devices every day.
• Consumers are most likely to perform voice searches to find further
information on local businesses they already know about.

49
 ANDREW W. PEARSON

• Voice searchers are most likely to look for restaurants, grocery stores,
and food delivery.
• Just 18% of consumers have used smart speakers for local voice
searches.
It should be obvious that creating a consolidated customer view is a necessary
component of personalization, but, unfortunately, “most marketers today are
working with customer data that is decentralized, spread across the organization
in multiple databases that are updated in batch processes. To find success,
marketers must prioritize consolidating data into a single database,” states
Jones.61
Psychographics is the study and classification of people according to their
attitudes, aspirations, and other psychological criteria, especially in market
research. As data about people and their behaviors becomes more abundant,
this line of research will become more and more important for customer
intelligence. The Cambridge Analytica-Facebook scandal is only now starting to
reveal how powerful this kind of information is and, in chapter five, I delve
further into this fascinating subject.

Identity-related data sources used for

personalization

Email address 57%

45%
Location 41%
40%
Cookies 34%
33%
Social ID 30%
25%
Device ID 22%
22%
Postal address 20%
18%
Location-related data 18%
17%
Specific to your business 15%
15%
Lifestyle Details 15%
13%
Psychographics 8%
5%
Other (please specify) 3%

Figure 3: Identity-related data sources used for personalization

Source: VB Insights67

50
 THE A.I. CRUISE LINE

Figure 3 lists the identity-related data sources that can be used for
personalization and it is a considerable amount of data that must be culled
through, siloed, and understood for personalization marketing to work properly
and effectively.
Another important step to bringing personalization efforts up to a user’s
expectation level will be by using behavioral data. “In order to create these types
of customer experiences, marketers must strategically collect and utilize
customer data, including real-time signals of intent, which are typically not
captured today,” argues Jones.61

Email data 57.1%

CRM data 54.8%
Search engine data 50.8%
Social media profile data 49.0%
Proprietary transaction data 40.3%
Digital ad tracking 39.5%
Third Party transactional data 36.2%
Loyalty program data 35.5%
Ratings data 34.7%
Geo-spatial/location data 30.9%
Mobile app data 30.6%
Cookie and pixel tracking 29.3%
Free text data from chat systems and reviews 15.1%
Internet of things 14.8%
Imagery and video analysis 8.4%
We do not gather data on our customers 7.4%
Other 5.4%

Which of the Following Do You Gather to Generate Insight into Your

Customers?

Figure 4: State of Data in Travel Survey, 2017

Source: eyefortravel.com68

Figure 4 shows the current data types that a typical travel company might be
collecting and utilizing.
With customer attitudes towards personalized content being shaped by today’s
online recommendation engines, consumers are becoming more used to
receiving what they want, when they want it, and on whatever channel they

51
 ANDREW W. PEARSON

want it consumed on.61 Cruise lines must keep this in mind when developing
personalization programs. The consumer has become highly sophisticated and
he or she expects the level of sophistication received on platforms like Amazon,
Pandora, Spotify, and Netflix to filter over to all their other business
communications; don’t waste a customer’s time with non-matching offers or he
or she will go down the street to a competitor’s store.
According to VB Insights, “Email is the dominant channel for personalized
content, yet is often limited to field insertion (e.g. “Dear ”). Most personalization
efforts are also based on transaction history and limited demographic data,
meaning personalization is not done to a high degree in most cases.”67 Figure 5
breaks down the different digital channels that brands can utilize to connect with
their customers.

Digital channels in which personalized messages/experiences are

delivered

Email 80%
Social media messaging 42%
Web: landing page 37%
Web: home page 36%
Web: content 34%
Mobile web 26%
Advertising: display 26%
Advertising: search 24%
Web media 24%
Advertising: social 24%
Web products/ecommerce 24%
Mobile: SMS 22%
Community 19%
Advertising: retargeting 19%
Mobile in-app messaging 19%
Web: dialog/chat 12%
Digital signs 7%
Other (please specify) 3%
None of the above 3%

Figure 5: Digital Channels for personalized messages/experiences

Source: VB Insights67

52
 THE A.I. CRUISE LINE

In her article 3 AI-driven strategies for cruise lines in 201969, Giselle Abramovich
claims that, “Personalization is table stakes for today’s cruise lines, who are
increasingly competing to be relevant in the hearts and minds of shoppers.” This
is a great analogy as personalization will soon be the base level upon which
customers will accept marketing from the companies they choose to buy from.
VB Insights claims that, “Although we may think of the sales process as ‘top-
down,’ most companies implement personalization with a bottom-up approach.
That is, most companies begin their personalization efforts based on ‘known’
prospects or customers. The channels where personalization is being employed
reinforce this finding.”67
In its article Creating the Ultimate Single Customer View 70, the Adobe Experience
Cloud team argues that, real-time access to data enables an “organization to
trigger personalized messages and outreach in the moment of highest impact.
What’s more, leveraging a variety of signals emerging from the buying process,
marketers can engage with a customer when it’s both most relevant for her, and
when she’s most likely to convert — maximizing the value on both sides.”
Of course, not all campaign management systems are alike or have the
functionality that helps businesses deliver experiences at the moment of highest
impact.70 Systems that have lags in data access due to third-party partnerships
and integrations are especially susceptible to problems as they are forced to do
double duty — first, they have to remind customers how they felt earlier and,
second, they have to encourage shoppers to act based on those earlier
experiences and emotions; this is far from ideal.70
In its Creating the Single Customer View with Adobe Campaign 71, the Adobe
Experience Cloud team recommends businesses “Rely on sales-centric campaign
management tools and you’ll be hard-pressed to create these single views, let
alone construct meaningful mosaics that adapt and evolve in real time. And if
you can’t capture the granular details surrounding customer interactions — if
you can’t understand the data you do have — it’s virtually impossible to deliver
personalized experiences at scale and build a loyal customer base.”
“Personalization is critical to any cross-channel strategy — and at the heart of
any personalization strategy is the ability to segment,” argues the Adobe
Experience Cloud team.71 Unsurprisingly, “Tapping into more complex
segmentation strategies helps organizations deliver better, more meaningful
cross-channel experiences.”71
AI and machine learning a whole new level of depth and detail to customer
segmentation modeling. “Being able to easily create control vs. test groups
based on nuanced criteria helps arrive at the insight necessary to design optimal
experiences for different sets of customers,” argues the Adobe Experience Cloud
team.71 “Applying the same nuanced criteria to the delivery of those experiences
is how that insight is transformed into personalization at scale,” they contend.71

53
 ANDREW W. PEARSON

Adobe argues that the numbers don’t lie — segmented and targeted emails
generate 58 percent of all revenue.71 That’s total revenue, not lift, which is a
highly impressive number.
“Without advanced filtering it’s virtually impossible to extract detailed data and
uncover the nuances behind the numbers. Beyond that, though, creating and
managing lists also becomes a challenge,” warns Adobe. 71 “Want to target
customers based on their preferred device? If filtering isn’t native to your
campaign management system, that simple task is going to be time-consuming
and costly at best. Personalization becomes a trade-off between quality and
speed,” says Adobe.71
“Businesses should also focus on solutions that utilize artificial intelligence (AI)
in a tangible and effective way,” argues Adobe.71 At best, “AI can take the grunt
work of data stitching, data cleaning, and anomaly detection off your plate,
freeing you up for more meaningful marketing work — gaining a better
understanding of customer wants and needs, for example, then spending time
designing perfectly personalized experiences,” recommends Adobe. 71
Abby Parasnis, Adobe’s chief technology officer argues that Adobe Sensei “gives
marketers and analysts new visibility into which segments are most important to
their businesses, and allows them to target overlapping or adjacent segments,
making it possible to acquire customers much more efficiently.” 71
“By having an integrated customer profile that combines online and offline data,
marketers can more easily provide truly meaningful customer experiences that
reinforce the brand message across all channels,” says Bruce Swan, senior
product manager for Adobe Campaign. 71 “The results include increased
engagement as well as a higher likelihood for conversion, long-term loyalty, and
brand advocacy,” adds Swan.71
A unified or single customer view can help marketers “harvest the insights they
need to develop targeted marketing campaigns — that, in turn, drives customer
loyalty, purchases, and conversions.”71 It is a virtuous cycle that feeds upon itself,
as long as the customer continues to see the value in loyalty. “Data-driven
marketing also speeds time-to-market, and reduces overall campaign costs,”
believes the Adobe team.71
Ultimately, Adobe concludes that, “it comes down to one key consideration:
your customers deserve to be treated like individuals — and you need to deliver.
You need to collect cross-channel insights that can be pulled together into a
cohesive single view. You need to have the capabilities to adjust that view in real
time, as your consumers pivot — and even change course. And you need both
the powerful insights and powerful technology to drive consistent, cohesive, and
meaningful cross-channel journeys for every customer.”71
The A.I. Cruise line can personalize the customer experience in the following

54
 THE A.I. CRUISE LINE

ways:
• Customer Service:
o Geo-locating a customer when he or she signs onto a cruise
line’s customer Wi-Fi system.
o Video analytics with facial recognition technology to spot
and/or confirm a customer’s identity.
o Social media customer service can cut down on normal
customer service expenses, as well as connect with customers
on the channels that they prefer, i.e., Facebook, WhatsApp,
WeChat, or Instagram.
o Chatbots can automate customer service requests, as well as
disseminate info seamlessly.
• E-Commerce
o Clickstream analysis can allow personalized offers to be sent to
a potentially returning customer when he or she is browsing
the company’s website or making a purchase.
• Customer Management:
o The ecommerce department can get more accurate attribution
analysis — so that the cruise line can understand which
advertising is associated with which user, making the
advertising more quantifiable and, therefore, much more
actionable.
o CRM systems can add social media as a channel feeding
targeted messages to only those customers who are most
likely to respond to them.
o The amount of promotions available and channels through
which to market through increases exponentially as campaign
lift can be assessed in terms of hours, rather than in terms of
days or weeks.
o Customer acquisition is accelerated because business users
throughout the company can quickly derive answers to the
following questions:
▪ Which combinations of campaigns accelerate
conversion?
▪ What behavior signals churn?
▪ Do web search key words influence deal size?
▪ Which product features do users struggle with?
▪ Which product features drive product adoption and
renewal?
▪ What drives customers to use costly sales channels?
o Customer interaction data can rapidly be turned into business
opportunities.

55
 ANDREW W. PEARSON

o Powerful recommendation engines can ingest data from a

multitude of sources and then be made available to frontline
staff, who can react in near real-time to customer requests.
• Point-of-Sale:
o Brands can better target merchandise, sales, and promotions
and help redesign store layouts and product placement to
improve the customer’s experience.
In its Pega Marketing Product Overview72, Pega Systems explains that Next-Best-
Action (NBA) marketing is a “Customer-centric approach to marketing that
combines traditional business rules with predictive and adaptive analytics to
provide real-time and batch marketing offers and treatments to drive customer
lifetime value.” “This includes the capabilities to allow marketers to create a
cross-channel engagement strategy that continuously look at customer history
and many different customizable attributes to determine the top offer, best
time, specific treatment, and best channel to interact with customers and
prospects.”72
Pega’s solution includes, “An agent-assisted channel application that provides a
best practice implementation of Next-Best-Action for use by call center agents,
retail users and brokers.”72 It “Includes UI panels and decision logic for offer
presentment, bundle negotiation, what-if analysis, needs assessment and
house-holding.”72
Pega’s Paid Media Manager can extend “Next-Best-Action to paid channels
including Facebook Ads, Google Ads, and LinkedIn Ads.”72 “Which offer or offers
to show advertisements to each individual can now be decisioned on based on
sophisticated strategies and AI. The full fidelity of enterprise data within the
firewall can be safely leveraged to power these paid decisions just as they do on
owned channels,” explains Pega.72
The “Paid Media Manager (PMM) enables marketers to leverage enterprise data
and insights to drive intelligent and effective advertising.” 72 “PMM is highly
integrated with NBA Designer. When an NBA decision is made on what offer or
offers an individual is eligible for PMM can then communicate those decisions to
ad platforms by placing customers and prospects in the appropriate
audiences.”72 “Inclusion of audience indicates the offer an individual will be
targeted with as well as the bid level,” explains Pega.72
Historically, ad decisions are based only on broad segmentation models which
are based on anonymous data.72 With PMM, ad decisions can be driven by the
full fidelity of a company’s proprietary enterprise data and the insights that drive
a company’s owned channels.72 “PMM takes a privacy first approach by only
communicating the decision over the firewall to ad platforms. The backing
enterprise data and AI-based insights stay safely within your firewall.”72
“These decisions can be communicated by leveraging direct CRM server-to-

56
 THE A.I. CRUISE LINE

server APIs to Google Ads, Facebook Ads, and LinkedIn Ads,” explains Pega.
“Traditional pixel-based APIs are also supported by Adobe Audience Manager as
well.”72 “PMM also enable segments to be communicated safely and securely to
ad platforms using these same paid channel integrations.”72
In her article The 5 Biggest Marketing Trends for 2019 73, Giselle Abramovich
quotes Stacy Martinet, VP of marketing strategy and communications at Adobe,
saying, “Companies that want to provide truly transformative customer
experiences need customer data that is real-time, intelligent, and predictive.”
Martinet adds that, “In 2019 we’ll see enterprises focused on building a seamless
flow of connected customer data — behavioral, transactional, financial,
operational, and more — to get a true end-to-end view of their customers for
immediate actionability.”73
Giselle Abramovich believes marketers have long been talking about
personalization marketing, but they are still at a very basic level of
personalization.73 “To truly unlock the value of personalization, companies must
first create a unified view of their customers,” contends Anudit Vikram, SVP of
audience solutions at Dun & Bradstreet.73
Jason Heller, partner and global lead, digital marketing operations at McKinsey
agrees, claiming, “The single view of the customer is the single most important
asset that a modern marketer can have, and it’s the core of their personalization
efforts.”73 “It also becomes the core of their next-generation marketing ROI
capability, as well,” he adds.73
One of the keys to personalization at scale is internal structure.73 Heller “expects
companies in 2019 will work on building agile marketing execution models in
which cross-functional teams can experiment, leveraging the data and
technology stack to capture value.”73
“Privacy, of course, will play a big role in an organization’s personalization
strategy,” says Abramovich.73 “New laws such as GDPR — plus California’s
privacy law, which comes into effect in January 2020 — means marketers must
be focused on ensuring ethical data collection practices and earning consumers’
trust,” claims Martinet.73
“When choosing partners to work with, brands need to look for products and
services that protect the data that is entrusted to them and are designed with
privacy in mind,” says Martinet.73 “Privacy is about respecting your customers
and giving them control over how their data is being used. Be transparent and
help them understand the value proposition,” adds Martinet. 73
Abramovich believes that, in 2019, many organizations will have what McKinsey
refers to as a “consent management” function, which includes “having an ethical
view of how the organization manages customers’ data, protects that data, and
establishes governance around how that data is utilized.” 73

57
 ANDREW W. PEARSON

“I think this is an absolute obligation that we have regardless of whether the

regulations exist or not because eventually they will exist,” says Heller. 73 “So
starting to operate that way today will only set you up for more success in the
future.”73
In terms of privacy, as reported in CB Insights What’s Next in AI? Artificial
Intelligence Trends74, Google’s federated learning approach aims at adding a
layer of privacy by utilizing a person’s mobile messaging while also keeping it
private. “In a nutshell, your data stays on your phone. It is not sent to or stored
in a central cloud server. A cloud server sends the most updated version of an
algorithm — called the ‘global state of the algorithm’ — to a random selection
of user devices.’”74
“Your phone makes improvements and updates to the model based on your
localized data. Only this update (and updates from other users) are sent back to
the cloud to improve the “global state” and the process repeats itself,” explains
CB Insights.74 Real world examples include Firefox’s use of federated learning to
“rank suggestions that appear when a user starts typing into the URL bar.”
Google Ventures-backed AI startup OWKIN is using the approach to protect
sensitive patient drug discovery data.74 “The model allows different cancer
treatment centers to collaborate without patients’ data ever leaving the
premises,” claims CB Insights.74
According to Adobe’s Indelible content, incredible experiences75, “Marketers
want to surface the right content precisely when and where customers need it.
But to be efficient, you want to create once and deliver everywhere, with
content automatically adjusting to fit connected experiences on any channel.
Machine learning lets you do that — finding better ways to optimize layout and
copy wherever they’re used.” For example, “Adobe’s smart summarisation can
take your product manager’s blog post about gourmet hot dogs and trim the
redundant content for a news clip or email.”75

Customer Relationship Management

As previously mentioned, CRM is a strategy used to learn more about a
customer’s needs and behaviors in order to develop a stronger relationship with
him or her, thereby creating a value exchange on both sides.
As Lovelock and Wirtz state in Services Marketing, People, Technology,
Strategy76, “from a customer perspective, well-implemented CRM systems can
offer a unified customer interface that delivers customization and
personalization.” Lovelock and Wirtz argue that at each transaction point, such
relevant patron data as a customer's personal preferences, as well as his or her
overall past history transactions are available to the clerk serving the customer,
giving them valuable information about how to interact with that person.76 This

58
 THE A.I. CRUISE LINE

is not an easy thing to do, however, especially when unstructured data like social
media feeds are added to the mix.
According to Lovelock and Wirtz, most CRM solutions contain the following
stages76:
• Data collection: the system captures customer contact details, such as
demographics, purchasing history, service preferences, etc.
• Data analysis: data captured is analyzed and categorized into a unique
set of criteria. This information is then used to tier the customer base
and tailor service delivery accordingly.
• Sales force automation: sales leads, cross-sell, and up-sell opportunities
can be effectively identified and processed, and the entire cycle from
lead generation to close of sales and after-sales service can be tracked
and facilitated through the CRM system.
• Marketing automation: the mining of customer data can help a
company achieve one-to-one marketing to each one of its customers.
Loyalty and retention programs can reduce costs, which can result in an
increase of marketing expenditure ROI. By analyzing campaign
responses, CRM systems can easily assess a marketing campaign's
quantifiable success rate.
• Call center automation: with customer information available right at
their fingertips, call center staff can improve customer service levels
because they will be able to immediately identify a customer's tier level,
as well as compare and contrast him or her against similar customers so
that only promotions that are likely to be accepted are offered.
Most cruise lines will have plenty of data collection, data analysis, sales force
automation, marketing automation, and call center automation software to help
them in their CRM endeavors, but it is not easy getting all of these complicated
systems and processes working together to provide a level of personalized
service that wows the customer.
Beyond simple CRM (which, I guess, is never really that simple), Social CRM
(SCRM) adds a whole new level of sophistication and complexity to the mix.
SCRM is the use of “social media services, techniques and technology to enable
organizations to engage with customers.”77 In his article Time to Put a Stake in
the Ground on Social CRM78, Paul Greenberg states that:

“Social CRM is a philosophy and a business strategy, supported

by a technology platform, business rules, workflow, processes
and social characteristics, designed to engage and react
accordingly in a collaborative conversation in order to promote
mutually beneficial value in a trusted and transparent business
environment. It’s the company’s response to the customer’s

59
 ANDREW W. PEARSON

ownership of the conversation.”

One aspect of “Social CRM” is “Social Media Monitoring,” the process by which
companies monitor sites like Facebook, Twitter, LinkedIn, Weibo, YouTube,
Instagram, and others for relevant brand and anti-brand comments and
mentions. Social media monitoring tools allow for continuous and strong
customer engagement.
When it comes to social media and implementing it into a cruise line, the one
constant question should be is, “How does this affect my ROI?” For many cruise
lines, there is the sense that social media is an ethereal, unquantifiable thing,
but this shouldn’t be the case. As Figure 6 shows, social media listening can be
used in a multitude of ways, like anticipating customer problems, understanding
and identifying sentiment, measuring a company’s share of voice, as well as
keeping track of a company’s brand. All of these are important in their own right
and, together, they can give a cruise line deep detail into a marketing campaign’s
performance and assist quantifying attribution analysis, which can help with
planning and implementing future marketing campaigns.

Figure 6: Social Media Listening objectives

Source: www.intelligencia.co

A good example of how a company can test whether a social media solution
would work for it is to consider the experience of a telecommunication company
that proactively adopted social media recently, as mentioned in R.E. Divol’s
article Demystifying social media79. “The company had launched Twitter-based
customer service capabilities, several promotional campaigns built around social
contests, a fan page with discounts and tech tips, and an active response

60
 THE A.I. CRUISE LINE

program to engage with people speaking with the brand.”79

In social-media terms, the investment was not insignificant, and the company’s
senior executives wanted quantifiable ROI not anecdotal evidence that the
strategy was paying off.79 “As a starting point, to ensure that the company was
doing a quality job designing and executing its social presence, it benchmarked
its efforts against approaches used by other companies known to be successful
in social media.”79 According to Divol, “the telecommunication company
advanced the following hypotheses79:
• “If all of these social-media activities improve general service
perceptions about the brand, that improvement should be reflected in
a higher volume of positive online posts.
• “If social sharing is effective, added clicks and traffic should result in
higher search placements.
• “If both of these assumption hold true, social-media activity should help
drive sales — ideally, at a rate even higher than the company achieves
with its average gross rating point (GRP) of advertising expenditures.” 79
The company tested its options. “At various times, it spent less money on
conventional advertising, especially as social-media activity ramped up, and it
modeled the rising positive sentiment and higher search positions just as it
would using traditional metrics.”79
The results were conclusive: “social-media activity not only boosted sales but
also had higher ROIs than traditional marketing did. Thus, while the company
took a risk by shifting emphasis toward social-media efforts before it had data
confirming that this was the correct course, the bet paid off.” 79 Just as
importantly, the company had now created an analytic baseline that gave it the
confidence to continue exploring a growing role for social media.79 It is very easy
to quantify search rankings and it is pretty obvious that if a cruise line ranks
higher in Google search, it should garner more business.
CRM is an integral part of what businesses hope will be a value exchange on both
sides of the customer-company equation, one that will, hopefully, create loyal
customers who become apostles for the business.
Lovelock and Wirtz created the “Wheel of Loyalty” as an organizing structure to
help businesses build customer loyalty and it is highly relevant to the retail
industry.76 The first of its three sequential steps include building a foundation for
loyalty, including “targeting the right portfolio of customer segments, attracting
the right customers, tiering the service, and delivering high levels of
satisfaction.”76
The second step — creating loyalty bonds that either deepen the relationship
through cross-selling and bundling or adding value to the customer through
loyalty rewards and higher level bonds — can be achieved by the cruise lines

61
 ANDREW W. PEARSON

gaining a fuller understanding of the patron. 76 It is important to understand as

much about the patron as possible, his wants, desires and needs, all the way
down to his preferred purchase items and any other preference he or she might
want to share with the cruise line.
The third factor — identify and reduce the factors that result in “churn” — is also
extremely important to a cruise line’s bottom line.7679 Engagement is paramount
here and mobile apps and social media are great channels to keep customers
interested and engaged.
Cruise lines should also feel compelled to reward their customers through
Facebook, Twitter, WeChat, Line, WhatsApp, Instagram, and Weibo, or any
number of social network, blogging, and/or micro-blogging services. The beauty
of using these channels is the ability of the customer to share these awards or
stories of these awards with friends and family.
It shouldn’t be too hard to get patrons to share their social media accounts,
either, as a cruise line can ask patrons for their social media accounts at sign up
through methods likes social bridging. Social media is now often a preferred
contact channel and it does make connecting with users in a real-time way
exceptionally easy.
According to Jones and Sasser, “the satisfaction-loyalty relationship can be
divided into three main zones: Defection, indifference, and affection. The zone
of defection occurs at low satisfaction levels. Customers will switch unless
switching costs are high or there are no viable or convenient alternatives.” 80 This,
obviously, isn’t the case with a cruise line, where switching often constitutes
little more than browsing to a competing cruise line's website or walking into
another cruise line’s store, often in the same mall. With the vast echo chamber
of social media against them, losing only one disgruntled patron could be the
least of the cruise line’s problems.
Jones and Sasser warn that, “Extremely dissatisfied customers can turn into
‘terrorists,’ providing an abundance of negative feedback about the service
provider.”80 Through social media channels, negative feedback can reverberate
around the world within seconds. Today, more than ever, cruise lines must spot
dissatisfied customers and approach them before they do irreparable harm to
the company’s image and reputation and social media is one of the best channels
in which to engage them. Like the proverbial canary in the coal mine, the A.I.
Cruise line will have systems set in place that can warn the business about these
customers before they become brand terrorists.
Cruise lines need to empower their patrons to post on Facebook or WeChat or
Weibo or Twitter or comment about their cruise line experiences and, hopefully,
turn them into apostles.
In Jones and Sasser’s zone of affection, satisfaction levels are high and

62
 THE A.I. CRUISE LINE

“customers may have such high attitudinal loyalty that they don’t look for
alternative service.”80 It is within this group that “Apostles” — members who
praise the firm in public — reside and this is the group responsible for improved
future business performance.81 The A.I. Cruise Line will not only be able to spot
these apostles, but also understand them on such a unique and personal level
that their loyalty and patronage will almost be guaranteed.
As Darrell Rigby explains in Bain & Company’s Management Tools 2015 An
Executive’s Guide82, CRM “is a process companies use to understand their
customer groups and respond quickly — and at times, instantly — to shifting
customer desires. CRM technology allows firms to collect and manage large
amounts of customer data and then carry out strategies based on that
information.”82
Retailer operators can utilize CRM to:
• Create databases of customers segmented into buckets that allow more
effective marketing.
• Generate more accurate sales leads.
• Gather market research on customers.
• Rapidly coordinate information between the sales and marketing staff
and front-facing hosts and reps, thereby increasing the customer
experience.
• Enable reps to see and understand the financial impact of different
product configurations before they set prices.
• Accurately gauge the return on individual promotional programs and
the effect of integrated marketing activities, and redirect spending
accordingly.
• Accumulate data on customer preferences and problems for product
and service designers.
• Increase sales by systematically identifying, managing, and automating
sales leads.
• Improve customer retention by uncovering the reason(s) for customer
churn.
• Design proactive customer service programs.
Today, CRM is evolving into what has been dubbed “Customer Centric
Relationship Management” (CCRM), a style of CRM that focuses on customer
preferences above all else. CCRM attempts to understand the client in a deep,
behavioral way, and it engages customers in individual, interactive relationships
through tailored marketing and one-to-one customer service. This
personalization can help a cruise line retain customers, build brand loyalty, as
well as provide customers not only with the information that they really want
but also with the rewards that they might actually use. Today’s technology allows
cruise lines to not only surface the information that they need to know about

63
 ANDREW W. PEARSON

their customers, but it can also provide front-facing employees with offers that
these customers will actually like and, therefore, probably use.
In Bain & Company’s Management Tools 2015 An Executive’s Guide 82, Darrell K.
Rigby claims CRM requires managers to:
1. “Start by defining strategic ‘pain points’ in the customer relationship
cycle. These are problems that have a large impact on customer
satisfaction and loyalty, where solutions would lead to superior
financial rewards and competitive advantage.”82
2. “Evaluate whether — and what kind of — CRM data can fix those pain
points.”82
3. “Calculate the value that such information would bring the company.”82
4. “Select the appropriate technology platform, and calculate the cost of
implementing it and training employees to use it.”82
5. “Assess whether the benefits of the CRM information outweigh the
expenses involved.”82
6. “Design incentive programs to ensure that personnel are encouraged to
participate in the CRM program. Many companies have discovered that
realigning the organization away from product groups and toward a
customer-centered structure improves the success of CRM.”82
7. “Measure CRM progress and impact. Aggressively monitor participation
of key personnel in the CRM program. In addition, put measurement
systems in place to track the improvement in customer profitability
with the use of CRM. Once the data is collected, share the information
widely with employees to encourage further participation in the
program.” 82

Segmentation
Once a cruise line implements a CRM program, data segmentation can begin.
According to Wikipedia, market segmentation “is the process of dividing a broad
consumer or business market, normally consisting of existing and potential
customers, into sub-groups of consumers (known as segments) based on some
type of shared characteristics.”83
In dividing or segmenting markets, cruise lines can look for shared
characteristics. Market segmentation tries to identify high yield segments — i.e.,
those segments that are likely to be the most profitable or that have outsized
growth potential — so that these can be selected for special attention (i.e.,
become target markets).
Rigby states that customer segmentation “is the subdivision of a market into
discrete customer groups that share similar characteristics.”82 He adds:
“Customer Segmentation can be a powerful means to identify unmet customer
needs. Companies that identify underserved segments can then outperform the

64
 THE A.I. CRUISE LINE

competition by developing uniquely appealing products and services.” 82 Rigby

believes that customer segmentation is most effective when a company can
discover its most profitable segments and then tailor offerings to them, thereby
providing the customer with a distinct competitive advantage. 82
As Rigby points out, customer segmentation requires managers to82:
• “Divide the market into meaningful and measurable segments
according to customers’ needs, their past behaviors or their
demographic profiles.”82
• “Determine the profit potential of each segment by analyzing the
revenue and cost impacts of serving each segment.”82
• “Target segments according to their profit potential and the company’s
ability to serve them in a proprietary way.”82
• “Invest resources to tailor product, service, marketing and distribution
programs to match the needs of each target segment.”82
• “Measure performance of each segment and adjust the segmentation
approach over time as market conditions change decision making
throughout the organization.”82
For a cruise line, the pain points might be things like customer loyalty and the
marketing department should be asking things like, “Why does it cost so much
money to retain customers?” “Can we not find cheaper but more meaningful
offers that show understanding of the customer?” Also, “How can we drive
customer loyalty to such a degree that our customers rave about us on social
media?”
Beside the above methodologies, customer segmentation can be used to:
• Prioritize new product development efforts.
• Develop customized marketing programs.
• Choose specific product features.
• Establish appropriate service options.
• Design an optimal distribution strategy.
• Determine appropriate product pricing.
Market segmentation assumes that different market segments require different
marketing programs — that is, different offers, prices, promotion, distribution
or some combination of marketing variables. Market segmentation is not only
designed to identify the most profitable segments, but also to develop profiles
of key segments to better understand their needs and purchase motivations.
Insights from segmentation analysis are subsequently used to support marketing
strategy development and planning.
Many marketers use the S-T-P approach; Segmentation→ Targeting →
Positioning to provide the framework for marketing planning objectives. That is,
a market is segmented, one or more segments are selected for targeting, and

65
 ANDREW W. PEARSON

products or services are positioned in a way that resonates with the selected
target market or markets. With real-time technology, segmentation can reach a
whole new customer experience complexity level.
The process of segmenting the market is deceptively simple. Seven basic steps
describe the entire process, including segmentation, targeting, and positioning.
In practice, however, the task can be very laborious since it involves poring over
loads of data, and it requires a great deal of skill in analysis, interpretation and
some judgment. Although a great deal of analysis needs to be undertaken, and
many decisions need to be made, marketers tend to use the so-called S-T-P
process as a broad framework for simplifying the process outlined here:
• Segmentation:
o Identify market (also known as the universe) to be segmented.
o Identify, select, and apply base or bases to be used in the
segmentation.
o Develop segment profiles.
• Targeting:
o Evaluate each segment's attractiveness.
o Select segment or segments to be targeted.
• Positioning:
o Identify optimal positioning for each segment.
o Develop the marketing program for each segment.
Markets can be broken down into the following segments:
• Geographic segment.
• Demographic segment.
• Psychographic segment.
• Behavioral segment.
• Purchase/usage occasion.
• Generational segment.
• Cultural segmentation.
For the cruise line, customers can also be further segmented into the following
areas:
• Game preference.
• Day of week.
• Time of day.
• Length of session.
• Size of stake.
• Most and least profitable customers.
Although customer segmentation is a common business practice, it has received
the following criticisms:

66
 THE A.I. CRUISE LINE

• That it fails to identify sufficiently meaningful clusters.

• That it is no better than mass marketing at building brands.
• That in competitive markets, segments rarely exhibit major differences
in the way they use brands.
• Geographic/demographic segmentation is overly descriptive and lacks
enough insights into the motivations necessary to drive
communications strategy.
• Difficulties with market dynamics, notably the instability of segments
over time and structural change that leads to segment creep and
membership migration as individuals move from one segment to
another.
Market segmentation has many critics, but, in spite of its limitations, it remains
one of the most enduring concepts in marketing and it continues to be widely
used in practice.
As Wikipedia explains83, there are no formulas for evaluating the attractiveness
of market segments and a good deal of judgment must be exercised.
Nevertheless, a number of considerations can be used to evaluate market
segments for attractiveness, including:
• Segment Size and Growth:
o How large is the market?
o Is the market segment substantial enough to be profitable?
o Segment size can be measured in number of customers, but
superior measures are likely to include sales value or volume.
o Is the market segment growing or contracting?
o What are the indications that growth will be sustained in the long
term? Is any observed growth sustainable?
o Is the segment stable over time?
• Segment Structural Attractiveness:
o To what extent are competitors targeting this market segment?
o Can we carve out a viable position to differentiate from any
competitors?
o How responsive are members of the market segment to the
marketing program?
o Is this market segment reachable and accessible?
• Company Objectives and Resources:
o Is this market segment aligned with the company's operating
philosophy?
o Do we have the resources necessary to enter this market segment?
o Do we have prior experience with this market segment or similar
market segments?
o Do we have the skills and/or know-how to enter this market
segment successfully?

67
 ANDREW W. PEARSON

The latest developments in CRM technology are adding AI to the process. As

explained in the MIT Technology Review article Transform Customer Experience
by Harnessing the Power of AI in CRM84;
“Unlike traditional customer-facing platforms that deliver a
fragmented view of the buyer, an intelligent platform presents
a single aggregated view of customer data. The built-in
intelligence layer helps businesses spot trends, anticipate
needs, and respond more proactively. With that complete
picture, for example, a business knows exactly when the
customer last purchased a product, what that product was,
whether he or she had a problem, and, if so, exactly how it was
resolved.”
MIT Technology Review adds that, “the wide range of machine-learning (ML)
models learns from what is collected to unearth and match patterns, as well as
act on correlations that would otherwise remain hidden.” 84 Using a consumer
shopping experience as an example, the MIT Technology Review article explains
that, “AI models embedded within the CRM system’s personalization engine take
into account the catalog that any given shopper sees and the context on how the
merchant is engaging with that shopper, and then ranks every product for that
buyer in terms of relevance from search results, making the most targeted and
personalized results ever.”84 This can be done at scale, with a constant refining
of shopper recommendations based on the ML, explains the MIT Technology
Review.84
“The data for such recommendations include both historic and real-time click-
stream data from multiple sources.”84 MIT Technology Review warns that, “What
is challenging is that the algorithms used to create predictions are as
heterogeneous as the sources of data used. To deliver predictive
recommendations with the highest accuracy, a range of different algorithms is
applied.”84
“In selecting the right algorithm, the champion-challenger model is used,
meaning that every time an algorithm yields accuracy, it is automatically set to
default over other models within the platform,” states MIT Technology Review
article.84 “That way, the path to personalization is very short,” notes the MIT
Technology Review article. “The end result of such accurate recommendations is
that both customer conversions and the overall potential value of the
merchant’s inventory goes up,” the article concludes.84
Beyond product recommendations, the MIT Technology Review article states
that, “other powerful capabilities that enhance CRM for both employee and
customer experiences include the algorithms for speech recognition, sentiment
analysis, intent, content summarization through natural language processing,
and question answering based on tables of data.”84

68
 THE A.I. CRUISE LINE

“Predicting customers’ future behaviors and needs often turns on the ability to
parse their emotions, more than just their past purchases, and creating a shared
bond,” claims the MIT Technology Review article.84 The article describes a CRM
system that online cruise line Fanatics utilizes to ensure that it understands its
customers in the best possible way, through his or her emotional team bond,
because, as Fanatics puts it, “sports merchandise is an emotional business.” 84
“We want to deliver the most relevant merchandise to you, at the right time, for
your team,” explains Jonathan Wilbur, the company’s director of CRM. 84 “If
you’re a Yankees fan, we want to make sure we’re never showing you anything
Boston Red Sox,” adds Wilbur.84
“The company’s ability to engage with customers around the biggest sporting
events in near real time is unmatched,” claims the MIT Technology Review.84
Fanatics carries merchandise for more than 1,000 professional and college
teams, including from the NFL, MLB, NBA, NHL, NASCAR, and football leagues
from around the world.84 “The company is event-driven, engaging fans around
everything from the World Series to football star Peyton Manning’s retirement
announcement,” explains the MIT Technology Review.84 “Multiply a thousand
teams by an endless stream of sports news events, and you’ve got billions of e-
mails going to fans each year.”84
“In 2015, we sent about 3.5 billion messages,” says Wilbur.84 “When a team wins
the Super Bowl, we can have 350 products live with a press of a button three
seconds after the game,” adds Wilbur.84 His team “built scripts that searched
customer data to display fans’ favorite teams, pulled in real-time scores and stats
from vendor feeds, and personalized branding using partner IDs.” 84
“The resulting campaigns were customized according to multi-tier segments.
Fanatics was able to deliver merchandise relevant to fans and their teams at just
the right time,” explains the MIT Technology Review.84 “Carolina Panther fans
didn’t get e-mails about “Super Bowl Champs” T-shirts after their team lost the
big game, but Denver Broncos fans had offers in their inboxes within minutes
after the final whistle.”84
“Fanatics is even building automatically triggered rules-driven campaigns based
on dynamic information,” notes the MIT Technology Review.84 “Any time a
baseball player hits three homes runs in a game, we’ll send an e-mail featuring
his jersey. Set it and forget it,” Wilbur adds.84 Although hitting three home runs
in a game is a rare event, Fanatics needs to be careful not to overdo it. Fans will,
of course, want to celebrate if such an occurrence happens, but businesses
should be careful not to hit their customers too often. In terms of marketing,
even if it is contextually aware and hyper-personalized, there can be too much
of a good thing.
Companies like Adobe, IBM, Oracle, Microsoft, SAP, Pega Systems,
Salesforce.com, and SugarCRM all have products that not only include contact

69
 ANDREW W. PEARSON

management systems that integrate emails, documents, jobs and faxes, but also
integrate with mobile and social media accounts as well, so the market doesn’t
lack product but this will be a case where one side doesn’t fit all.
Social media can help amplify the “relationship” in “Customer Relationship
Management”, thereby enabling organizations to connect and engage
consumers in a unique way, as well as personalize and monetize customer
relationships on a sustained basis, which should increase profitability. 110 “Social
media also provides a path to richer customer analysis, using technologies
capable of funneling and consolidating customer insights.”110 Insights derived
from this analysis can help companies to “dynamically calibrate, anticipate, and
offer products and services that meet perpetually shifting consumer demands in
a hyper-competitive marketplace.”110
Specifically, for a cruise line, it would be advantageous to link a patron’s
customer account with his or her social media accounts so that the cruise line
could get a heads-up on what a patron might be saying about them on social
media. These channels are also becoming two-way customer service channels
and direct messaging patrons to their social accounts is probably one of the
quickest ways to reach a patron. Of course, cruise lines should tread carefully in
this area, especially in China, but there’s no reason why platforms like Facebook,
Twitter, WhatsApp, Line, WeChat, Instagram, and others couldn’t be prominent
customer connection and/or customer service channels.
In his article Customer Analytics in the Age of Social Media 85 for TDWI, David
Stodder reports that the importance of customer analytics is in the boardroom;
“overwhelmingly, respondents cited giving executive management customer
and market insight (71%) as the most important business benefit that their
organization seeks to achieve from implementing customer analytics.” 85 “This
percentage rises to 81% when survey results are filtered to see only the
responses from those who indicated ‘strong acceptance’ of data-driven
customer analytics over gut feel.”85The second highest benefit cited, at 62%, was
“the ability to react more quickly to changing market conditions, which speaks
to the need for customer data insights to help decision makers address
competitive pressures from rapid product or service commoditization.” 85
With the commoditization of products and services, customer loyalty can be
elusive; innovation must be constant and it should help to reveal why an
organization might be losing its customer base.85 “Information insights from
analytics can help organizations align product and service development with
strategic business objectives for customer loyalty.”85 These insights can also help
an organization be selective about how they deploy their marketing campaigns
and customer-touch processes so that they emphasize features in new products
and services that are important to each specific customer.85
Customer analytics can also provide answers to questions like85:

70
 THE A.I. CRUISE LINE

• When in the life cycle are customers most likely to churn?

• What types of products or services would prevent them from churning?
• When should customers be offered complimentary items?
• When is it too costly to try to keep certain customers?
Cruise lines can realize significant ROI from investing in customer analytics as it
can improve the marketing department’s efficiency, effectiveness 85, and reach.
However, customer analytics ROI is a difficult thing to quantify. Better customer
knowledge equates to more optimized marketing spend because a business can
focus its resources on those campaigns that have the highest predicted chances
of success for particular segments, as well as cutting off or avoiding those that
have the least.85
“By using analytics to eliminate mismatches of campaigns targeting the wrong
customers or using the wrong messages and offers, marketing functions can
reduce wasteful spending and increase gains relative to costs.” 85 Customer
segmentation allows organizations to move “away from one-size-fits-all, brand-
level-only marketing and toward the ‘market of one’: that is, personalized, one-
to-one marketing.”85
Reaching a customization and customer service level that makes a customer feel
as though he or she is a preferred customer is not easy, scaling that up so that
an entire database of customers feel that they are unique and receiving
outstanding customer service is even more challenging. However, in this day and
age of hyper-personalization, it is almost a necessity if a company wants to
provide good and engaging customer service.
TDWI Research85 examined the importance of accomplishing various objectives
for gaining positive ROI from customer analytics (see Figure 7). “Using customer
analytics to target cross-sell and up-sell opportunities was the objective cited by
the biggest percentage of respondents (54%).” 85 This objective is about gaining
more value from existing customers, understanding their purchasing habits, and
trying to get them to buy more products more often.85
“Some organizations (18%) are implementing an advanced technique called
‘uplift modeling’ (also called incremental or true-lift modeling), which enables
marketers to use data mining to measure the impact and influence of marketing
actions on customers.”85 Insights such as these allow marketers to develop new
kinds of predictive models to determine the best prospects for up-sell and cross-
sell offerings.85 “As firms scale up to execute large numbers of campaigns across
multiple channels, the efficiency gained from predictive modeling can be critical
to marketing spending optimization,” argues Stodder. 85

71
 ANDREW W. PEARSON

In your organization, which of the following marketing objectives are

most important to achieve for customer analytics to deliver a return
on investment? (Please select all that apply.)

54%

49%

47%

42%

39%

37%

35%

32%

26%

23%

18%

14%

Figure 7: Which Are the Most Important Business Objectives When It Comes to
Customer Analytics?
Source: TDWI Research85, Based on 1,625 responses from 432 respondents;
almost four responses per respondent, on average.

Analytics can improve marketing performance by quantifying a customer’s

lifetime value as well as customer worth at the many different stages in the
customer’s life cycle.85 Armed with this kind of information, managers can align
their deployment of resources to achieve the highest value, as well as avoid the
costs and inefficiencies of marketing to the wrong people at the wrong time. 85
Organizations have long used demographics such as gender, household size,
education, occupation, and income to segment customers, but data mining
techniques let organizations segment much larger customer populations and,
perhaps, more importantly, determine whether to apply new characteristics that
refine segmentation to fit the specific attributes of the organization’s products
and services.85 AI can increase these demographic, education, occupation, and
incomes variants exponentially, as well as add behavioral ones as well.

72
 THE A.I. CRUISE LINE

“Customer analytics using data mining tools improves the speed of segmentation
analysis over manual and spreadsheet efforts that are often used in less mature
organizations.”85 Speed is a vital ingredient for marketing initiatives that are time
sensitive, particularly for those companies that need to provide real-time cross-
sell and up-sell offers to customers clicking through web pages.
With social media added to the mix, as well as clickstreams, and other behavioral
data, the volume and variety of data is exploding, and that can be a godsend for
cruise lines that want to increase personalization.85
Social networking sites such as Facebook, Twitter, and LinkedIn “have files
containing petabytes of data, often in vast Hadoop clusters.”85 Weibo and
WeChat add hundreds of millions of users into the mix and, with it, petabytes of
data as well.
Text analytics can be used to increase the speed, depth, and consistency of
unstructured content analysis far greater than what can be done manually. 85
“More advanced analytics can look for correlations between satisfaction ratings,
commented sentiments, and other records, such as first-call-resolution
metrics.”85
“To analyze data generated by social media networking services such as Twitter,
Facebook, Weibo, and LinkedIn, many organizations are implementing Hadoop
and NoSQL technologies, which do not force a schema on the source data prior
to storage, as traditional BI and data warehousing systems do.” 85 Because of this,
the discovery analytics processes can run against the raw data. 85
“Customer analytics tools need to be able to consume data from sources such as
Hadoop clusters and then integrate the insights into overall customer profiles,”
says Stodder.85 The data sources can be varied for these technologies and
methods, says Stodder; “they include transaction data, clickstreams, satisfaction
surveys, loyalty card membership data, credit card purchases, voter registration,
location data, and a host of [other] demographic data types.” 85
In his article Control Group Marketing — With or Without CRM Software
Systems86, Rick Cook states that, “The basic idea of a control group is simple.
Select a random (or nearly random) sample from your campaign's marketing list
and exclude them from the promotion. Then measure the control group's
activity and compare it to the activity of the group targeted via a campaign. The
difference between the control and campaign group gives you a pretty good
notion of how effective — and profitable — the campaign is.
“The theory is that a certain fraction of the customers in the campaign are going
to purchase from you anyway during the campaign period. The control group lets
you filter out that effect, as well as the effects of other channels which may be
influencing behavior, such as display advertising, and shows you how much the
campaign has affected customer behavior,” explains Cook.86

73
 ANDREW W. PEARSON

Although control groups should be used to test out the effects of marketing
campaigns, few companies include them in their marketing processes.86
“Marketing control groups become even more effective when combined with
the customer analytics found in most marketing automation or customer
relationship management systems,” notes Cook.86
Cook argues that, “With a CRM system and a control group you can also detect
the halo effect of your campaign. These are purchases and other actions which
are influenced by the campaign but don't come in through the normal campaign
channels.”86 For example, a customer could be so inspired by one particular
campaign that he or she picks up the phone and orders products directly from
the company instead of going through the call-to-action channel.86 “Another
example is the customer who doesn't use the promotional coupon you included
in your marketing campaign but who purchases the product anyway.”86 Cook
notes that brands “can assume that customers in the test group who respond in
unconventional methods are still influenced by the campaign and so should be
counted as part of the campaign effect.” 86 “Because CRM software lets you track
all points of customer contact, and not just the direct response to the campaign,
it can capture these halo customers,” concludes Cook. 86
The size of the control group is usually 10 percent of the size of the campaign or
test group.86 Ideally you want the control group to be a truly random sample
from the company’s campaign list, but this is difficult to attain in practice as
complete randomness is hard to achieve.86 “Many companies select their control
group by a simpler process, such as selecting every 10th name on the list to make
up the control group,” but there are other more scientific ways to choose the
participants, which could and should be utilized.86

Customer Lifecycle
Today, we can safely say that the mass marketing experience is over. According
to Gartner, there are five stages of customer experience maturity — initial,
developing, defined, managed, and optimizing. The goal here is to improve the
customer experience through a systematic process to improve customer
satisfaction, loyalty, and advocacy.
In its 15 Applications of Artificial Intelligence in Marketing87, Huguesrey maps out
the most effective AI technologies for marketing across the customer lifecycle.
“All the techniques are 'AI' in the sense that they involve computer intelligence,
but we've broken them down into 3 different types of technology — Machine
Learning Techniques, Applied Propensity Models, and AI Applications,” says
Huguesrey.87 The steps are broken down into the customer lifecycle RACE
framework (See Figure 8), which contains four separate groups — Reach, Act,
Convert, and Engage.87

74
 THE A.I. CRUISE LINE

Figure 8: A.I. for marketers across the customer lifecycle

Source: Huguesrey87

“Each different application has major implications for marketers, but the
applications have different roles to play across the customer journey. Some are
better for attracting customers, whilst others are useful for conversion or re-
engaging past customers,” says Huguesrey.87
According to Huguesrey, reach “involves using techniques such as content
marketing, SEO and other 'earned media' to bring visitors to your site and start
them on the buyer's journey.”87 “AI & applied propensity models can be used at
this stage to attract more visitors and provide those that do reach your site with
a more engaging experience.”87
AI-generated content can be a good place to start. “AI can't write a political
opinion column or a blog post on industry-specific best practice advice, but there
are certain areas where AI generated content can be useful and help draw
visitors to your site,” says Huguesrey.87 AI content writing programs like
Wordsmith can pick elements from a dataset and structure a “human sounding”
article from it.87
“AI writers are useful for reporting on regular, data-focused events. Examples
include quarterly earnings reports, sports matches, and market data,” says
Huguesrey.87 If you operate in a niche such as financial services or sports, then

75
 ANDREW W. PEARSON

“AI generated content could form a useful component of your content marketing
strategy.”87 AI-powered content curation allows brands to better engage
visitors and customers on their site by showing them relevant content. 87
Huguesrey sees it as “a great technique for subscription businesses, where the
more someone uses the service, more data the machine learning algorithm has
to use and the better the recommendations of content become.” 87 The systems
becomes somewhat of a self-fulfilling prophecy, like it has become for
companies like Netflix, Pandora, and Amazon.
Cruise lines should have plenty of content to showcase about their upcoming
destinations and many cruise passengers are repeat customers so AI-created
content could prove to be a strong lead generator.
In the coming years, voice search is expected to change the future of SEO and
brands need to keep up with the changing times.87 Huguesrey believes “A brand
that nails voice search can leverage big gains in organic traffic with high purchase
intent thanks to increased voice search traffic due to AI driven virtual personal
assistants.”87
Programmatic media buying — the algorithmic purchase and sale of
advertisements in real time — “can use propensity models generated by
machine learning algorithms to more effectively target ads at the most relevant
customers.”87 AI can ensure programmatic ads don’t appear on questionable
websites and/or remove them from a list of sites that the advertiser doesn’t
want them to appear on.87
In her article Programmatic Advertising 101: How it Works88, Sara Vicioso states
that, “programmatic advertising is the automated process of buying and selling
ad inventory through an exchange, connecting advertisers to publishers.” This
process uses artificial intelligence technologies “and real-time bidding for
inventory across mobile, display, video and social channels — even making its
way into television.”88
Vicioso adds that, “Artificial intelligence technologies have algorithms that
analyze a visitor’s behavior allowing for real time campaign optimizations
towards an audience more likely to convert. Programmatic companies have the
ability to gather this audience data to then target more precisely, whether it’s
from 1st party (their own) or from a 3rd party data provider.” 88
Programmatic media buying includes the use of demand-side platforms (DSPs),
supply-side platforms (SSPs) and data management platforms (DMPs). 88 DSPs
facilitate the process of buying ad inventory on the open market, as well as
provide the ability to reach a brand’s target audience due to the integration of
DMPs.88 “DMPs collect and analyze a substantial amount of cookie data to then
allow the marketer to make more informed decisions of whom their target
audience may be,” says Vicioso.88

76
 THE A.I. CRUISE LINE

“On the publisher side of things, publishers manage their unsold ad inventory
through an SSP,”88 which reports such clickstream activity as how long a visitor
was on a specific site or how many pages were viewed per visit.88 Vicioso explains
that, “SSPs will ultimately be in charge of picking the winning bid and will serve
the winning banner ad on the publisher’s site.”88
As Allie Shaw notes in her article AI could save television advertising with
advanced personalization89, “In short, AI programs draw from data pools to make
decisions about where and when to buy or sell ad space according to
demographic and cost-versus-benefit information.” “Essentially, your TV can
learn about your habits in the way your web browser already does, allowing
advertisers to present you with ads based on that information — so you’ll see
fewer repetitive ads that you don’t care about. This means you and your
neighbors may all be watching the premiere of The Walking Dead but seeing
different ads based on your unique interests,” explains Shaw. 89
“Thanks to programmatic TV advertising, advertisers can know how many people
have viewed their ads, where these viewers are located, and what their viewing
history looks like — with information updating by the minute,” says Shaw.89
“They’re also able to get more accurate data about an ad’s cost per impression
(CPM, or the cost for each 1,000 people who see the ad), allowing for more
relevant and cost-efficient targeting,” she explains.89
The second step of the RACE framework is “Act”. Brands must draw visitors in
and make them aware of the company’s product and/or services. Machine
learning algorithms can build propensity models that can predict the likelihood
of a given customer to convert, the price at which a customer is likely to convert,
and/or what customers are most likely to turn into repeat customers. 87
“Propensity models generated by machine learning can be trained to score leads
based on certain criteria so that your sales team can establish how 'hot' a given
lead is, and if they are worth devoting time to,” explains Huguesrey. 87 “This can
be particularly important in B2B businesses with consultative sales processes,
where each sale takes a considerable amount of time on the part of the sales
team,” says Huguesrey.87
The machine learning algorithms can run through vast amounts of historical data
to establish which ads perform best on which people and at what stage in the
buying process.87 Using this data, ads can be served to them with the most
effective content at the most effective time. 87 By using machine learning to
constantly optimize thousands of variables, businesses can achieve more
effective ad placement and content than traditional methods.87 However,
humans will still be needed for the creative parts.87
The third step of the RACE framework — “Content” — is one of the most
important steps and it includes dynamic pricing, re-targeting, web and app
personalization, and chatbots.87

77
 ANDREW W. PEARSON

All marketers know that discount sales are one of the most effective ways of
moving product, but they can also hurt the financial bottom line.87 Sales are so
effective because they get people to buy a product that they might not have
previously considered because they couldn’t justify the cost of the purchase. But
sales also mean people who would have paid the higher price pay less than they
would have. The trick is to understand the threshold between buying and not
buying and this is where dynamic pricing comes in.
By targeting special offers only at those who are likely to need them in order to
convert, brands can ensure they don’t give offers to people who have the
propensity to pay full price.87 “Machine learning can build a propensity model of
which traits show a customer is likely to need an offer to convert, and which are
likely to convert without the need for an offer,” says Huguesrey. 87 This means
companies can increase sales, while also maximizing their profit margins. 87
By using a propensity model to predict a customer’s stage in the buying cycle,
cruise lines can serve the customer, either through an app or on a web page,
with the most relevant and timely content.87 “If someone is still new to a site,
content that informs them and keeps them interested will be most effective,
whilst if they have visited many times and are clearly interested in the product
then more in-depth content about a product’s benefits will perform better,”
states Huguesrey.87
Another way to convert customers is with chatbots that mimic human
intelligence by interpreting a consumer’s queries and potentially complete an
order for them.87 Chatbots are relatively easy to build and Facebook is
simplifying the process of developing chatbots for brands.87 Facebook “wants to
make its Messenger app the go-to place for people to have conversations with a
brand’s virtual ambassadors.”87 Facebook has created the wit.ai bot
engine, which allows brands to train bots with sample conversations and have
these bots continually learn from customer interactions.87
“Much like with ad targeting, machine learning can be used to establish what
content is most likely to bring customers back to the site based on historical
data,” says Huguesrey.87 By building an accurate prediction model of what
content works best to win back different customer types, machine learning can
help optimize a brand’s retargeting ads to make them as effective as possible.87
The final step of the RACE framework is “Engage”. As previously mentioned, it is
far easier to sell to an existing customer than it is to find and attract new ones,
therefore keeping current customers happy is paramount. 87 “This is particularly
true in subscription-based business, where a high churn rate can be extremely
costly,” contends Huguesrey.87
“Predictive analytics can be used to work out which customers are most likely to
unsubscribe from a service, by assessing what features are most common in
customers who do unsubscribe,” says Huguesrey. 87 “It’s then possible to reach

78
 THE A.I. CRUISE LINE

out to these customers with offers, prompts or assistance to prevent them from
churning,” contends Huguesrey.87
Marketing automation techniques usually involve a series of business rules,
which, once triggered, initiate or continue interactions with a given customer. 87
However, these rules can be quite arbitrary.87 “Machine learning can run through
billions of points of customer data and establish when are the most effective
times to make contact, what words in subject lines are most effective and much
more,” says Huguesrey.87 “These insights can then be applied to boost the
effectiveness of your marketing automation efforts,” he adds.87
“In a similar fashion to marketing automation, applying insights generated from
machine learning can create extremely effective 1:1 dynamic emails,” says
Huguesrey.87 Propensity models can “establish a subscribers propensity to buy
certain categories, sizes and colors through their previous behavior and displays
the most relevant products in newsletters.”87 The product stock, deals, pricing
specifically individualized for each customer would all be correct at the time the
customer opens the offer email.87
For most businesses, customer information housed in an EDW would include
things like transactional data, customer and CRM data, mobile, social, and
location data, as well as information from web logs that track its user’s web
behavior, and online advertising bid management systems. EDWs should also
give a business the ability to do analytics on the fly, which could help the
customer’s experience in a multitude of ways.
Today, most big companies which have large customer databases have loyalty
programs that are part of a CRM and/or an SCRM initiative. These companies
should provide their customers with an intimate experience that will make them
want to return to again and again and again.
Obviously, creating a consolidated customer view is a necessary component of
personalization. Another important step of bringing personalization efforts up to
a user’s expectation level will be using behavioral data in the process. In order
to create these types of customer experiences, businesses need to strategically
collect and utilize customer data, including real-time signals of intent, which
aren’t always captured today. They are not easy to capture, either.
In their article Knowing What to Sell, When, and to Whom90, authors V. Kumar,
R. Venkatesan, and W. Reinartz showed how, by simply understanding and
tweaking behavioral patterns, they could increase the hit rate for offers and
promotions to consumers, which then had an immediate impact on revenue.
By applying statistical models based on the work of Nobel prize-winning
economist Daniel McFadden, researchers accurately predicted not only a specific
person’s purchasing habits, but also the specific time of the purchase to an
accuracy of 80%.91

79
 ANDREW W. PEARSON

Obviously, the potential to market to an individual when he or she is primed to

accept the advertising is advantageous for both parties involved. By utilizing data
from past campaigns and measures generated by a predictive modeling process,
cruise lines can track actual campaign responses versus expected campaign
responses, which can often prove wildly divergent. Additionally, cruise lines can
generate upper and lower “control” limits that can be used to automatically alert
campaign managers when a campaign is over or underperforming, letting them
focus on campaigns that specifically require attention.
One of the benefits of automating campaigns is that offers based on either stated
or inferred preferences of patrons can be developed. Analysis can identify which
customers may or may not be more responsive to an offer centered around a
particular trip. The result: more individualized offers are sent out to the cruise
line and, because these offers tap into a customer’s wants, desires, needs and
expectations, they are more likely to be used; more offers used mean more
successful campaigns, which means more money coming into the cruise line’s
coffers.
With predictive analytics, a cruise line can even predict which low-tier and mid-
tier customers are likely to become the next high rollers. In so doing, the cruise
line can afford to be more generous in its offers as it will know that there is a
high likelihood that these customers will appreciate the personalized attention
and therefore become long term — and, hopefully, highly profitable — patrons.
Once the patron leaves the cruise and returns home, the marketing cycle begins
anew. RFM models can project the time at which a patron is likely to return and
social media should be checked for any comments, likes or uploads left by a
customer, something that should already be occurring.
A campaign management solution can enable the cruise line to develop and
manage personalised customer communications strategies and the delivery of
offers. It will also allow users to rapidly create, modify and manage multi-
channel, multi-wave marketing campaigns that integrate easily with any
fulfilment channel, automatically producing outbound (contact) and inbound
(response) communication history.
Users can define target segments, prioritise selection rules, rank offers across
multiple campaigns and channels, select communication channels, schedule and
execute campaigns, and perform advanced analyses to predict and evaluate the
success of customer communications.
The customer journey starts the moment a potential customer browses to a
cruise line’s webpage or notices an advertisement for a cruise on television, or
on the Internet, or in print. With a few browser click strokes, a cruise line’s
marketing department can create a click path analysis that reveals customer
interactions on the cruise line’s website. Descriptive analytical functionalities can
then provide a deeper understanding of the customer journey. Column

80
 THE A.I. CRUISE LINE

dependencies can visually display the strength of a relationship between

attributes within any dataset. This helps users better understand the
characteristics of their data and is often used to help target further analytics.
A recommendation engine can also help predict a person’s interest based on
historical data from many users. This is useful in increasing client engagement,
recommending more relevant choices and increasing customer satisfaction. For
example, recommendations can predict interest in the cruise line’s games and
services.
Rapid advancements in facial-recognition technology have reached the point
where a single face can be compared against 36 million others in about one
second.92 A system made by Hitachi Kokusai Electric and reported by DigInfo TV
shown at a security trade show a few years ago was able to achieve this blazing
speed by not wasting time on image processing. Using edge analytics, it takes
visual data directly from the camera to compare the face in real time. 92 The
software also groups faces with similar features, so it is able to narrow down the
field of choices very quickly. The usefulness to the cruise line’s security
enforcement is pretty obvious, but it can also be used by multiple departments;
facial recognition technology can be set up to send alerts to hosts, store clerks
or managers, or just about anyone needing it.
Predictive modeling is only useful if it is deployed and it creates an action. Taking
advantage of the more powerful, statistically based segmentation methods,
customers can be segmented not only by dollar values, but also on all known
information, which can include behavioral information gleaned from resort
activities, as well as the patron’s simple demographic information. This more
detailed segmentation allows for more targeted and customer-focused
marketing campaigns.
Models can be evaluated and reports generated on multiple statistical measures,
such as neural networks, decision trees, genetic algorithms, the nearest neighbor
method, rule induction, and lift and gains charts. Once built, scores can be
generated in a variety of ways to facilitate quick and easy implementation. The
projects themselves can be re-used and shared to facilitate faster model
development and knowledge transfer.
In his paper Predictive Analytics93, Wayne Eckerson advises creating predictive
models by using the following six steps:
1. Define the business objectives and desired outcomes for the project
and then translate them into predictive analytic objectives and tasks.
2. Explore and analyze the source data to determine the most appropriate
data and model building approach and then scope the effort.
3. Prepare the data by selecting, extracting, and transforming the data,
which will be the basis for the models.
4. Build the models, as well as test and validate them.

81
 ANDREW W. PEARSON

5. Deploy the models by applying them to the business decisions and

processes.
6. Manage and update the models accordingly.
By utilizing data from past campaigns and measures generated by the predictive
modeling process, a cruise line can track actual campaign responses versus
expected campaign responses, which can often prove wildly divergent.
Additionally, a cruise line can generate upper and lower “control” limits that can
be used to automatically alert campaign managers when a campaign is over or
underperforming, letting them focus on campaigns that specifically require
attention.
By understanding what type of patron is on its website, why they are there, and
what they like to do while they are there, a cruise line can individualize its
marketing campaigns so that they can be more effective, thereby increasing the
cruise line’s ROI.
All of a patron’s captured information can now become part of the master
marketing profile that will be the basis for future marketing efforts. Combining
the daily, weekly, and monthly master marketing profiles will also allow the
cruise line to develop insightful macro views of its data, views that could help
with labor management and vendor needs.

Customer Loyalty
Loyalty is so important to a cruise line company because, as repeated studies
have shown, customers become more profitable over time. In their study Zero
Defections: Quality Comes to Service94, Reichheld and Sasser demonstrated that
a customer’s profitability rises as his or her loyalty increases. In this study, the
authors found that it usually took more than a year to recoup any customer
acquisition costs, but then profits increased as customers remained with the
service or firm.
Here are a few other facts and figures regarding customers and their loyalty:
• On average, loyal customers are worth up to 10 times as much as their
first purchase.95
• It is 6-7 times more expensive to acquire a new customer than it is to
keep a current one.95
• News of bad customer service reaches more than twice as many ears as
praise for a good service experience.95
• For every customer who bothers to complain, 26 other customers
remain silent.95
Loyalty is so important to a cruise line because, as repeated studies have shown,
customers become more profitable over time. Reichheld and Sasser

82
 THE A.I. CRUISE LINE

demonstrated that a customer’s profitability increases as his or her loyalty

increases.94 In this study, the authors found that it usually took more than a year
to recoup any customer acquisition costs, but then profits increased as
customers remained with the service or the firm.
Reichheld and Sasser believe there are four factors for this growth and, in order
of their importance, they are94:
1. Profit derived from increased purchases: as a customer ages, he or she
will probably become more affluent, therefore will have more money
to spend for company products/services.
2. Profit from reduced operating costs: As customers become more
experienced, they should make fewer demands on the business,
perhaps taking advantage of available self-service options.
3. Profit from referrals to other customers.
4. Profit from price premiums: long-term customers are more likely to
pay regular prices for services rather than being tempted into using a
businesses’ lower profit products and/or services.
As previously mentioned, customer satisfaction is the foundation of true
customer loyalty, while customer dissatisfaction is the key factor that drives
customers away. This may sound obvious, but its importance cannot be
overstated. The number one thing that creates loyalty in anybody (that includes
your customers) is the social construct of reciprocity — the social norm that's
been evaluated and debated since the days of Aristotle.94 Many scholars believe
it to be one of the single most defining aspects of social interaction that keeps
society whole.94 Reciprocity doesn’t have to be a bar of gold, like some cruise
lines in Macau like to offer their high rollers, it could simply be an
acknowledgement of poor customer service along with the promise to do better
in the future.94
As previously mentioned, Jones and Sasser warned that, “Extremely dissatisfied
customers can turn into ‘terrorists,’ providing an abundance of negative
feedback about the service provider.”80 Through social media channels, negative
feedback can reverberate around the world within seconds. Today, more than
ever, companies must spot dissatisfied customers and approach them before
they do irreparable harm to a company’s image and reputation.
In the zone of indifference, customers willingly switch if they can find a better
alternative, while in the zone of affection, satisfaction levels are high and
“customers may have such high attitudinal loyalty that they don’t look for
alternative services.”80 It is within this group that “Apostles” — members who
praise the firm in public — reside and this is the group that is responsible for
improved future business performance.80 In the social media world, these people
are more likely to be known as “influencers” and I will go into much more detail
about this type of marketing in chapter five.

83
 ANDREW W. PEARSON

A consumer’s engagement with a brand can be measured along a continuum

from no awareness, through early engagement, and, hopefully, if everything
goes right, into advocacy.45 As for the customer-company relationship, “the
strength of feeling will develop and vary over time and, as in any healthy
relationship, both parties should be aware of feelings so they can react
accordingly,” advises Woodcock.45
As was shown in Nielsen’s 2012 Global Trust in Advertising96 survey, consumers
trust their friends and colleagues much more than they trust TV advertising or
corporate communications. Today, consumers communicate with each other
like never before, through a multitude of social and mobile media channels and
these channels should be exploited as much as possible.
“SCRM is the connection of social data (wherever it is) with existing customer
records (customer database) that enable companies to provide new forms of
customer insight and relevant context.”45 With SCRM, marketers can
“understand the mood, find new sales leads, respond faster to customer needs
and maybe even anticipate needs by listening into their conversations and taking
action.”45
SCRM doesn’t replace CRM systems; it adds value by augmenting traditional
systems.45 As Woodcock notes, “SCRM is a great hunting ground for businesses
to find and acquire consumers to full ‘traditional’ CRM programs as well as
identify key influencers who can be considered as high value customers. It offers
companies an organized approach, using enterprise software that connects
business units to the social web giving them the opportunity to respond in near
real time, and in a coordinated fashion.”45
Social media can help amplify the “relationship” in “Customer Relationship
Management”, thereby enabling organizations to connect and engage
consumers in a unique way, as well as personalize and monetize customer
relationships on a sustained basis, which should increase profitability.110 “Social
media also provides a path to richer customer analysis, using technologies
capable of funneling and consolidating customer insights.”110 Insights derived
from this analysis can help companies to “dynamically calibrate, anticipate, and
offer products and services that meet perpetually shifting consumer demands in
a hyper-competitive marketplace.”110
Marketers can also “listen into what customers are saying, to better understand
their needs, their voices and tie it back to actual customer profiles”45, which
could contain their Facebook or WeChat pages or Twitter handles. “In addition,
marketers will be able to catch leads in ‘mid-air’ by listening for keywords that
suggest a customer is getting ready to buy, then sending real-time alerts to sales
teams to respond.”45
For a cruise line, it would be advantageous to link a patron’s account with his or
her social media accounts so that the cruise line could get a heads-up on what a

84
 THE A.I. CRUISE LINE

patron might be saying about them on social media. A tip off about an upcoming
and/or a last-minute trip to Vegas, Macau, KL, Manila, or Singapore could be
captured from social media, then acted upon accordingly.
In a lot of cases, ROI is an enormously tricky thing to measure, but social media
is providing unique ways for businesses to quantify their social media spend;
gone are the days of companies wasting endless amounts of time and money
building up what amounts to a useless group of Facebook followers, or, at least,
they should be.
“Short-term campaign ROI as the main measure for individual campaigns will
evolve into correlation analysis between activities, engagement and sales. This
will be unsettling for many traditional marketers.” 45 “The explicit use of active
and control groups, and experimentation of using different treatments will help
marketers understand the impact of specific SM activities.”45 More direct
marketing type disciplines will be required, in a world where there is real-time
feedback on attitude and behavior and a plethora of data.”45 This has become a
much more demanding world in terms of capturing and utilizing all of this data,
but making the effort to turn this data into actionable intelligence will be noticed
by fickle consumers, I have no doubt.
In its Retail Analytics: Game Changer for Customer Loyalty 97, Cognizant argues
that in the retail industry, “predictive models can be used to analyze past
performance to assess the likelihood that a customer will exhibit a specific
behavior in order to improve marketing effectiveness.” This can help with
“predicting customer reactions to a given product and can be leveraged to
improve basket size, increase the value of the basket and switch the customer
to a better and more profitable offering.”97 Predictive models can also help tailor
pricing strategies that take into account both the need for competitive pricing
and the company’s financial bottom line.97 Both of these processes can be done
in the cruise industry as well, obviously.
Predictive analytics and data mining are used to discover which variables out of
possibly hundreds are most influential in determining customer loyalty within
certain segments.85 “Advanced analytics generally involves statistical,
quantitative, or mathematical analysis and centers on developing, testing,
training, scoring, and monitoring predictive models.”85
Models can be created that will uncover patterns, affinities, anomalies, and
other useful insights for marketing campaigns and for determining cross-sell and
up-sell opportunities.85 “The tools and techniques are also used for developing
and deploying behavioral scoring models for marketing, deciding whether to
adjust customers’ credit limits for purchases, and a variety of highly time-
sensitive analytic processes,” notes Stodder.85
“As more online customer behavior is recorded in Web logs and tracked through
cookies and other observation devices, sizeable amounts of information are

85
 ANDREW W. PEARSON

becoming available to organizations that seek a more accurate view of a

customer’s path to purchase,” says Stodder. 85 Attribution analysis is, first and
foremost, a big-data problem, given the quantity and variety of data available
from today’s multiple platforms.85
Businesses that are performing attribution analysis will frequently employ
Hadoop and/or MapReduce, with analytic software solutions such as R, SAS’s
eMiner, SAP’s InfiniteInsights, Python, and IBM’s SPSS, amongst others.85 This
allows a business to run sophisticated algorithms against detailed data to find
the correct path to purchase. This analysis can then be integrated with analysis
from other data types and sources, including those which might have been
generated by some offline customer activity.85
Attribution analysis can reveal such things as what kinds of campaigns most
influence customer behavior. 85 “The analysis can help organizations determine
where to allocate marketing resources to gain the highest level of success, as
well as how to more accurately assign the percentage of credit due to specific
marketing and advertising processes,” concludes Stodder. 85
On August 13, 2014, Facebook announced a major step forward in the area of
attribution analysis.98 In his article Facebook Now Tells Whether Mobile Ads Lead
to Desktop Purchases98, T. Peterson says that Facebook “would start telling
advertisers on what device people saw an ad and on what device they took an
action, such as buying a product or signing up for a test drive, as a result of seeing
that ad. That means Facebook will be able to credit mobile ads that lead to
desktop sales and desktop ads that result in mobile purchases.”
Peterson notes that, “Advertisers can already track conversions through
Facebook on desktop and on mobile, but to date Facebook hasn't broken out
conversions by device type for advertisers to see. For example, advertisers have
been able to see if their desktop and mobile ads lead to conversions, but they
didn't know on which device type those conversions were taking place.”98
However, Facebook’s new cross-device conversion measurement only works for
advertisers who place specific Facebook trackers on their websites and mobile
apps.98 “Without sharing users' personal information with the advertiser, those
trackers can see that a Facebook user is checking out the advertisers' site or app
and whether they've converted in the advertiser-specified fashion.”98 If the
person does convert, “Facebook's trackers can trace back to see if that person
has seen an ad from that advertiser on Facebook, which may have directly or
indirectly led to the conversion.”98 Of course, nothing is 100% certain when it
comes to attribution analysis, but this is a big step in the right direction.

Cruise Line Engagement and Loyalty Platform

The Cruise Line Engagement and Loyalty Platform (see Figure 9) shows how a
cruise line would engage its customers in a loyalty platform that utilizes social

86
 THE A.I. CRUISE LINE

media as an important part of the process. The Cruise line Engagement and
Loyalty Platform can be implemented in a multitude of ways. In the Listening part,
cruise lines should define and look out for triggers such as photos, hashtags,
keywords, likes, video views, etc., etc. This runs the gamut, from staying on top of
keywords and hashtags on Twitter, Facebook, Instagram, and a whole host of other
potential image-related social sites.
Check-ins and geo-posts from sites like Foursquare, WeChat, Instagram, Facebook,
WhatsApp, YouTube, as well as a whole host of other social networks can help
cruise lines connect with a nearby audience. Sports betting operators should also
be listening to comment boards or short-term blogging sites like Tumblr or social
news aggregation sites like Reddit for comments about their products and services.

Figure 9: Cruise line Engagement and Loyalty Platform

Source: chirpify.com

The Rules Engine step is pretty straightforward; cruise lines are already creating
considerable business rules for their establishments and these should be extended
to the company’s defined rewards program, their reward’s economy, and the
marketing of the program.
Rewards programs are difficult to implement and costly to maintain because there
are so many moving parts; each reward point and free offer has to be correlated
against the department that offered it, the right budget it should be assigned to and
every reward point has a monetary value that has to be enumerated properly.
Building a rules engine can simplify the marketing process by defining who gets
what, when he or she gets it, and through which channel it gets delivered on. With
mobile and social being added to the customer channel mix, things are going to get

87
 ANDREW W. PEARSON

exponentially more complex very quickly, so building a rules engine that lays things
out in a highly definable way is imperative.
Once the rules engine is in place, automation must kick in. With cruise lines now
handling databases filled with millions of customers, it would be impossible to
market to customers without considerable automation going on behind the scenes.
Segmenting customers and building campaigns that market to thousands of
individuals would be impossible as well.
Understanding the ROI of each marketing campaign is imperative, and, with today’s
real-time personalization capabilities, cruise lines can quickly understand who is
accepting their marketing and how much revenue it is driving. Adding a real-time
element to the process would be impossible without strict rules set in place and
powerful marketing automation tools that not only send out marketing offers but
also quantify them once they are utilized.
In terms of marketing and customer service, Facebook bots could be created and
automated to answer standard customer service questions and this should lighten
the load on a cruise line’s customer service department.
Moderating boards and UGC posts are a great channel to connect with customers
and/or potential customers. They are also good places to pick up both customer
service issues and competitor information.
As Chirpify sees it: “Moderation allows brands to increase social efficiency and
effectiveness by uniting automated listening triggers while giving moderators the
ability to manually review posts and user content for fit before determining their
qualification for a reward. This helps brands better personalize the reward based
on the user while making sure that the reward is one that the customer appreciates
and/or makes them feel special.”99
Rewards and marketing content can deliver points, discounts, reminders, as well as
contest entries in real-time, but reaching today’s audiences can be tricky.
Marriott won the Chief Marketer’s 2017 Gold for Best Loyalty Marketing for the
development of a loyalty engagement platform that uses rewards points as social
currency to incentivize engagement on Twitter, Facebook and Instagram.100
In the Chief Marketer write up about the award, the magazine noted that, “The
platform was designed as a reciprocal ecosystem where members are
empowered to engage via social media and advocate on behalf of the program.
The platform engages guests one-to-one at scale with personalized content and
instant rewards, enabling members to engage with the loyalty program even
when they aren’t on property.”99
“By connecting their social media accounts with the platform, members can earn
points by engaging with a variety of triggers throughout the year. The platform
is integrated directly with Marriott Rewards’ member database, allowing the

88
 THE A.I. CRUISE LINE

program to recognize and reward members the moment they post,” Chief
Marketer notes.99
According to Chief Marketer, engagement was huge. “When it came time to
amplify the launch of Marriott Rewards’ Reward-a-Friend enrollment program,
connected members helped spread the word with a simple retweet. The
campaign generated 7.4M earned media impressions in four days.” 99
“Connected members generated more than 65 million positive earned media
impressions in 2016 on behalf of Marriott Rewards. More than 84 million
Rewards points were awarded in response to 326,000 social media
engagements, equivalent to earning roughly 11,000 free nights.” 99 This kind of
engagement comes at a cost, obviously (the free rooms), but the enormous
amount of engagement is worth the cost because it is word-of-mouth marketing
and it allows verifiable visibility on the engagement. The ROI shouldn’t be hard
to quantify either.
Most major cruise line companies today have customer loyalty programs that
are a part of a CRM and/or a SCRM initiative to provide their customers with an
intimate experience that will make them want to return to the cruise line again
and again and again.
Obviously, creating a consolidated customer view is a necessary component of
personalization. Another important step of bringing personalization efforts up to
a user’s expectation level will be using behavioral data. In order to create these
types of customer experiences, cruise line companies will strategically collect
and utilize customer data, including real-time signals of intent.
A campaign management solution can enable the cruise line to develop and
manage personalised customer communications strategies and the delivery of
offers. It will also allow users to rapidly create, modify and manage multi-
channel, multi-wave marketing campaigns that integrate easily with any
fulfilment channel, automatically producing outbound (contact) and inbound
(response) communication history. Users can define target segments, prioritise
selection rules, prioritise offers across multiple campaigns and channels, select
communication channels, schedule and execute campaigns, and perform
advanced analyses to predict and evaluate the success of customer
communications.
The customer journey starts a long time before the customer even enters the
cruise ship. It begins the moment a potential customer browses to a cruise
operator’s webpage or notices an advertisement for a cruise line on television,
or on the Internet, or in print, or on a billboard. It can even be while connecting
with a cruise line’s social media accounts, or even the moment the customer
actually enters the ship.
With a few browser click strokes, a cruise operator’s ecommerce department

89
 ANDREW W. PEARSON

can create a click path analysis that reveals customer interactions on the cruise
line’s websites. Descriptive analytical functionalities can then provide a deeper
understanding of the customer journey.
A recommendation engine can help predict a person’s interest based on
historical data from many users. This is useful in increasing client engagement,
recommending more relevant choices and increasing customer satisfaction. For
example, recommendations can predict interest in cruise line products or
services, room types, etc., etc.
Like the proverbial butterfly that flaps its tiny wings in Brazil, and sets off a
typhoon in Manila, any customer who shows any inkling towards visiting the
property can be quantified and analyzed so that not only is their trip a rewarding
one, but also that their customer touch points are reduced as much as possible
so the labor needs of the cruise line are kept to a minimum.
By understanding what type of patron is on its ships, why they are there, and
what they like to do while they are there, a cruise line operator can individualize
its marketing campaigns so that they can be more effective, thereby increasing
the cruise line operator’s ROI.
One thing to keep in mind when it comes to rewards is uniqueness and what has
become known as “social rewards”. In his article Getting it right: mixing social and
economic rewards in hotel loyalty programmes 101, Lee Jin-soo argues that:
“The concept of relationship marketing is prevalent in the local
hotel industry, giving rise to numerous loyalty or reward
programmes that offer preferential rewards for members in
proportion to how often they patronise an establishment. The
rewards offered — usually in the form of economic or social
benefits — make members feel special, important, and
appreciated. Common examples of economic benefits include
free room, room upgrade and discounts. Social benefits, on the
other hand, are more diverse and include any preferential
treatment or personalised recognition and attention given to
individual customers.”
Lee believes that social rewards are important in today’s world that is filled with
customer loyalty programs. In comparison to typical rewards, “social rewards
usually work better in building stronger relationships since social benefits
enhance a customer’s intrinsic reasons for sustaining and reinforcing emotional
commitment, and thus his attachment to a specific hotel.” 101
The author of the article, Lee Jin-soo is also an associate professor of the School
of Hotel and Tourism Management at the Hong Kong Polytechnic University and
the article discusses a study the university did exploring the impact of economic
and social rewards on the relational behavior of loyalty program members. 101

90
 THE A.I. CRUISE LINE

The university “conducted an online survey of 334 participants privy to various

hotel reward programmes in the United States to study their responses towards
economic reward based initiatives versus those based on social benefits. ”101
The findings revealed that a social reward dominant programme was “more
effective than an economic reward oriented programme in encouraging
favourable relational behaviours toward programme providers,” states Jin-
soo.101
The relational effects were especially apparent in the following areas: “openness
(e.g. “I would feel comfortable telling the hotel when I think something needs
improvement.”), advocacy (e.g. “I would defend the hotel to others if I hear
someone speaking poorly about it.”), and immunity (e.g. “Even if I hear some
negative information about the hotel, I would not switch to a competing
hotel”).”101
“When program members were exposed to customised, intrinsic offerings from
a loyalty programme, they tend to be intrinsically motivated to stay with a hotel
by developing emotional attachment, internal enjoyment, and affective
commitment,” Lee noted.
In short, the study concluded that financial incentives alone wouldn’t prevent
customers from switching over to a competing program.101 To foster relational
behaviors and gain long-term benefits from a loyalty program, hotels needed to
consider offering both social and economic rewards.101
In the article, Lee recommends the following:
“Loyalty programme managers should consider economic
benefits as a defensive relational strategy and social rewards
as an offensive relational strategy in stimulating the relational
behaviours of customers. Given the limitations of financial
benefits in maintaining long-lasting customer relationships,
and the practical need for striving cost-effectiveness in
business operation, loyalty programme operators should adopt
the strategy of providing economic rewards just to the extent
of merely meeting, rather than greatly exceeding, members’
expectations. At the same time, they should develop and offer
more customised offers, which can intrinsically motivate
customers to foster a sense of belonging and emotional
attachment to their hotel.”101
Lee uses Thailand’s Phuket Marine Biology Centre and the Royal Thai Navy’s
annual release of “baby turtles at a leading hotel in cooperation with the
governor of Phuket and that hotel’s guests”101 as an example of a customized
social reward.
At the event, “only members of the hotel’s superclass loyalty programme are

91
 ANDREW W. PEARSON

allowed to release baby turtles into the sea, which is a rewarding and memorable
experience, especially for children.”101 “Such a preferential treatment is so
meaningful to customers that they become reluctant to switch to any competing
hotel that offers equal or even better financial incentives,” Lee concludes. 101
To make this entire system work, we are obviously talking about a huge amount
of data flowing through these systems, utilizing everything from a typical EDW,
to Hadoop clusters, as well as stream processing applications like Spark, Flink,
Storm, IBM’s Infosphere, Hitachi’s HDS, TIBCO’s StreamBase, etc., etc.
In many cases, acquiring a social identity that is tied to a customer record is as
simple as asking for it; “Like” my page, “Follow” us, “heart” this, “Pin” that board,
“Snap” to our story, “Tweet or retweet” one of our offers and/or a story about your
trip to our resorts. These are all wonderful ways to engage an audience, either
connecting with a new one or continuing to build a relationship with a current one.

Natural Language Processing

According to skymind.ai9, “Natural language refers to language that is spoken
and written by people, and natural language processing (NLP) attempts to
extract information from the spoken and written word using algorithms.”
In their article How Artificial Intelligence and Machine Learning Can Impact
Market Design19, Paul R. Milgrom and Steve Tadelis give some interesting use
cases for NLP. Online marketplaces like eBay, Taobao, Airbnb, along with many
others have seen exponential growth since their inception because they provide
“businesses and individuals with previously unavailable opportunities to
purchase or profit from online trading.”19 Besides the new marketplaces created
for these wholesalers and retailers, “the so called ‘gig economy’ is comprised of
marketplaces that allow individuals to share their time or assets across different
productive activities and earn extra income.” 19
“The amazing success of online marketplaces was not fully anticipated,” Milgrom
and Tadelis infer, “primarily because of the hazards of anonymous trade and
asymmetric information. Namely, how can strangers who have never transacted
with one another, and who may be thousands of miles apart, be willing to trust
each other?”19 “Trust on both sides of the market is essential for parties to be
willing to transact and for a marketplace to succeed,” claim Milgrom and
Tadelis.19
eBay’s early success is often attributed to its innovative feedback and reputation
mechanism, which has been replicated by practically every other marketplace
that came after eBay.19 Milgrom et al. believe that these online feedback and
reputation mechanisms provide a modern-day version of more ancient
reputation mechanisms used in the physical marketplaces that were the
medieval trade fairs of Europe.102

92
 THE A.I. CRUISE LINE

The problem for Milgrom and Tadelis is that “recent studies have shown that
online reputation measures of marketplace sellers, which are based on buyer-
generated feedback, don’t accurately reflect their actual performance. 19 A
growing body of research reveals that “user-generated feedback mechanisms
are often biased, suffer from ‘grade inflation,’ and can be prone to manipulation
by sellers.”19 “For example, the average percent positive for sellers on eBay is
about 99.4%, with a median of 100%. This causes a challenge to interpret the
true levels of satisfaction on online marketplaces,” state Milgrom and Tadelis.19
For Milgrom and Tadelis, a natural question emerges: “can online marketplaces
use the treasure trove of data it collects to measure the quality of a transaction
and predict which sellers will provide a better service to their buyers?” 19 After
all, these online marketplaces and gig-economy sites collect vast amounts of
data as part of the process of trade.19 The millions of transactions, searches and
browsing that occur in these marketplaces every day could be leveraged to
create an environment that promotes trust, similar to the way institutions
emerged in the medieval trade fairs of Europe that helped foster trust. 19 Milgrom
and Tadelis believe that AI can be applied to these marketplaces to help create
a more trustworthy and better buying experience to consumers.19
“One of the ways that online marketplaces help participants build trust is by
letting them communicate through online messaging platforms,” explain
Milgrom and Tadelis.19 On eBay, buyers question sellers about their products,
“which may be particularly useful for used or unique products for which buyers
may want to get more refined information than is listed.” 19 Airbnb also “allows
potential renters to send messages to hosts and ask questions about the
property that may not be answered in the original listing.” 19
Using NLP, “marketplaces can mine the data generated by these messages in
order to better predict the kind of features that customers value.” 19 However,
Milgrom and Tadelis claim, “there may also be subtler ways to apply AI to
manage the quality of marketplaces.”19 The messaging platforms are not only
restricted to pre-transaction inquiries, they also provide both parties the ability
to send messages to each other post-transaction.19 The obvious question that
emerges for Milgrom and Tadelis is, “how could a marketplace analyze the
messages sent between buyers and sellers post the transaction to infer
something about the quality of the transaction that feedback doesn't seem to
capture?”19
This question was posed and answered in the paper Canary in the e-commerce
coal mine: Detecting and predicting poor experiences using buyer-to-seller
messages103 by Masterov et al. Milgrom and Tadelis explain19:
“By using internal data from eBay’s marketplace. The analysis
they performed was divided into two stages. In the first stage,
the goal was to see if NLP can identify transactions that went

93
 ANDREW W. PEARSON

bad when there was an independent indication that the buyer

was unhappy. To do this, they collected internal data from
transactions in which messages were sent from the buyer to
the seller after the transaction was completed and matched it
with another internal data source that recorded actions by
buyers indicating that the buyer had a poor experience with the
transactions. Actions that indicate an unhappy buyer include a
buyer claiming that the item was not received, or that the item
was significantly not as described, or leaves negative or neutral
feedback, to name a few.”
The simple NLP approach Milgrom and Tadelis use “creates a ‘poor-experience’
indicator as the target (dependent variable) that the machine learning model will
try to predict, and uses the messages’ content as the independent variables.” 19
“In its simplest form and as a proof of concept, a regular expression search was
used that included a standard list of negative words such as ‘annoyed,’
‘dissatisfied,’ ‘damaged,’ or ‘negative feedback’ to identify a message as
negative,” explain Milgrom and Tadelis.19 Messages void of these designated
terms were considered neutral.19 Using this classification, the researchers
grouped transactions into three distinct types: “(1) No post-transaction
messages from buyer to seller; (2) One or more negative messages; or (3) One
or more neutral messages with no negative messages.” 19
In the second stage of the analysis, using the fact that negative messages are
associated with poor experiences, Masterov et al. constructed a novel measure
of seller quality based on the idea that sellers who receive a higher frequency of
negative messages are bad sellers.103 According to Masterov et al., the measure,
which is “calculated for every seller at any point in time using aggregated
negative messages from past sales, and the likelihood that a current transaction
will result in a poor experience,”103 is a monotonically increasing relationship.103
This simple exercise shows that, using a marketplace’s message data and a
simple NLP procedure, businesses can predict which sellers will create poor
experiences better than one inferred from highly inaccurate and wildly inflated
feedback data.19
Of course, eBay is not unique in allowing “parties to exchange messages and the
lessons from this research are easily generalizable to other marketplaces.” 19
“The key is that there is information in communication between market
participants, and past communication can help identify and predict the sellers or
products that will cause buyers poor experiences and negatively impact the
overall trust in the marketplace,” conclude Milgrom and Tadelis. 19

Creating a market for feedback

Besides the over-inflation of customer feedback as described above, another

94
 THE A.I. CRUISE LINE

problem with customer feedback forums is the fact that few buyers even bother
leaving feedback.19 In fact, Milgrom and Tadelis argue, “through the lens of
mainstream economic theory, it is surprising that a significant fraction of online
consumers leave feedback. After all, it is a selfless act that requires time, and it
creates a classic free-rider problem.”19 Additionally, “because potential buyers
are attracted to buy from sellers, or products, that already have an established
good track record, this creates a ‘cold start’ problem,” 19 i.e., new sellers with no
feedback face a high barrier-to-entry because buyers are hesitant to try them
out.19
Li et al. address this problem in their paper Buying Reputation as a Signal of
Quality: Evidence from an Online Marketplace 104 by “Using a unique and novel
implementation of a market for feedback on the huge Chinese marketplace
Taobao where they let sellers pay buyers to leave them feedback.” 19 Of course,
it might be concerning to allow “sellers to pay for feedback as it seems like a
practice in which they will only pay for good feedback and suppress any bad
feedback, which would not add any value in promoting trust.” 19 However,
Milgrom and Tadelis explain that “Taobao implemented a clever use of NLP to
solve this problem: it is the platform, using an NLP AI model, that decides
whether feedback is relevant and not the seller who pays for the feedback.”19
“Hence, the reward to the buyer for leaving feedback was actually managed by
the marketplace, and was handed out for informative feedback rather than for
positive feedback,” note Milgrom and Tadelis.19
“Specifically, in March 2012, Taobao launched a ‘Rebate-for-Feedback’ (RFF)
feature through which sellers can set a rebate value for any item they sell (cash-
back or store coupon) as a reward for a buyer's feedback,” says Milgrom and
Tadelis.19 Sellers who choose this option guarantee that the rebate will be
transferred from the seller's account to a buyer who leaves high-quality feedback
that is, most importantly, informative about the purchased product, rather than
whether the feedback is positive or negative.19 “Taobao measures the quality of
feedback with an NLP algorithm that examines the comment's content and
length and finds out whether key features of the item are mentioned,” explains
Milgrom and Tadelis.19 The marketplace actually manages “the market for
feedback by forcing the seller to deposit at Taobao a certain amount for a chosen
period, so that funds are guaranteed for buyers who meet the rebate criterion,
which itself is determined by Taobao.”19
Taobao wanted to promote more informative feedback, but as Li et al. note,
“economic theory offers some insights into how the RFF feature can act as a
potent signaling mechanism that will further separate higher from lower quality
sellers and products.”104
Building upon the work of Philip Nelson in his influential article Information and
Consumer Behavior105 that suggested advertising acts as a signal of quality, say
Milgrom and Tadelis suggest that, “According to the theory, advertising — which

95
 ANDREW W. PEARSON

is a form of burning money — acts as a signal that attracts buyers who correctly
believe that only high-quality sellers will choose to advertise.”19 “Incentive
compatibility is achieved through repeat purchases: buyers who purchase and
experience the products of advertisers will return in the future only if the goods
sold are of high enough quality,” argue Milgrom and Tadelis. 19 “The cost of
advertising can be high enough to deter low quality sellers from being willing to
spend the money and sell only once, because those sellers will not attract repeat
customers, and still low enough to leave profits for higher quality sellers. Hence,
ads act as signals that separate high quality sellers, and in turn attract buyers to
their products,” state Milgrom and Tadelis.19
Li et al. argue that Taobao’s “RFF mechanism plays a similar signaling role as ads
do, which can be seen as signals that separate high quality sellers, and in turn
attract buyers to their products.”104 Assuming “consumers express their
experiences truthfully in written feedback, any consumer who buys a product
and is given incentives to leave feedback, will leave positive feedback only if the
buying experience was satisfactory.”19
Li et al. believe that a seller will offer RFF incentives to buyers if he or she expects
positive feedback, which usually only happens if the seller provides a high quality
item and/or service.19 “If a seller knows that their goods and services are
unsatisfactory, then paying for feedback will generate negative feedback that
will harm the low-quality seller,” contend Milgrom and Taledis.19 “Equilibrium
behavior,” Milgrom and Tadelis argue, “implies that RFF, as a signal of high
quality, will attract more buyers and result in more sales.” 19 “The role of AI was
precisely to reward buyers for information, not for positive feedback,” state
Milgrom and Tadelis19, and that is as it should be.
Li et al. analyzed data “from the period where the RFF mechanism was featured,
and confirmed that first, as expected, more feedback was left in response to the
incentives provided by the RFF feature.” 104 Li et al. also discovered that “the
additional feedback did not exhibit any biases, suggesting that the NLP
algorithms used were able to create the kind of screening needed to select
informative feedback.”104 Li et al. conclude that, “the predictions of the simple
signaling story were borne out in the data, suggesting that using NLP to support
a novel market for feedback did indeed solve both the free-rider problem and
the cold-start problem that can hamper the growth of online marketplaces.” 104

Reducing Search Friction with A.I.

“An important application of AI and machine learning in online marketplaces is
the way in which potential buyers engage with the site and proceed to search
for products or services,” Milgrom and Tadelis note.19 At Google, Facebook, and
Amazon AI-powered search engines are trained to maximize what the provider
believes to be the right objective.19 “Often this boils down to conversion, under
the belief that the sooner a consumer converts a search to a purchase, the

96
 THE A.I. CRUISE LINE

happier the consumer is both in the short and the long run,” say Milgrom and
Tadelis.19 The rationale: “search itself is a friction, and hence, maximizing the
successful conversion of search activity to a purchase reduces this friction.” 19
Although this is consistent with economic theory, which posits “search as an
inevitable costly process that separates consumers from the products they
want”19 this isn’t really the case. “Unlike the simplistic models of search
employed in economic theory, where consumers know what they are looking for
and the activity of search is just a costly friction, in reality, people’s search
behavior is rich and varied,” claim Milgrom and Tadelis.19
In their paper Returns to Consumer Search: Evidence from eBay106, Blake, Nosko,
and Tadelis use “comprehensive data from eBay to shed light on the search
process with minimal modeling assumptions.” Blake et al.’s data showed that
consumers search significantly more than in previous studies, which were
conducted with limited access to search behavior over time.106
“Furthermore, search often proceeds from the vague to the specific. For
example, early in a search a user may use the query ‘watch’, then refine it to
‘men’s watch’ and later add further qualifying words such as color, shape, strap
type, and more,” explain Blake et al.106 This behavior suggests that consumers
aren’t looking specifically at first and are exploring their own tastes, and what
product characteristics might exist, as part of their search process.106 Blake et al.
showed that the average number of terms in a user’s query “rises over time, and
the propensity to use the default ranking algorithm declines over time as users
move to more focused searches like price sorting.”106
“These observations suggest that marketplaces and retailers alike could design
their online search algorithms to understand search intent so as to better serve
their consumers,” recommend Milgrom and Tadelis.19 Consumers in the
exploratory phases of the search process, should be provided some general
offerings to better learn their tastes as well as all available options in the
market.19
Once the consumer shows the desire to purchase something in particular, the
offering should be narrowed to a set of products that match the consumer’s
preferences.19 “Hence, machine learning and AI can play an instrumental role in
recognizing customer intent,” contend Milgrom and Tadelis.19 This is good news
for any company looking to utilize social media in its marketing plans.
AI and machine learning not only helps “predict a customer’s intent, but given
the large heterogeneity on consumer tastes, AI can help a marketplace or cruise
line better segment the many customers into groups that can be better served
with tailored information.”19
Using AI for more refined customer segmentation, or even personalized
experiences, does raise price discrimination concerns.19 “For example, in 2012

97
 ANDREW W. PEARSON

the Wall Street Journal reported107 that ‘Orbitz Worldwide Inc. has found that
people who use… Mac computers spend as much as 30% more a night on hotels,
so the online travel agency is starting to show them different, and sometimes
costlier, travel options than Windows visitors see.”19 Whether these practices of
utilizing consumer data and AI to adjust pricing helps or harms consumers is up
for discussion, but economic theory states that price discrimination can either
increase or reduce consumer welfare. 19 “If on average Mac users prefer staying
at fancier and more expensive hotels because owning a Mac is correlated with
higher income and tastes for luxury, then Orbitz practice is beneficial because it
shows people what they want to see and reduces search frictions. However, if
this is just a way to extract more surplus from consumers who are less price
sensitive, but do not necessarily care for the snazzier hotel rooms, then it harms
these consumers,” contend Milgrom and Tadelis. 19 Either way, price elasticity
systems can be set up if brands choose to set them up.

Conclusion
We live in a real-time, 24-7 world, a world where 280-character Twitter
messages foment political revolutions; a world where marketers should fear not
the power of the pen, but the destructive force of the critical tweet or the far-
reaching viral impact of an inflammatory social media diatribe that can encircle
the digital world in seconds, laying waste to a reputation that might have taken
decades to develop.
Conversely, it is also a world where an advertiser’s message can go viral and
reach more eyeballs in less than an hour than a multi-million-dollar television
commercial campaign can in a year. Customer intelligence is imperative in a
world that moves so quickly and, thankfully for the cruise line industry, great
strides have been made in developing tools and setting up architectures that
simplify the customer intelligence and customer experience (CX) process.
The cruise line loyalty programs that started in the 80s and 90s should be
morphing into complex data collecting and data crunching machines that
capture not just every dollar (or pound or euro or won or even every mop) that
a customer spends while on a cruise line’s ship, but also every post a customer
submits to his or her social channels so that an online psychometric profile can
be built for each and every customer almost in real-time.
This social profile can be used to both keep track of each customer’s behavior as
well as create competitive intel, i.e., is this particular customer frequenting a
competitor’s cruise line? These psychological profiles can also be the basis for
micro and macro psychometric profiles that a cruise line can utilize to
understand the type of customer it is attracting. This could be highly valuable
information that can be utilized by the cruise line’s marketing department in its
future advertising campaigns. Better client understanding should also go a long

98
 THE A.I. CRUISE LINE

way in closing the gap between customer service expectations and a company’s
delivery of service quality, which is currently quite divergent and trending
possibly trending away.
Cruise lines should recognize that the customer journey is an important concept
to understand and implement. Each customer should be viewed through his or
her unique customer experience maturity stage, which includes “initial”,
“developing”, “defined”, “managed”, and “optimized”. Questions of loyalty,
customer satisfaction, and customer dissatisfaction should be viewed according
to which maturity stage the customer is in. A problem that arises for a new
customer should be handled in a very different way as compared to an issue
arising with a returning patron.

61 Jones, Andrew (2015, December 15). Study finds marketers are prioritizing
personalization…but are further behind than they realize,
http://venturebeat.com/2015/12/14/study-finds-marketers-are-prioritizing-
personalization-but-are-further-behind-than-they-realize/ (accessed 26 November
2016).
62 Vatash, Prateek. (2018). 2018 Digital Trends I Retail. Adobe.
https://wwwimages2.adobe.com/content/dam/acom/uk/modal-
offers/pdfs/Econsultancy-2018-Digital-Trends-Retail_EMEA.pdf (Accessed 6 August
2018).
63 Klein, Michael. Adobe. Machine Learning and AI: If Only My Computer Had a Brain
Wired for Business. https://www.adobe.com/insights/personalization-with-machine-
learning-and-ai.html (Accessed 7 April 2019).
64 Tractica. Artificial Intelligence Revenue to reach 36.8 Billion Worldwide by 2025.
August 25, 2016. https://www.tractica.com/newsroom/press-releases/artificial-
intelligence-revenue-to-reach-36-8-billion-worldwide-by-2025/ (Accessed 7 April 2019).
65 Goode, Lauren. (2019). 8 Things to Expect from CES, Consumer Tech’s Big Shindig.
Wired. January 4, 2019. https://www.wired.com/story/ces-2019-what-to-expect/
(Accessed 6 January 2019).
66 https://www.brightlocal.com/learn/voice-search-for-local-business-study/
67 http://insight.venturebeat.com/report/marketing-personalization-maximizing-
relevance-and-revenue?utm_source=vb&utm_medium=refer&utm_content=editorial-
post&utm_campaign=personalization-report (accessed 26 November 2016)
68 Eyefortravel Limited. (2017).
https://www.eyefortravel.com/sites/default/files/data_in_travel_report_draft7.pdf
(Accessed 30 March 2019).
69 Abramovich, Giselle. (2018). CMO.com. 3 AI-driven strategies for cruise lines in 2019.
11 November 2018. https://www.cmo.com/features/articles/2018/11/15/3-ai-driven-
strategies-for-cruise lines-in-2019.html#gs.COMQdxdU (Accessed 23 February 2019).
70 Adobe Experience Cloud. (2018). Creating the Ultimate Single Customer View. Adobe.
10 October 2018. https://theblog.adobe.com/creating-the-ultimate-single-customer-
view-with-adobe-campaign/ (Accessed 16 January 2019)

99
 ANDREW W. PEARSON

71 Adobe Experience Cloud. (2018) Adobe.com. Creating the Single Customer View with
Adobe Campaign. https://theblog.adobe.com/creating-the-ultimate-single-customer-
view-with-adobe-campaign/ (Accessed 12 January 2019).
72 Pega Marketing Product Overview. (2019). Pegasystems Inc., Cambridge, MA.
http://pega.com.
73 Abramovich, Giselle. (2018). CMO. The 5 Biggest Marketing Trends for 2019.
https://www.cmo.com/features/articles/2018/12/12/the-5-biggest-marketing-trends-
for-2019.html#gs.GwcjmLKw (Accessed 11 January 2019).
74 CB Insights. (2019). What’s Next in AI? Artificial Intelligence Trends. CB Insights.
https://www.cbinsights.com/research/report/ai-trends-2019/ (Accessed 10 February
2019).
75 Enterprise Content Team. Adobe. Incredible content, incredible experiences.
https://www.adobe.com/insights/deliver-experience-with-content-intelligence.html
(Accessed 8 January 2019).
76 Lovelock, C. a. (2010). Services Marketing, People, Technology, Strategy, Seventh
Edition. Prentice Hall.
77 https://en.wikipedia.org/wiki/Social_CRM
78 Greenberg, P. (2009, July 6). Time To Put a Stake in the Ground On Social CRM.
Retrieved from ZDnet.com: http://www.zdnet.com/blog/crm/time-to-put-a-stake-in-
the-ground-on-social-crm/829
79 Divol, R. E. (2012, April). Demystifying social media. Retrieved from Mckinsey.com:
http://www.mckinsey.com/insights/marketing_sales/demystifying_social_media
80 The Customer Satisfaction-Loyalty Relationship from Thomas O. Jones and W. Earl
Sasser, Jr., “Why Satisfied Customers Defect” Harvard Business Review, Nov.–Dec. 1995,
p. 91. Reprinted by permission of Harvard Business School.
81 Wangenheim, F. v. (2005). Postswitching Negative Word of Mouth. Journal of Service
Research, 8, No. 1, 67-78.
82 Rigby, Darrell. 2015. Management Tools 2015. An Executive’s Guide. Bain & Company.
http://www.bain.com/publications/articles/management-tools-customer-relationship-
management.aspx
83 https://en.wikipedia.org/wiki/Market_segmentation
84 MIT Technology Review. (2016). Transforming Customer Experiences by Harnessing
the Power of AI in CRM. www.technologyreview.com/media (Accessed 10 January 2018.
85 Stodder, David. TDWI Research. Best practices report, customer analytics in the age of
social media. Third quarter 2012. https://tdwi.org/research/2012/07/best-practices-
report-q3-customer-analytics-in-the-age-of-social-media.aspx?tc=page0 (Accessed 30
April 2019).
86 Cook, Rick. Control Group Marketing—With or Without CRM Software Systems.
crmsearch http://www.crmsearch.com/marketing-control-groups.php (Accessed 19
November 2017).
87 Huguesrey.com. 15 Applications of Artificial Intelligence in Marketing. October 26,
2017. https://huguesrey.wordpress.com/2017/10/26/15-applications-of-artificial-
intelligence-in-marketing-source-robert-allen/ (Accessed 18 March 2019).
88 Vicioso, Sara. Seer Interactive. Programmatic Advertising 101: How it Works. August
27, 2015. https://www.seerinteractive.com/blog/programmatic-advertising-101-works/
(Accessed 26 March 2019).
89 Shaw, Allie. (2017). VentureBeat. AI could save television advertising with advanced
personalization. October 28, 2017. https://venturebeat.com/2017/10/28/ai-could-save-
100
television-advertising-with-advanced-personalization/ (Accessed 15 April 2018).
 THE A.I. CRUISE LINE

90 Kumar, V. V. (2006). Knowing What to Sell, When, and to Whom. Harvard Business
Review.
91 Pierson, Lillian. Solving Real-World Problems with Nearest Neighbor Algorithms.
www.dummies.com. http://www.dummies.com/programming/big-data/data-
science/solving-real-world-problems-with-nearest-neighbor-algorithms/ (Accessed 20
November 2017).
92 Bea, Francis. March 25, 2012. Goodbye, anonymity: latest surveillance tech can search
up to 36 million faces per second. www.digitaltrends.com
http://www.digitaltrends.com/cool-tech/goodbye-anonymity-latest-surveillance-tech-
can-search-up-to-36-million-faces-per-second/
93 Eckerson, Wayne. 2007. Predictive Analytics, Extending the Value of Your Data
Warehouse Investment. TDWI Best Practices Report.
https://www.sas.com/events/cm/174390/assets/102892_0107.pdf
94 Reichheld, F. and Sasser, W.E. (1990). Zero defections: quality comes to services.
Harvard Business Review, 105-111.
95 White House Office of Consumer Affairs
96 Nielsen Company. (2012). Global Trust in Advertising and Brand Messaging. Nielsen
Company.
97 Cognizant. (2014, January). Retail Analytics: Game Changer for Customer Loyalty.
Retrieved from congnizant.com: http://www.cognizant.com/InsightsWhitepapers/Retail-
Analytics-Game-Changer-for-Customer-Loyalty.pdf (Accessed 20 November 2017).
98 Peterson, T. (2014, August 13). Facebook Now Tells Whether Mobile Ads Lead to
Desktop Purchases. Retrieved from AdAge: http://adage.com/article/digital/facebook-
makes-link-mobile-ads-desktop-purchases/294568/ (Accessed 20 November 2017).
99 https://www.chirpify.com/announcing-chirpify-moderation-brands-can-now-
moderate-social-triggers/ (Accessed September 7, 2017).
100 http://www.chiefmarketer.com/pro-awards-winners-2017 (Accessed September 7,
2017)
101 Lee Jin-soo. 24 March, 2017. Getting it right: mixing social and economic rewards in
hotel loyalty programmes. South China Morning Post.
http://www.scmp.com/business/companies/article/2081624/getting-it-right-mixing-
social-and-economic-rewards-hotel-loyalty
102 Milgrom, P.R., North, D.C. and Weingast, B.R. (1990). The role of institutions in the
revival of trade: The law merchant, private judges, and the Champagne fairs,” Economics
and Politics, 2(1):1-23.
103 Masterov, D. V., Mayer, U. F., and Tadelis, S. (2015) “Canary in the e-commerce coal
mine: Detecting and predicting poor experiences using buyer-to-seller messages,” In
Proceedings of the Sixteenth ACM Conference on Economics and Computation, EC '15,
pp81-93.
104 Li, L.I., Tadelis, S., and Zhou, X. (2016). Buying Reputation as a Signal of Quality:
Evidence from an Online Marketplace. NBER Working Paper No. 22584.
105 Nelson, P. (1970). Information and Consumer Behavior. Journal of Political
Economy, 78(2), 311-329. Retrieved from http://www.jstor.org/stable/1830691
106 Blake, Thomas, Nosko, Chris, and Tadelis, Steven. (2016), Returns to consumer
search: Evidence from ebay. In Proceedings of the 2016 ACM Conference on Economics
and Computation, pages 531–545. http://www.hbs.edu/faculty/conferences/2016-
dids/Documents/Tadelis_BNT_search_EC_vST.pdf (Accessed 22 January 2019).
107 Mattolli, Dana. (2012). Wall Street Journal. On Orbitz, Mac users steered to higher
pricier hotels. 101
August 23, 2012.
 ANDREW W. PEARSON

https://www.wsj.com/articles/SB10001424052702304458604577488822667325882
(Accessed 3 February 2019).

102
 CHAPTER THREE: ANALYTICS

Overview
As previously mentioned, Annie Eissler points out that, according to Nucleus
Research “analytics and business intelligence solutions deliver, on average,
$13.01 for every dollar spent.”27 She adds that, “We’re at a point where the hype
surrounding data analytics has converted into real, documented returns for
companies of all sizes and across all industries. But the truth is, leading
companies have been achieving double-digit return on investment (ROI) from
their analytics investments for several years now.”27
According to Nina Sandy, a Nucleus Research analyst, “Companies don’t have the
luxury anymore to wait weeks for reports on the profitability of business
decisions in increasingly fast paced markets.”27 “New analytics solutions are
being developed around this need where businesses can make better decisions,
faster.”27
The fact that so many software vendors are adding analytics to their standard
data mining, CRM, social media, marketing automation, and other offerings is
reducing prices across the board. For price, you obviously can’t beat open
source, but there is no free lunch in the software business and these open source
products do require skilled consultants to write the code and build the systems,
but these open source solutions can reduce the sting of the yearly
license/maintenance fees that comes with commercial software.
Eissler warns that, “You need the technology to enable analytics, but if you don’t
understand the technology that enables the analytics — or the business
application — then it won’t provide any value,”27 which is an accurate
assessment; “junk in, junk out,” as any good analyst will tell you. Eissler
concludes that, “The real value comes when you take the technological
component of analytics and apply it to a business component that — once
optimized — produces a solid ROI that continues to pay off over time.” 27
Analytics is, of course, a huge field. In this chapter, I will mostly focus on
customer analytics, which, when coupled with insights from social media data,
can enable organizations to make faster strides in predicting retention, attrition,
and return rates, with the goal of reducing customer churn, raising customer lift,
and/or increasing a whole host of other metrics.108
Sources such as transactions data, clickstream, as well as service and call center
records are also important for customer analytics.108 These can both improve
how a cruise line decides on characteristics for customer segmentation, and also

103
 ANDREW W. PEARSON

provide clues to emerging characteristics for the definition of new segments.108

As David Stodder explains in his article Customer Analytics in the Age of Social
Media108, “Firms can employ predictive modeling to test and learn from
campaigns so that they are able to select the most persuasive offers to put in
front of the right customers at the right time.”108
Webopedia.com says customer analytics “exploits behavioral data to identify
unique segments in a customer base that the business can act upon. Information
obtained through customer analytics is often used to segment markets, in direct
marketing to customers, predicate analysis, or even to guide future product and
services offered by the business.”109
In IBM’s Achieving Customer Loyalty with Customer Analytics110, IBM argues that
customer analytics can uncover “patterns and trends in customer behavior and
sentiment hidden among different types of customer data such as transactions,
demographics, social media, survey and interactions.” “The results of the
analysis are then used to predict future outcomes so businesses can make
smarter decisions and act more effectively.”110 Results from these models can
then be presented back to the business users in easily digestible dashboards and
scorecards.110 “Self-learning predictive models ensure that each new iteration of
customer analytics insight and the business decisions it drives become more
accurate and effective,” argues IBM.110
Customer analytics can also help determine which of a cruise line’s advertising
campaign or advertising partner’s pages have the highest landing rates, as well
as show conversion rates for all of a cruise line company’s advertising and
marketing budgets.
Mobile analytics can also display how many visitors downloaded material from a
site, which can help in factoring a company’s advertising and marketing budgets.
And, finally, mobile analytics can display which pages have the highest exit rates.
With this type of analysis, marketers can rapidly adjust marketing campaigns to
exploit the most effective ones and, conversely, trim the non-performing ones.
The biggest problem with any analytics procedure is filtering out the noise
associated with the data. Without clean data, “the trends, patterns, and other
insights hidden in the raw data are lost through aggregation and filtering.”108
Organizations need an unstructured place “to put all kinds of big data in its pure
form, rather than in a more structured data warehousing environment.”108 This
is because what might be considered just “noise” in the raw data from one
perspective could be full of important “signals” from a more knowledgeable
perspective.108 “Discovery, including what-if analysis, is an important part of
customer analytics because users in marketing and other functions do not always
know what they are looking for in the data and must try different types of
analysis to produce the insight needed.”108 As per Stodder, among the frequent
targets for analysis are the following108:

104
 THE A.I. CRUISE LINE

• Understanding sentiment drivers.

• Identifying characteristics for better segmentation.
• Measuring the organization’s share of voice and brand reputation
compared with the competition.
• Determining the effectiveness of marketing touches and messages in
buying behavior, i.e., attribution analysis.
• Using predictive analytics on social media to discover patterns and
anticipate customers’ problems with products and/or services.

Very important Somewhat important Somewhat unimportant

Not important Don't know

Marketing 52 29 7 7 5
Campaign management 47 27 10 9 7
Sales/sales reporting 45 34 8 8 5
Market research 43 36 9 7 5
Customer services/order management 43 32 10 11 5
Executive management 40 32 14 8 6
Advertising 39 26 12 18 5
Fraud/risk management 39 24 16 13 8
Finance 38 29 16 11 6
Product development 33 32 14 14 7
Call/contact center 33 29 12 18 8
Operations management/research 32 36 15 11 6
Web storefront/online presence 32 30 14 16 8
New media/social media dept. 31 31 17 13 8
Regulatory complaine/data governance 29 33 15 15 8
Public relations 26 32 20 14 8
Distribution, fulfillment, or logistics 25 27 16 23 9
Supply Chain 20 21 22 26 11
Event Management 14 35 23 18 10
Procurement 13 23 28 24 12

0% 20% 40% 60% 80% 100%

Figure 10: Importance of Customer Analytics Technology

Based on one answer per business function from 452 responses.
Source: TWDI Research108

105
 ANDREW W. PEARSON

TWDI’s research108 found that “the business functions or operations for which
respondents considered customer analytics most important were marketing
(81%, with 52% indicating “very important”), sales and sales reporting (79%, with
45% “very important”), and campaign management (74%, with 47% “very
important”).108 Market research (43% “very important”) and customer services
and order management (also 43% “very important”) were also high among
business functions regarded as critical to developers and consumers of customer
analytics (see Figure 10).108
The marketing department, “which in most organizations is empowered with the
responsibility for identifying, attracting, satisfying, and keeping customers, is
clearly the main stage for customer analytics.”108 Marketing departments and
functions are becoming increasingly qualitative.108 “Gut feelings” are being
replaced by data-driven decision-making.108 “Data drives the pursuit of efficiency
and achievement of measurable results. Marketing functions are key supporters
of ‘data science,’ which is the use of scientific methods on data to develop
hypotheses and models and apply iterative, test-and-learn strategies to
marketing campaigns and related initiatives.”108
Customer analytics can be a very effective tool for micro-targeting customers
with customized marketing offers and promotions. 110 Obviously, when an
organization “attempts to cross-sell or up-sell a customer, a product or service
they desire, it can enhance satisfaction.”110 However, unwanted marketing
campaigns can do just the opposite, annoying customers, thereby eroding loyalty
and, potentially, hurting sales.110 Even worse, unwanted marketing campaigns
can give customers the impression that the organization doesn’t care about their
wants, desires, needs and preferences.”110
Customer analytics can help determine which marketing interactions are likely
to please individual customers and which will not.” 110 Sales functions can be
important beneficiaries of customer analytics as well.108 Stodder argues that,
“Sales reports typically focus on providing visibility into the pipeline. Managers
can use data insights to improve sales forecasting of potential revenues based
on deeper knowledge of priority opportunities, most valued customer segments,
and more.”108
“Customer service and order management can use customer analytics to get a
more subtle and substantial view of what actions impact customer experiences
and satisfaction.”108 Contact centers can utilize “customer analytics to help tune
performance metrics closer to real time, so that each day’s agents are guided, if
not incentivized, to interact with customers in beneficial ways.”108
Analytics can also “help service and order management functions move away
from one-size-fits-all approaches to customers and instead tune and tailor
interactions more personally based on knowledge of particular types or
segments, such as regions or nationalities.”108 “Finally, through integrated views

106
 THE A.I. CRUISE LINE

of customer data and analytics, service and order management functions are
able to work in better synchronicity with the organization’s marketing, sales, and
other business functions.”108 Customer analytics can be used to understand
where marketing campaigns are working as well.
In the words of business management guru Thomas Davenport, “Organizations
are competing on analytics not just because they can — business today is awash
in data crunchers — but also because they should.”111 Although these words
were said more than ten years ago, they might be more relevant today than ever
before. Davenport adds, “Business processes are among the last remaining
points of differentiation. And analytics competitors wring every last drop of
value from those processes.”111 “Customer analytics helps organizations
determine what steps will give them competitive advantages, increase
profitability, and identify waste in business processes,” Davenport argues.111
With the steep drop in RAM prices, in-memory solutions are all the rage these
days and they allow analytics to reach a whole new level of sophistication. Today,
creativity is becoming the differentiator; today’s overriding philosophy might be
“Those who analyze best win.”
With products and services being commoditized at such a rapid rate today,
customer loyalty has become more elusive than ever.108 “Innovation must be
constant and must immediately address why an organization is losing customers.
Information insights from analytics can help an organization align product and
service development with strategic business objectives for customer loyalty.”108
In addition, these insights can help cruise line organizations be selective in how
they deploy marketing campaigns and customer-touch processes so that they
emphasize features in new products and services that are important to
customers.
When TDWI Research examined the business benefits sought from customer
analytics (see Figure 11), respondents cited giving executive management
customer and market insight as the most important benefit (71%). 108 The second
most important benefit was being able to react more quickly to changing market
conditions (62%).108
Improving customer satisfaction and gaining a complete picture of a customer’s
activity across business channels — two areas that would be considered a part
of the “Customer Experience Management” (CEM) process — are critical to
identifying what steps an organization must take to build and retain customer
loyalty.108 The remaining items fall mainly into the categories of business
intelligence, marketing, and brand management and they are extremely
important to a cruise line company as well.

107
 ANDREW W. PEARSON

What are the most important business benefits that your organzation
seeks to achieve from implementing customer analytics technologies
and methods? (Please select all that apply)
Give executive management customer/market
71%
insight
React more quickly to changing market
62%
conditions
Improve customer satisfaction with experiences
60%
and engagement
Gain complete view of the customer activty
56%
across channels

Identify potential competitive advantages 50%

Discover what influences buying behavior 49%

Apply insights to product/service development 44%

Manage and target marketing mix 43%

Identify financial impact of marketing actions 42%

Develop more effective loyalty programs 32%

Manage brands effectively in social media 31%

Gain accurate attribution of conversion to

26%
marketing touches

0% 10% 20% 30% 40% 50% 60% 70% 80%

Figure 11: What Are the Important Business Benefits of Customer Analytics?
Based on 2,573 responses from 454 respondents; almost six responses, on
average.
Source: TDWI Research108

Organizations are becoming open to customer analytics because they are

interested in discovering how a marketing department can be more effective,
not just more efficient.108 “Whereas other types of applications for e-commerce,
fulfillment, or marketing automation help organizations determine how to get
things done (e.g., getting goods delivered at the right time, executing a
marketing campaign), customer analytics helps organizations answer who, what,
when, where, and why questions,” argues Scott Groenendal, program director
of customer analytics market strategy for IBM Business Analytics.108 “They can
find answers to questions such as: What channel should I communicate through?
When is the best time to target this person, and why would they be receptive to

108
 THE A.I. CRUISE LINE

this message?” adds Groenendal.108

Individual creativity, personal experiences, customer behavior and marketing
context are critical components of consumer marketing decisions.108 “The role
of customer analytics is not necessarily to replace these, but to help decision
makers come to fact-based conclusions through better knowledge of the
organization’s customers and markets.”108
Just as importantly, analytics are needed for scalability.108 “Just as automation is
necessary to run hundreds or thousands of marketing campaigns, customer
analytics processes are important for supplying intelligence and guidance to
those automated routines. Customer analytics can provide the brains to match
the marketing systems’ brawn.”108
With the commoditization of products and services, customer loyalty can be
elusive; innovation must be constant and it should help to reveal why an
organization might be losing its customer base.108 “Information insights from
analytics can help organizations align product and service development with
strategic business objectives for customer loyalty.”108 These insights can also
help an organization be selective about how they deploy their marketing
campaigns and customer-touch processes so that they can emphasize features
in new products and services that are important to each specific customer. 108
In its Achieving Customer Loyalty with Customer Analytics110, IBM describes one
of its studies that asked some of the world’s leading company CEOs and CMOs
what their number one priority was.110 The CEOs answered that it was to engage
customers, while the CMOs said it was to enhance customer loyalty.110 The study
argued that forward-thinking companies were using customer analytics to110:
• Guide front-line interactions with customers.
• Create and execute customer retention strategies.
• Prompt people or systems to proactively address customer satisfaction
issues.
• Guide product planning to fulfill future customer needs.
• Hire and train employees to act upon customer insights and improve
loyalty.
• Align operations to focus on satisfying customers.
The Customer Analytics in the Age of Social Media 108 report concluded that the
importance of customer analytics is in the boardroom; “overwhelmingly,
respondents cited giving executive management customer and market insight
(71%) as the most important business benefit that their organization seeks to
achieve from implementing customer analytics.”108 “This percentage rises to
81% when survey results are filtered to see only the responses from those who
indicated ‘strong acceptance’ of data-driven customer analytics over gut feel.”108
The second highest benefit cited at 62% was “the ability to react more quickly to

109
 ANDREW W. PEARSON

changing market conditions, which speaks to the need for customer data insights
to help decision makers address competitive pressures from rapid product or
service commoditization.”108
Customer analytics can also provide answers to questions like, “When in the life
cycle are customers most likely to churn? What types of products or services
would prevent them from churning, and when should they be offered
complimentary items? When is it too costly to try to keep certain customers?108
Businesses can realize significant ROI from investing in customer analytics as it
can improve the marketing department’s efficiency and effectiveness.108
However, customer analytics ROI is a difficult thing to fully quantify — especially
in the analytics space. Better customer knowledge equates to more optimized
marketing spend because a business can focus its resources on those campaigns
that have the highest predicted chances of success for particular segments, as
well as cutting off or avoiding those that have the least.108
“By using analytics to eliminate mismatches of campaigns targeting the wrong
customers or using the wrong messages and offers, marketing functions can
reduce wasteful spending and increase gains relative to costs.”108 Customer
segmentation allows organizations to move “away from one-size-fits-all, brand-
level-only marketing and toward the ‘market of one’: that is, personalized, one-
to-one marketing.”108
Reaching a customization and customer service level that makes a customer feel
as though he or she is a preferred customer is not easy, scaling that up so that
an entire database of customers feel that they are unique and receiving
outstanding customer service is even more challenging, but, in this day and age,
it is almost a necessity if a company wants to provide good and engaging
customer service.
TDWI Research108 examined the importance of accomplishing various objectives
for gaining positive ROI from customer analytics (see Figure 12). “Using customer
analytics to target cross-sell and up-sell opportunities was the objective cited by
the biggest percentage of respondents (54%).”108 This objective is about gaining
more value from existing customers by understanding their purchasing habits
and trying to get them to buy more products more often. 108 “Some organizations
(18%) are implementing an advanced technique called ‘uplift modeling’ (also
called incremental or true-lift modeling), which enables marketers to use data
mining to measure the impact and influence of marketing actions on
customers.”108 Insights such as these allow marketers to develop new kinds of
predictive models to determine the best prospects for up-sell and cross-sell
offerings.108 “As firms scale up to execute large numbers of campaigns across
multiple channels, the efficiency gained from predictive modeling can be critical
to marketing spending optimization,” argues Stodder.108

110
 THE A.I. CRUISE LINE

In your organization, which of the following marketing

objectives are most important to achieve for customer analytics
to deliver a return on investment? (Please select all that apply.)

54%

49%

47%

42%

39%

37%

35%

32%

26%

23%

18%

14%

Figure 12: Which Are the Most Important Business Objectives When it Comes
to Customer Analytics?
Source: TWDI Research108, Based on 1,625 responses from 432 respondents;
almost four responses per respondent, on average.

Analytics can improve marketing performance by quantifying a customer’s

lifetime value as well as customer worth at the many different stages in the
customer life cycle.108 “If organizations can identify their most valuable
customers they can determine if they are worthy of retention efforts and
resources because of the returns they will provide.”110 For instance, it may not
be worth the time, effort, and expense to retain a low value customer, unless
customer analytics reveals that this low-spend customer actually has a lot of
social influence.110 Armed with this information, managers can align their
deployment of resources to achieve the highest value, as well as avoid the costs
and inefficiencies of marketing to the wrong people at the wrong time.108
Organizations have long used demographics such as gender, household size,
education, occupation, and income to segment customers.108 Data mining
techniques let organizations segment much larger customer populations and,

111
 ANDREW W. PEARSON

perhaps, more importantly, determine whether to apply new characteristics that

refine segmentation to fit the specific attributes of the organization’s products
and services.108
“Customer analytics using data mining tools improves the speed of segmentation
analysis over manual and spreadsheet efforts that are often used in less mature
organizations.”108 Speed is a vital ingredient for marketing initiatives that are
time sensitive, particularly for those companies that need to provide real-time
cross-sell and up-sell offers to customers clicking through web pages.108 Today,
personalized web pages can be rendered during the web page load and elements
of the page can take into account past purchase history, clickstream information,
as well as a whole host of other things.
In its Achieving Customer Loyalty with Analytics110, IBM argues that customer
analytics can provide businesses with the ability to:
• Analyze all data types to gain a 360-degree view of each individual
customer.
• Employ advanced algorithms that uncover relevant patterns and causal
relationships that impact customer satisfaction and loyalty.
• Build predictive models that anticipate future outcomes.
• Learn from every customer interaction and apply lessons to future
interactions and strategies.
• Deploy customer insights to decision-makers and front-line systems.
• Improve sales forecasting and help minimize sales cycles.
• Measure and report on marketing performance.
“The next most common objectives in the research were predicting retention,
attrition, and churn rates (47%) and determining lifetime customer value
(42%).”108 Churn can cost organizations heavily, both from the loss of profits
from existing customers as well as in the high price of attracting new ones.
“Attrition or churn analysis methods are aimed at discovering which variables
have the most influence on customers’ decisions to leave or stay.”108
With data mining and predictive analytics, organizations can learn which attrition
rates are acceptable or expected for particular customer segments and which
rates could be highly detrimental to the bottom line.108 “Predictive customer
analytics can play a major role in enabling organizations to discover and model
which customers are most likely to leave, and from which segments.”108
With social media added to the mix, as well as clickstreams, and other behavioral
data, the volume and variety of data is exploding.108 “Social networking sites such
as Facebook, Twitter, LinkedIn, and MySpace have files containing petabytes of
data, often in vast Hadoop clusters.”108 Weibo and WeChat add another
hundreds of millions of users to the mix and with it petabytes of data.
“Advertising concerns are recording tens of millions of events daily that

112
 THE A.I. CRUISE LINE

organizations want to mine in near real time to identify prospects,” Stodder

notes.108 Businesses of all kinds want to use predictive models and score event
and transaction details as fast as they come in so that they can gain insight into
individual shopping behavior.108 Insights that they hope will give them an
advantage over their competitors, but this is dangerous and expensive territory
to chart, especially if done incorrectly.
The “data sources most commonly monitored for customer analytics are
customer satisfaction surveys (57%) and customer transactions and online
purchases (55%). Just under half (44%) are monitoring Web site logs and
clickstream sources. In addition to monitoring customer satisfaction surveys,
about half (48%) of organizations surveyed are studying call and contact center
interactions.”108
Customer satisfaction surveys are usually conducted in person, on a website,
over the phone or through traditional mail and e-mail channels.108 Because this
includes both semi-structured data and unstructured comments, data collection
can be difficult.108 “Standard questions inquire about a customer’s satisfaction
with purchases, the services they received, and the company’s brands overall.
Other questions address the customer’s likelihood of buying from the company
again and whether they would recommend the firm to others.”108
Text analytics can be used to increase the speed, depth, and consistency of
unstructured content analysis far greater than what can be done manually. 108
“More advanced analytics can look for correlations between satisfaction ratings,
commented sentiments, and other records, such as first-call-resolution
metrics.”108
“To analyze data generated by social media networking services such as Twitter,
Facebook, Weibo, and LinkedIn, many organizations are implementing Hadoop
and NoSQL technologies, which do not force a schema on the source data prior
to storage, as traditional BI and data warehousing systems do.”108 Because of
this, the discovery analytics processes can run against the raw data.108
“Customer analytics tools need to be able to consume data from sources such as
Hadoop clusters and then integrate the insights into overall customer profiles,”
advises Stodder.108
The data sources can be varied for these technologies and methods; “they
include transaction data, clickstreams, satisfaction surveys, loyalty card
membership data, credit card purchases, voter registration, location data, and a
host of [other] demographic data types.”108

Data Mining
In his paper The CRISP-DM model, the new blueprint for data mining 112, C.
Shearer introduces the concept of the “Cross-industry standard process for data

113
 ANDREW W. PEARSON

mining”, which is more commonly known by its acronym CRISP-DM. It is a “data

mining process model that describes commonly used approaches that data
mining experts use to tackle problems.”112 It is currently the de facto standard
for developing data mining and data discovery projects.112
In their paper Methods for mining HTS data113, Harper and Pickett break the
CRISP-DM process of data mining into the following six major phases:
1. Business understanding — focuses on understanding the project
objectives and requirements purely from a business perspective, and
then “converting this knowledge into a data mining problem definition,
and a preliminary plan designed to achieve the objectives.” 113 “A
decision model, especially one built using the Decision Model and
Notation standard can be used.”113
2. Data understanding — this starts with an “initial data collection and
proceeds with activities in order to get familiar with the data, to identify
data quality problems, to discover first insights into the data, or to
detect interesting subsets to form hypotheses for hidden
information.”113
3. Data preparation — phase that covers “all activities to construct the
final dataset (data that will be fed into the modeling tool(s)) from the
initial raw data. Data preparation tasks are likely to be performed
multiple times, and not in any prescribed order. Tasks include table,
record, and attribute selection as well as transformation and cleaning
of data for modeling tools.”113
4. Modeling — various modeling techniques are selected and applied in
this phase, and their parameters are calibrated to optimal values. 113
“Typically, there are several techniques for the same data mining
problem type. Some techniques have specific requirements on the form
of data. Therefore, stepping back to the data preparation phase is often
needed.”113
5. Evaluation — At this project stage, model (or models) that appear to
have high quality from a data analysis perspective should have been
made.113 “Before proceeding to final model deployment, it is imperative
to more thoroughly evaluate the model, and review the steps executed
to construct the model, to be certain it dovetails with the business
objectives.”113 A key objective here is to determine if any important key
business objective has been left out.113 At the end of this phase, a
decision on whether to use the data mining results should be
reached.113
6. Deployment — Creation of the model is generally not the end in and of
itself.113 “Even if the purpose of the model is to increase knowledge of
the data, the knowledge gained will need to be organized and
presented in a way that is useful to the customer.” 113 “Depending on
the requirements, the deployment phase can be as simple as generating

114
 THE A.I. CRUISE LINE

a report or as complex as implementing a repeatable data scoring (e.g.

segment allocation) or data mining process.”113 “In many cases it will be
the customer, not the data analyst, who will carry out the deployment
steps. Even if the analyst deploys the model it is important for the
customer to understand up front the actions which will need to be
carried out in order to actually make use of the created models.” 113
The sequence of the phases (see Figure 13) is not strict and Harper and Pickett
argue that moving back and forth between different phases is often required so
flexibility is important.113 “The arrows in the process diagram indicate the most
important and frequent dependencies between phases,” contend Harper and
Pickett.113 “The outer circle in the diagram symbolizes the cyclic nature of data
mining itself,” add Harper and Pickett.113
The data mining processes continues long after a solution has been deployed,
argue Harper and Pickett.113 The “lessons learned during the process can trigger
new, often more focused business questions and subsequent data mining
processes will benefit from the experiences of previous ones,” the writers
conclude.113

Figure 13: CRISP DM

Source: Wikipedia

In the SAS Institute Best Practices paper Data Mining and the Case for
Sampling114, SAS defines data mining “as the process used to reveal valuable

115
 ANDREW W. PEARSON

information and complex relationships that exist in large amounts of data.” 114
For SAS, data mining is an iterative process, divided into five stages that are
represented by the acronym SEMMA.114 “Beginning with a statistically
representative sample of data, the SEMMA methodology — which stands for
Sample, Explore, Modify, Model, and Assess — makes it easy for business
analysts to apply exploratory statistical and visualization techniques, select and
transform the most significant predictive variables, model the variables to
predict outcomes, and confirm a model’s accuracy,” argues SAS. 114 According to
SAS, the SEMMA methodology is broken down into the following steps 114:
• “Sample the data by creating one or more data tables. The samples
should be big enough to contain the significant information, yet small
enough to process quickly.”114
• “Explore the data by searching for anticipated relationships,
unanticipated trends, and anomalies in order to gain understanding and
ideas.”114
• “Modify the data by creating, selecting, and transforming the variables
to focus the model selection process.”114
• “Model the data by allowing the software to search automatically for a
combination of data that reliably predicts a desired outcome.” 114
• “Assess the data by evaluating the usefulness and reliability of the
findings from the data mining process.”114
SEMMA is itself a cycle, in which the internal steps can be performed iteratively,
as needed.114 SAS advises that projects following SEMMA “can sift through
millions of records and reveal patterns that enable businesses to meet data
mining objectives such as”114:
• Segmenting customers accurately into groups with similar buying
patterns.
• Profiling customers for individual relationship management.
• Dramatically increasing response rate from direct mail campaigns.
• Identifying the most profitable customers and the underlying reasons
for his or her popularity.
• Understanding why customers leave for competitors (attrition, churn
analysis).
• Uncovering factors affecting purchasing patterns, payments and
response rates.
• Increasing profits by marketing to those most likely to purchase.
• Decreasing costs by filtering out those least likely to purchase.
• Detecting patterns to uncover non-compliance.114

116
 THE A.I. CRUISE LINE

Artificial Intelligence & Machine Learning

According to Wikipedia, Machine Learning (ML) is the subfield of computer
science that “explores the construction and study of algorithms that can learn
from data. Such algorithms operate by building a model based on inputs and
using that to make predictions or decisions, rather than following only explicitly
programmed instructions.”115
ML “evolved from the study of pattern recognition and computational learning
theory in artificial intelligence” and it “explores the study and construction of
algorithms that can learn from and make predictions on data — such algorithms
overcome following strictly static program instructions by making data driven
predictions or decisions, through building a model from sample inputs.” 115

Figure 14: Machine Learning for cruise lines

Source: www.intelligencia.co

As per Wikipedia, ML can be broken down into the following three categories
(see Figure 14)115:

117
 ANDREW W. PEARSON

1. Supervised learning: The computer is presented with example inputs

and their desired outputs, given by a “teacher”, and the goal is to learn
a general rule that maps inputs to outputs.
2. Unsupervised learning: No labels are given to the learning algorithm,
leaving it on its own to find structure in its input. Unsupervised learning
can be a goal in itself (discovering hidden patterns in data) or a means
towards an end (feature learning).
3. Reinforcement learning: A computer program interacts with a dynamic
environment in which it must perform a certain goal (such as driving a
vehicle), without a teacher explicitly telling it whether it has come close
to its goal or not. Another example is learning to play a game by playing
against an opponent.
There are so many use cases for ML and deep learning in the cruise line industry
that it is impossible to create an exhaustive list here, but it is particularly useful
for marketing personalization, customer recommendation, spam filtering,
network security, optical character recognition (OCR), voice recognition,
computer vision, fraud detection, predictive asset maintenance, optimization,
language translations, sentiment analysis, and online search, amongst many
others uses.
ML can be used to spot credit card or transaction fraud in process; ML can build
predictive models of credit card transactions based on their likelihood of being
fraudulent and the system can compare real-time transactions against these
models. When the system spots potential fraud it can alert either the bank or
the cruise ship where the transaction occurs.
Although ML and data mining often employ the same methods and overlap
significantly, they do differ quite significantly. As Wikipedia explains115:
“While machine learning focuses on prediction, based on
known properties learned from the training data, data mining
focuses on the discovery of (previously) unknown properties in
the data (this is the analysis step of Knowledge Discovery in
Databases). Data mining uses many machine learning
methods, but with different goals; on the other hand, machine
learning also employs data mining methods as “unsupervised
learning” or as a preprocessing step to improve learner
accuracy. Much of the confusion between these two research
communities (which do often have separate conferences and
separate journals, ECML PKDD being a major exception) comes
from the basic assumptions they work with: in machine
learning, performance is usually evaluated with respect to the
ability to reproduce known knowledge, while in Knowledge
Discovery and Data Mining (KDD) the key task is the discovery
of previously unknown knowledge. Evaluated with respect to

118
 THE A.I. CRUISE LINE

known knowledge, an uninformed (unsupervised) method will

easily be outperformed by other supervised methods, while in
a typical KDD task, supervised methods cannot be used due to
the unavailability of training data.”
WEKA, a comprehensive collection of machine-learning algorithms for data
mining tasks written in Java and released under the GPL, contains tools for data
pre-processing, classification, regression, clustering, association rules, and
visualization. It has a very minimal learning curve compared to products like
SAS’s Enterprise Miner. However, unlike SAS, it can become quite inefficient with
larger datasets, at least as of the date of this publication.
Python and R are the most popular open source solutions used for ML and they
both have a large user base community. Scikit-learn combined with Pandas,
Numpy, Seaborn and Matplotlib make implementing ML algorithms in Python
very versatile and these provide more customization and utilization than R.
The R community has a large and active user base. R’s libraries contain a wide
variety of statistical and graphical techniques as well. These include linear and
nonlinear modeling, classical statistical tests, time-series analysis, classification,
clustering, amongst others. Due to its S heritage, R also has strong object-
oriented programming capabilities.
Other ML software includes Matlab, Scikit, Accord, Apache’s Mahout, Spark’s
MLLib, H2O on Hadoop, ConvNteJS, SPSS, SAP’s Predictive Analytics library, even
SQL Server is powerful enough to build some ML models on it.
ML can help a cruise line discover customer segments that they may not realize
they had. Armed with this kind of information, cruise lines can understand what
matters most to its customers at the individual and personalization level, which
will enable them to anticipate their customer’s needs before even the customers
are aware of them. Even more, cruise lines can understand key characteristics of
their most profitable customers and recognize the next important ones when he
or she happens to login onto the cruise line’s website or steps onto a cruise ship.
The use of deep neural networks and image classifiers can analyze and parse
images, which can enable cruise line marketers to monitor the images that
provide the highest selling and conversion rates through each ecommerce
channel.
ML can also be used to compute dynamic clusters of customers to create fluid
segmentation in real-time. As consumer buying habits or booking patterns
evolve, fluid segmentation ensures the cruise line continues to reach the right
guests, at the right time, at the right price, through the right channels, with the
right offer.
Today, the importance of personalization in customer experience initiatives can’t
be underestimated. In his article 5 Ways AI Will Boost Personalization in Digital

119
 ANDREW W. PEARSON

Marketing116, Dirk Vogel argues that AI will radically change the marketing
landscape by allowing the following:
• Personal shopping for everyone.
• Utilizing chatbots to increase customer service.
• Seamless programmatic media buying.
• Predictive customer service.
• Optimizing marketing automation.
According to Vogel, “Shopping online creates rich data footprints regarding the
individual preferences, spending habits and preferred channels of individual
consumers. Feeding these digital breadcrumbs into an AI-engine helps bring
curated shopping journeys to mass audiences.” 116
As Amazon, Pandora, and Netflix have proven, personalized shopping for
everyone is a winning formula; “Using an advanced recommendation-AI, e-
commerce leader Amazon creates more than 35% of its total revenues with
personalized shopping recommendations,” states Vogel. 116 “Taking
personalization to the next level, artificial intelligence also allows for predicting
the kinds of purchases consumers are going to make before they even know it,”
notes Vogel.116
“Customer service is still where today’s brands are dropping the ball,” Vogel
believes, adding that, “Only 35% of companies are able to identify their
customers at the moment of contact (Selligent survey) — with customers
potentially unfriending brands and taking their business elsewhere.”116
It may sound counter-intuitive, Vogel argues, “but automated bots can create
lifelike, seamless customer service experiences, addressing the consumer on
their purchase history and known preferences.” 116 One of the standouts, Vogel
notes is Facebook’s “M” technology, which is embedded in the Messenger
app.116 “The AI delivers personalized product, travel and restaurant
recommendations, while troubleshooting technical problems,” explains
Vogel.116
Although chatbots are cheaper than handling customer service inquiries over the
phone, there’s a catch as chatbots can only deliver highly personalized and
contextual assistance if they have access to universal consumer profiles that are
populated by real-time data.116 This means, done correctly, developing chatbots
is an expensive upfront investment, it is an investment that should be done
company-wide, not siloed by just the marketing or customer service department,
as information that chatbots tap into are useful throughout an organization.
On the marketing side, AI may deliver that extra dash of relevancy programmatic
advertising has been waiting for all these years.116 “On the consumer side, AI
helps create individualized display ads that website visitors want to see,” 116 while
on the accounting side, “the bots handle invoicing and payment for these ad

120
 THE A.I. CRUISE LINE

transactions, giving marketers more time to focus on the big picture.” 116
With AI, predictive customer service and marketing could be just around the
corner.116 “What may sound like a scenario from Minority Report is already being
beta tested: Intel subsidiary Saffron has created an artificial intelligence that is
able to predict with 88% certainty why, on which channel, and for which product
individual customers will seek help next. ‘We’ve been expecting your call,’ never
rang more true,” Vogel states.116
One of the biggest problems in corporate marketing is hitting the customer with
automated marketing offers too often.116 In the future, “AI will analyze a
consumer’s purchase history and email habits to choose the optimal time for
hitting the inbox with content that’s bound to boost open rates and
conversions,” Vogel contends.116
Another interesting use of AI is what Pinterest is doing with its visual search
technology. According to Lauren Johnson’s Adweek article Pinterest Is Offering
Brands Its Visual Search Technology To Score Large Ad Deals 117, “The visual
search technology is Pinterest’s version of AI and human curation that lets
consumers snap a picture of IRL things and find similar items online. Taking a
picture of a red dress for example, pulls up posts of red dresses that consumers
can browse through and shop,” states Johnson.117
“The idea is to give people enough ideas that are visually related so that they
have a new way to identify and search for things,” said Amy Vener, retail vertical
strategy lead at Pinterest.117 “From a visual-discovery perspective, our
technology is doing something similar where we’re analyzing within the image
the colors, the shapes and the textures to bring that to another level of
dimension,” Vener adds.117
Utilizing the technology, someone who points his or her phone’s camera at a
baby crib will receive recommendations for similar baby products.117
“Eventually, all of Target’s inventory will be equipped with Pinterest’s
technology to allow anyone to scan items in the real world and shop similar items
through Target.com,” states Johnson. “Target is the first cruise line to build
Pinterest’s technology into its apps and website, though the site also has a deal
to power Bixby, Samsung’s AI app that works similarly.” 117
“We’re now in a place where we’re using Pinterest as a service to power some
visual search for other products,” Vener said. “I think there’s an opportunity for
cruise lines to be a little more of a prominent player when it comes to visual
discovery.”117

7 Patterns of AI
In her Forbes article The Seven Patterns of AI118, Kathleen Walch lays out a theory
that, regardless of the application of AI, there are seven commonalities to all AI

121
 ANDREW W. PEARSON

applications. These are “hyperpersonalization, autonomous systems, predictive

analytics and decision support, conversational/human interactions, patterns and
anomalies, recognition systems, and goal-driven systems.”118 Walch adds that,
“Any customized approach to AI is going to require its own programming and
pattern, but no matter what combination these trends are used in, they all follow
their own pretty standard set of rules. These seven patterns are then applied
individually or in various combinations depending on the specific solution to
which AI Is being applied.”118
The ‘Hyperpersonalization Pattern’, which can be boiled down to the slogan,
‘Treat each customer as an individual’ is defined as “using machine learning to
develop a profile of each individual, and then having that profile learn and adapt
over time for a wide variety of purposes including displaying relevant content,
recommend relevant products, provide personalized recommendations and so
on.”118 The objective here is to “treat each individual as an individual,” 118 which
is far from a simple thing to do.
”Implementations of the hyperpersonalization particular pattern include
creating personalized recommendations based off of browsing patterns and
searches,” says Walch.118 Not only limited to the marketing industry,
hyperpersonalization is being implemented in industries such as finance,
healthcare, as well as in personalized fitness and wellness applications. 118 For
example, in the US the FICO credit score is being augmented using AI.118
According to Walch, the FICO credit score “is used to lump individuals together
who might otherwise have vastly different amounts of credit worthiness and
penalizes groups of individuals who lack credit history.” 118 Walch argues that,
“By moving away from using the traditional FICO score into something that treats
each individual as an individual we might get more accurate pictures of
individuals to see just how likely they are to pay back loans.” 118
Walch’s second AI pattern is ‘Autonomous systems’, which should reduce the
need for manual labor, which is paramount in an industry like the cruise line
industry, which often has to utilize under skilled employees.
“Autonomous systems are physical and virtual software and hardware systems
that are able to accomplish a task, reach a goal, interact with their surroundings,
and achieve an objective with minimal human involvement,” says Walch. 118
“Where the primary objective of hyper-personalization is to treat people as
individuals, the goal of autonomous systems is to streamline things with as little
human interaction as possible,” notes Walch. 118 “The autonomous pattern
requires machine learning capability that can independently perceive the
outside world, predict the future behavior of external elements, and plan for
how to deal with those changes,” adds Walch.118
“Obvious applications of this pattern include autonomous machines and vehicles
of all sorts includes cars, boats, trains, airplanes, and more.”118 “However this

122
 THE A.I. CRUISE LINE

pattern also includes autonomous systems including autonomous

documentation and knowledge generation, autonomous business processes,
and cognitive autonomation. These include systems that can operate in close
proximity to humans, including preferential decision making,” notes Walch.118
AI powered predictive analytics, which are very much a topic of this chapter, “is
defined as using machine learning and other cognitive approaches to understand
how past or existing behaviors can help predict future outcomes or help humans
make decisions about future outcomes based on these patterns. The objective
of this pattern is helping humans make better decisions.” 118
Walch explains that, “Some uses of this pattern include assisted search and
retrieval, predicting some future value for data, predicting behavior, predicting
failure, assisted problem resolution, identifying and selecting best fit, identifying
matches in data, optimization activities, giving advice, and intelligent
navigation.”118 The idea is for AI powered predictive analytics to helps users
make better decisions, providing augmented intelligence capabilities. Machine
learning can be a constantly process, adapting over time to provide better
results.118
The ‘Conversational Pattern’ allows machines to communicate as humans do. 118
The conversational/human interaction pattern “is defined as machines and
humans interacting with each other through conversational forms of interaction
and content across a variety of methods including voice, text, and image
forms.”118 As Walch explains, “This includes machine to human, human to
machine, and back and forth human and machine interaction. The objective of
this pattern is enabling machines to interact with humans how humans interact
with each other.”118
Chatbots, voice assistants, and sentiment, mood and intent analysis all fit within
this category.118 As Walch points out, this systems are “trying to understand the
intent behind human interactions.”118 They “can also be used to facilitate human
to human interaction through translation.”118 “The big thing to remember is that
this pattern is used to create an easier way for humans to interact with each
other and machines through methods that are natural or comfortable for
humans,” concludes Walch.118
“Machine learning is particularly good at identifying patterns and finding
anomalies or outliers,” says Walch.118 She adds that, “The ‘pattern-matching
pattern’ is one of the repeating approaches to AI projects that has seen wide and
increasing adoption.”118 The goal of the ‘Patterns and Anomalies’ pattern of AI is
“to use machine learning and other cognitive approaches to learn patterns in the
data and learn higher order connections between data points to see if it fits an
existing pattern or if it is an outlier or anomaly.” 118 “The object of this pattern is
to find what fits with existing data and what doesn’t,” add Walch. 118
“Applications of this pattern include fraud and risk detection to see if things are

123
 ANDREW W. PEARSON

out of the ordinary or expectations are happening. Another application is finding

patterns among data, and helping to minimize or fix human mistakes,” notes
Walch.118 “This pattern also includes predictive text, where it can analyze
patterns in speech and grammar to help suggest words to choose to speed up
the writing process,” adds Walch.118
“One of the big advancements in machine learning is the use of deep learning to
greatly improve the accuracy of recognition-related tasks such as image, video,
audio, and object recognition, classification, and identification,” explains
Walch.118 “The recognition pattern is defined as using machine learning and
other cognitive approaches to identify and determine objects or other desired
things to be identified within image, video, audio, text, or other primarily
unstructured data,” says Walch.118 “The objective of this pattern is to have
machines identify and understand things,” notes Walch.118
“Image and object recognition, facial recognition, audio and sound recognition,
handwriting and text recognition, and gesture detection” are all examples of this
well-developed pattern, which computers excel at. 118 Google, Facebook, Apple,
Samsung, Huawei and a whole host of other companies are investing heavily in
recognition systems. In China, “one of the most well-funded AI companies,
Sensetime, is focused on facial recognition applications for the Chinese
government.”118
The ‘Goal-Driven Systems Pattern’ is the seventh of Walch’s AI patterns.118 For
decades, machines have been beating humans at easily conquered games like
checkers and chess. Now, through the power of reinforcement learning and
much more powerful computers, machines can now beat human at some of the
most complex games imaginable, including Go and multi-player games like Dota
2.118 “AlphaGo and AlphaZero were created by Google’s DeepMind division
under the theory that through goals, computers could learn anything through
game play,” explains Walch.118 DeepMind, the company behind AlphaGo and
AlphaZero, believes that, “Games are just the beginning to solutions that could
potentially even lead to breakthroughs in solving long-hoped for goals in
Artificial General Intelligence (AGI).”118
“Games are not the only possibility for goal-driven systems,” says Walch, adding
that, “With the power of reinforcement learning and other machine learning
techniques, organizations can apply machine learning and other cognitive
approaches to give their systems the ability to learn through trial and error.” 118
“This is useful for any situation where you want to have the system find the
optimal solution to a problem. The main learning approach for this pattern is
through reinforcement learning,” says Walch.118 “Examples in the pattern can
include game playing, resource optimization, iterative problem solving, and
bidding and real-time auctions,” explains Walch.118 Although the goal-driven
systems pattern is that widely utilized as some of the other patterns, it is gaining
rapid adoption.118

124
 THE A.I. CRUISE LINE

Although each one of these patterns can all be utilized individually, organizations
often combine one or more of these seven patterns to realize their goals. 118
Walch believes that, “By companies thinking of AI projects in terms of these
patterns it will help them better approach, plan, and execute AI projects.” 118 “In
fact,” she claims, “emerging methodologies are focusing on the use of these
seven patterns as a way to expedite AI project planning.”118 “Once you know that
you’re doing a recognition pattern, for example, you can gain insight into a wide
range of solutions that have been applied to that problem, insights into the data
that’s needed to power the pattern, use cases and examples of applications of
the pattern, algorithm and model development tips, and other insights that can
help speed up the delivery of high quality AI projects,” Walch concludes. 118

Cruise Line Analytics

In his article Patron Analytics in the Casino and Gaming Industry: How the House
Always Wins119, Scott Sutton lays out the backstory of how analytics is currently
used in the cruise line and gaming industry:
“In the 1980’s and 1990’s, casino patron loyalty programs,
originally called ‘slot clubs’, started popping up in many of the
larger casinos. These slot clubs encouraged customers to sign
up for player cards and, in return for loyalty to the casino,
patrons would receive rewards such as complimentary rooms,
access to special events, and other offers. This was
revolutionary, as it allowed casinos to track gaming behavior
down to the individual level, leading to more accurate
information about patrons’ gaming behavior and interests. The
information could then be used to better segment customers,
predict future behavior, and improve marketing outcomes. As
casino analytics advanced, casino resorts started incorporating
the relevant data from hotel, dining, retail, entertainment, and
other outlets to get a more complete view of a patron’s
behaviors. A recent development is that many of the major
gaming loyalty programs, especially those in competitive
markets such as Las Vegas, are now also rewarding non-
gaming spending in order to encourage customers to keep non-
gaming spending at their respective properties, in addition to
providing additional data about non-gaming behavior.”
This book is an attempt to chart the next course of analytics for cruise lines and
IRs by utilizing IoT, geo-location capabilities, ML, facial and emotional
recognition, AR, as well as many of the other technologies discussed here.
It could be argued that the most valuable use of predictive analytics at a cruise
line would be by the marketing and sales department, but that’s not the whole

125
 ANDREW W. PEARSON

story. Being able to accurately predict not only who are a cruise line’s best leads
and prospects, but when and how it is best to engage them is nice but
understanding how their acceptance of these marketing offers will affect the
overall cruise line’s bottom line is what The A.I. Cruise Line is all about.
As previously mentioned, there are four different types of analytics and they are:
• Descriptive analytics – What happened?
• Diagnostic analytics – Why did it happen?
• Predictive analytics – What will happen?
• Prescriptive analytics – How can we make it happen again?
Figure 15 contains examples of how each of these types of analytics can be
utilized by a cruise line, including at its onboard casino.

Figure 15: Analytics Value Escalator

Source: www.intelligencia.co
For a cruise line, descriptive analytics could include pattern discovery methods
such as customer segmentation, i.e., culling through a customer database to
understand a customer’s preferred type of trip. Simple cluster segmentation
models could also divide customers into their preferred choice of purchases.

126
 THE A.I. CRUISE LINE

Market basket analysis, which utilizes association rules, would also be

considered a descriptive analytics procedure. Cruise lines could use market
basket analysis to bundle and offer promotions as well as gain insight into its
customers’ buying habits. Detailed customer shopping and purchasing behavior
could also be used to develop future products.
Diagnostic analytics is a form of advanced analytics that examines data or
content to answer the question, “Why did it happen?” It attempts to understand
causation and behaviors by utilizing such techniques as drill-down, data
discovery, data mining and correlations. Building a decision tree atop a web
user’s clickstream behavior pattern could be considered a form of diagnostic
analytics as these patterns might reveal why a person clicked his or her way
through a cruise line’s website.
In his seminal article Predictive Analytics White Paper120, Charles Nyce states
that, “Predictive analytics is a broad term describing a variety of statistical and
analytical techniques used to develop models that predict future events or
behaviors. The form of these predictive models varies, depending on the
behavior or event that they are predicting. Most predictive models generate a
score (a patron rating, for example), with a higher score indicating a higher
likelihood of the given behavior or event occurring.”
Data mining, which is used to identify trends, patterns, and/or relationships
within a data set, can then be used to develop a predictive model.120 Prediction
of future events is the key here and these analyses can be used in a multitude of
ways, including forecasting behavior that could lead to a competitive advantage
over rivals. Gut instinct can sometimes punch you in the gut and predictive
analytics can help factor in variables that are inaccessible to the human mind and
often the amount of variables in an analytical problem are beyond human
comprehension.
Predictive analytics is the use of statistics, machine learning, data mining, and
modeling to analyze current and historical facts to make predictions about future
events. Said another way, it gives mere mortals the ability to predict the future
like Nostradamus. In recent years, data mining has become one of the most
valuable tools for extracting and manipulating data and for establishing patterns
in order to produce useful information for decision-making.
Whether you love it or hate it, predictive analytics has already helped elect
presidents, discover new energy sources, score consumer credit, assess health
risks, detect fraud, and target prospective buyers. It is here to stay, and
technological advances, ranging from faster hardware to software that analyzes
increasingly vast quantities of data, are making the use of predictive analytics
more creative and efficient than ever before.
Predictive analytics is an area of data mining that deals with extracting
information from data sets and using it to predict trends and behavioral patterns.

127
 ANDREW W. PEARSON

Often the unknown event of interest is in the future, but predictive analytics can
be applied to any type of unknown, whether that is in the past, the present, or
the future.
Predictive analytics uses many techniques from data mining to analyze current
data to make predictions about the future, including statistics, modeling,
machine learning, and AI. For example, logistic regression can be used to turn a
market basket analysis into a predictor so that a cruise line can understand what
items are usually purchased together.
For a cruise line, predictive analytics can also be used for CRM, collection
analysis, cross-sell, customer retention, direct marketing, fraud detection,
product prediction, project risk management, amongst many other things.
Predictive analytics utilizes the following techniques:
• Regression
• Linear regression
• Discrete choice models
• Logistic regression
• Multinomial logistic regression
• Probit regression
• Time series models
• Survival or duration analysis
• Classification and regression trees
• Multivariate adaptive regression splines
• Machine learning
• Neural networks
• Naïve Bayes
• K-nearest neighbors
Predictive modeling is only useful if it is deployed and it creates an action. Taking
advantage of the more powerful, statistically based segmentation methods,
customers can be segmented not only by dollar values, but also on all known
information, which can include behavioral information gleaned from patron’s
activities, as well as the patron’s simple demographic information. This more
detailed segmentation allows for more targeted and customer-focused
marketing campaigns.
Models can be evaluated and reports generated on multiple statistical measures,
such as neural networks, decision trees, genetic algorithms, the nearest neighbor
method, rule induction, and lift and gains charts. Once built, scores can be
generated in a variety of ways to facilitate quick and easy implementation. The
projects themselves can be reused and shared to facilitate faster model
development and knowledge transfer.
In his paper Predictive Analytics121, Wayne Eckerson advises creating predictive

128
 THE A.I. CRUISE LINE

models by using the following six steps:

1. Define the business objectives and desired outcomes for the project
and then translate them into predictive analytic objectives and tasks.
2. Explore and analyze the source data to determine the most appropriate
data and model building approach and then scope the effort.
3. Prepare the data by selecting, extracting, and transforming the data,
which will be the basis for the models.
4. Build the models, as well as test and validate them.
5. Deploy the models by applying them to the business decisions and
processes.
6. Manage and update the models accordingly.
By utilizing data from past campaigns and measures generated by the predictive
modeling process, cruise line operators can track actual campaign responses
versus expected campaign responses, which can often prove wildly divergent.
Additionally, cruise line operators can generate upper and lower “control” limits
that can be used to automatically alert campaign managers when a campaign is
over or underperforming, letting them focus on campaigns that specifically
require attention.
Prescriptive analytics tries to optimize a key metric, such as profit, by not only
anticipating what will happen, but also when it will happen and why it happens.
In its Prescriptive Analytics Makes Waves with Retail & CPG122, Profitect has one
of the best descriptions of prescriptive analytics, i.e., it is the “application of logic
and mathematics to data to specify a preferred course of action. The most
common examples are optimization methods, such as linear programming;
decision analysis methods, such as influence diagrams; and predictive analytics
working in combination with rules.” Profitect argues that prescriptive analytics
differs from descriptive, diagnostic and predictive analytics in that its output is a
decision.122
Prescriptive analytics can ingest a mixture of structured, unstructured, and semi-
structured data, and utilize business rules that can predict what lies ahead, as
well as advise how to exploit this predicted future without compromising other
priorities. Stream processing can add an entirely new component to prescriptive
analytics.
The analytics powerhouse SAS is finding its vaunted place atop the analytics
pyramid challenged not just by their typical acronymed competitors — SAP, IBM,
EMC, HDS, and the like — but also by the simpler visualization toolmakers like
Tableau, Qlik, and Alteryx, who are muscling their way into the mix, with
products that include data blending and in-memory technology that allows
business users to access complete datasets at the touch of a button. These
solutions offer less complex analytical capabilities, but such things as market
basket analysis or simple decision tree networks can be created with them and

129
 ANDREW W. PEARSON

the costs associated with them can be one quarter or one fifth of what the top
echelon providers charge.
Throughout the rest of this chapter, I will break down many of the different types
of analytical models that can be used to strengthen the customer experience for
cruise lines.
In its conference paper How Predictive Analytics is Changing the Retail
Industry123 from the International Conference on Management and Information
Systems, the writers argue that predictive models incorporate the following
steps:
• Project Definition: Define the business objectives and desired outcomes
for the project and translate them into predictive analytic objectives
and tasks.
• Exploration: Analyze source data to determine the most appropriate
data and model building approach, and scope the effort.
• Data Preparation: Select, extract, and transform data upon which to
create models.
• Model Building: Create, test, and validate models, and evaluate
whether they will meet project metrics and goals.
• Deployment: Apply model results to business decisions or processes.
This ranges from sharing insights with business users to embedding
models into applications to automating decisions and business
processes
• Model Management: Manage models to improve performance (i.e.,
accuracy), control access, promote reuse, standardize toolsets, and
minimize redundant activities.
Even though this was a paper around the subject of retail, it can be utilized for
model building in the cruise line industry as well.

Analytical Methods

Decision Trees
According to Wikipedia, a decision tree is “a decision support tool that uses a
tree-like graph or model of decisions and their possible consequences, including
chance event outcomes, resource costs, and utility. It is one way to display an
algorithm.”124
Lucidchart125 states that, “A decision tree is a map of possible outcomes of a
series of related choices. It allows an individual or organization to weigh possible
actions against one another based on their costs, probabilities, and benefits.
They can be used either to drive informal discussion or to map out an algorithm
that predicts the best choice mathematically.” Lucidchart also adds: “A decision
tree typically starts with a single node, which branches into possible outcomes.

130
 THE A.I. CRUISE LINE

Each of these outcomes leads to additional nodes, which branch off into other
possibilities. This gives it a treelike shape.”125
Decision trees are used to identify the strategy that is most likely to reach a goal.
It is a decision support tool that uses a graph or model of decisions and their
possible consequences, including chance event outcomes, resource costs, and
utility. Decision trees are sequential partitions of a set of data that maximize the
differences of a dependent variable (response or output variable). They offer a
concise way of defining groups that are consistent in their attributes, but which
vary in terms of the dependent variable.
A decision tree consists of three types of nodes:
1. Decision nodes — represented by squares represented by squares,
represented by squares, which shows a decision to be made.125
2. Chance nodes — represented by circles, represented by circles, which
shows the probabilities of the certain results.125
3. End nodes — represented by triangles, represented by triangles, which
reveals the final outcome of a decision path. 125
The construction of a decision tree is based on the principle of “divide and
conquer”: through a supervised learning algorithm, successive divisions of the
multivariable space are carried out in order to maximize the distance between
groups in each division (that is, carry out partitions that discriminate). The
division process finalizes when all of the entries of a branch have the same value
in the output variable, giving rise to the complete model. The further down the
input variables are in the tree, the less important they are in the output
classification (and the less generalization they allow, due to the decrease in the
number of inputs in the descending branches).
For the cruise line industry, decision trees can be utilized in operations
management and marketing, where they can predict whether a person will
respond to an offer or not, or whether they are likely to abuse an offer.119
According to Deng et al. in their paper Building a Big Data Analytics Service
Framework for Mobile Advertising and Marketing126, the decision tree algorithm
is:
“Used to classify the attributes and decide the outcome of the
class attribute. In order to construct a decision tree both class
attribute and item attributes are required. Decision tree is a
tree like structure where the intermediate nodes represent
attributes of the data, leaf nodes represents the outcome of the
data and the branches hold the attribute value. Decision trees
are widely used in the classification process because no domain
knowledge is needed to construct the decision tree.”
The main step in the decision tree algorithm is to identify the root node for any

131
 ANDREW W. PEARSON

given set of data.126 “Multiple methods exist to decide the root node of the
decision tree. Information gain and Gini impurity are the primary methods used
to identify the root node. Root node plays an important role in deciding which
side of the decision tree the data falls into. Like every classification methods,
decision trees are also constructed using the training data and tested with the
test data.”126
Advantages Disadvantages
• Possibility of creating complex
• Simple and robust decision trees for simple data
• Useful to predict the outcomes of • Replication problem makes the
future data decision trees complex. So
• Little cleansing is enough to remove the replicated data
remove the missing values data before constructing a decision
• Useful for large data sets tree
• Decision trees can handle both • Pruning is required to avoid
categorical and numerical data complex decision trees
• It is hard to find out the correct
root node
Table 2: Advantages and disadvantages of decision trees
Source: Researchgate126

k-Means Cluster
As its name suggests, the k-Means cluster is a clustering algorithm and it is one
of the most commonly used analytical models because of its simplicity and ease
of use. The fact that it is still going strong after fifty years speaks as much to its
ease-of-use as it does to the difficulty of designing a general-purpose clustering
algorithm.
According to Telgarsky and Vattani, “The goal of cluster analysis is to partition a
given set of items into clusters such that similar items are assigned to the same
cluster whereas dissimilar ones are not. Perhaps the most popular clustering
formulation is K-means in which the goal is to maximize the expected similarity
between data items and their associated cluster centroids.”127
Hartigan and Wong explain that the: “aim of the k-means algorithm is to divide
M points in N dimensions into k clusters so that the within-cluster sum of squares
is minimized. It is not practical to require that the solution has minimal sum of
squares against all partitions, except when M, N are small and k = 2. We seek
instead ‘local’ optima, solutions that no movement of a point from one cluster
to another will reduce the within-cluster sum of squares.”128
k-means clustering identifies and classifies items into groups based on their
similarity. K is the number of clusters that needs to be decided upon before the

132
 THE A.I. CRUISE LINE

clustering process begins.126 “The whole solution depends on the K value. So, it
is important to choose a correct K value. The data point is grouped into a cluster
based on the Euclidean distance between the point and the centroid of the
cluster,” explains Deng et al.126
For Deng et al, initial clustering can be done in one of three ways.
1. Dynamically Chosen: In this method, the first K items are chosen and
then assigned to K clusters.
2. Randomly Chosen: In this method, the values are randomly selected
and then assigned to K clusters.
3. Choosing from Upper and Lower Boundaries: In this method, the values
that are very distant from each other are chosen and they are used as
initial values for each cluster.”126
According to Deng at al., the k-Means methodology is as follows126:
• Step 1: Choose the initial values using one of the above three methods
• Step 2: For each additional value
• Step 3: Calculate the Euclidean distance between this point and
centroid of the clusters.
• Step 4: Move the value to the nearest cluster.
• Step 5: Calculate the new centroid for the cluster.
• Step 6: Repeat steps 3 to 5.
• Step 7: Calculate centroid of the cluster.
• Step 8: For each value
• Step 9: Calculate the Euclidean distance between this value and the
centroid of all the clusters.
• Step 10: Move the value to the nearest cluster.

133
 ANDREW W. PEARSON

Figure 16: Clustering Algorithm

Source: Researchgate126

Advantages Disadvantages
• Sensitive to noise
• Numbers of clusters must be
• Faster computations than decided before starting clustering
hierarchical clustering • Choosing correct initial clustering
• It produces tighter clusters than process
other clustering techniques • Choosing correct number of
• Gives best result when data sets clusters
are distinct • The centroid of the group changes
• Easy to understand because we calculate centroid
every time a new item joins the
cluster
• Large data sets needed to cluster
the data correctly
Table 3: Advantages and disadvantages of decision trees
Source: Researchgate126

k-Nearest Neighbors
First described in the early 1950s, the k-nearest neighbors method is a

134
 THE A.I. CRUISE LINE

classification (or regression) algorithm that, in order to determine the

classification of a point, combines the classification of the K nearest points. It is
supervised because you are trying to classify a point based on the known
classification of other points. It is labor intensive when given large training sets,
and it did not gain popularity until the computer revolution in the 1960s brought
processing power that was able to handle large data sets.126 Today, it is widely
used in the area of pattern recognition.126
As Deng et al. explain126:
“Nearest-neighbor classifiers are based on learning by analogy,
that is, by comparing a given test tuple with training tuples that
are similar to it. The training tuples are described by n
attributes. Each tuple represents a point in an n-dimensional
space. In this way, all of the training tuples are stored in an n-
dimensional pattern space. When given an unknown tuple, a k-
nearest-neighbor classifier searches the pattern space for the k
training tuples that are closest to the unknown tuple. These k
training tuples are the k ‘nearest neighbors’ of the unknown
tuple. When the ‘k’ closest points are obtained, the unknown
sample is then assigned to the most common class among
those k-points. In case of k=1, the unknown sample is assigned
to the closest point in the pattern space. The closeness is
measured using the distance between the two points.”
the k-means clustering and k-nearest neighbor methodologies seek to
accomplish different goals; k-nearest neighbors is a classification algorithm,
which is a subset of supervised learning, while k-means is a clustering algorithm,
which is a subset of unsupervised learning.
K-nearest neighbor techniques can be used to prevent theft in the retail and
gaming business. Modern surveillance systems are intelligent enough to analyze
and interpret video data on their own, utilizing k-nearest neighbor for visual
pattern recognition to scan and detect hidden packages in the bottom bin of a
shopping cart at check-out, for example.
As she explains in her article Solving Real-World Problems with Nearest Neighbor
Algorithms129, Lillian Pierson states that, “If an object is detected that’s an exact
match for an object listed in the database, then the price of the spotted product
could even automatically be added to the customer’s bill. While this automated
billing practice is not used extensively at this time, the technology has been
developed and is available for use.”
The K-nearest neighbor algorithm can also be used to detect patterns in credit
card usage to root out credit card fraud. “Many new transaction-scrutinizing
software applications use kNN algorithms to analyze register data and spot
unusual patterns that indicate suspicious activity,” Pierson adds.129

135
 ANDREW W. PEARSON

“If register data indicates that a lot of customer information is being entered
manually rather than through automated scanning and swiping, this could
indicate that the employee who’s using that register is in fact stealing customer’s
personal information,” warns Pierson.129 Another example would be “if register
data indicates that a particular good is being returned or exchanged multiple
times, this could indicate that employees are misusing the return policy or trying
to make money from doing fake returns.”129
kNN is not just about fraud. It can also be used to increase retail sales. “Average
nearest neighbor algorithm classification and point pattern detection can be
used in grocery retail to identify key patterns in customer purchasing behavior,
and subsequently increase sales and customer satisfaction by anticipating
customer behavior,” explains Pierson.129
Advantages Disadvantages
• KNN neither doesn’t follow any nor have any
standard for selecting the value ‘k’, which is
• It produces tighter
one of the key factors in the success of an
clusters than other
algorithm
clustering techniques
• As KNN is a Lazy Learner algorithm, it has high
• Gives best result
storage requirements and requires efficient
when data sets are
indexing techniques
distinct
• The efficiency of the KNN algorithm also
• Easy to understand
depends on the choice of the distance metric
used. The results of the algorithm differ for
each similarity metric
Table 4: Advantages and disadvantages of decision trees
Source: Researchgate.126

Logistic Regression
According to Wikipedia, logistic regression is a regression model where the
dependent variable (DV) is categorical, i.e., a variable that can take on one of a
limited, and usually fixed, number of possible values.130 This compares to a
variable that would be continuous. Developed in 1958 by statistician David Cox,
“The binary logistic model is used to estimate the probability of a binary
response based on one or more predictor (or independent) variables (features).
It allows one to say that the presence of a risk factor increases the probability of
a given outcome by a specific percentage.”130
In his article Using Logistic Regression to Predict Customer Retention131, Andrew
Karp explains that:
“Logistic regression is an increasingly popular statistical
technique used to model the probability of discrete (i.e., binary

136
 THE A.I. CRUISE LINE

or multinomial) outcomes. When properly applied, logistic

regression analyses yield very powerful insights in to what
attributes (i.e., variables) are more or less likely to predict
event outcomes in a population of interest. These models also
show the extent to which changes in the values of the
attributes may increase or decrease the predicted probability
of event outcomes.”
Logistic regression techniques may be used to classify a new observation whose
group is unknown, in one of the groups, based on the values of the predictor
variables. According to Karp, “Logistic regression models are frequently
employed to assess the chance that a customer will: a) re-purchase a product, b)
remain a customer, or c) respond to a direct mail or other marketing stimulus.” 131
Karp adds that “Economists frequently call logistic regression a ‘qualitative
choice’ model, and for obvious reasons: a logistic regression model helps us
assess probability which ‘qualities’ or ‘outcomes’ will be chosen (selected) by the
population under analysis.”131 As can be expected, Karp argues that, “When
proper care is taken to create an appropriate dependent variable, logistic
regression is often a superior (both substantively and statistically) alternative to
other tools available to model event outcomes.” 131
Karp uses a health care example to make his point that the analyst has several
independent variables to use in the modeling process, but this example can be
illustrative of how they could be used in the cruise line industry.131 Karp explains
that “An analyst developing a model predicting re-enrollment in a health
insurance plan may have data for each member’s interaction with both the
health plans administrative apparatus and health care utilization in the prior
‘plan year.’”131
The analyst can then construct variables such as the “number of times member
called the health plan for information, number of physician office visits, whether
or not the member changed primary care physicians during the previous ‘plan
year,’ and answers to a customer satisfaction survey.”131 These can be employed
in the modeling process and, once the model has been constructed, the analyst
must decide which variable can be employed as the “outcome” or the
“dependent” variable.131
In logistic regression analyses “it is often the analyst’s responsibility to construct
the dependent variable based on an agreed-upon definition of what constitutes
the ‘event of interest’ which is being modeled.”131
In the health care re-enrollment example, “a health plan’s management team
may define ‘attrition’ or ‘failure to re-enroll’ as situations where a member fails
to return the re-enrollment card within 30 days of its due date. Or, in a response
modeling scenario, a direct mail firm may define ‘non-response’ to an
advertisement as failure to respond within 45 days of mailout. ”131

137
 ANDREW W. PEARSON

Logistic regression models can be powerful tools to build models that

understand customer retention.131 “When applied properly, logistic regression
models can yield powerful insights into why some customers leave and others
stay. These insights can then be employed to modify organizational strategies
and/or assess the impact of the implementation of these strategies,” Karp
adds.131

A/B Testing
Also known as split testing or bucket testing, A/B testing is a method of
marketing testing by which a baseline control sample is compared to a variety of
single-variable test samples in order to improve response rates.
A classic direct mail tactic, this method has recently been adopted within the
interactive space to test tactics such as banner ads, emails, and landing pages.
As Scott Sutton explains in his article Patron Analytics in the casino and
Hospitality Industry: How the House Always Wins119, for casino marketers, A/B
testing is the most effective way to identify the best available marketing offer.119
It can test “two different offers against one another in order to identify the offer
that drives the highest response and the most revenue/profit.”119
As Dan Siroker and Peter Komen explain in their book A/B Testing: The Most
Powerful Way to Turn Clicks Into Customers132, “The hardest part of A/B testing
is determining what to test in the first place. Having worked with thousands of
customers who do A/B testing every day, one of the most common questions we
hear is, ‘Where do I begin?’”
The mistake many companies make is they jump in headfirst without any
detailed planning. Siroker and Komen propose the following deliberate five-step
process132:
1. Define success
2. Identify bottlenecks
3. Construct a hypothesis
4. Prioritize
5. Test
A/B testing is particularly good for website marketing, especially for uncovering
a company’s best landing page. As Siroker and Komen explain, “Defining success
in the context of A/B testing involves taking the answer to the question of your
site’s ultimate purpose and turning it into something more precise: quantifiable
success metrics. Your success metrics are the specific numbers you hope will be
improved by your tests.”132
Whereas a cruise line website could easily define its success metrics in terms of
revenue per visitor, it is also important to understand such things as traffic
sources, bounce rate, top pages, conversion rates, conversion by traffic source,

138
 THE A.I. CRUISE LINE

amongst other things.

Site Type Common Conversion & Aggregate Goals
e-Commerce • Completed purchase
A site that sells things • Each step within the checkout funnel
for users to purchase • Products added to cart
online. • Product page views

Media/Content • Page views

A site focused on • Articles read
article or other • Bounce rate (when measuring within an A/B
content consumption. testing tool, this is often measured by seeing if
the user clicked anywhere on the page)
Lead Generation • Form completion
A site that acquires • Clicks to a form page (links may read “Contact
business through us” for example)
name capture.
Donation • Form completion
• Clicks to a form page (links may read “Send a
donation” for example)

Table 5: Typical A/B conversion & aggregate goals

Source: A/B Testing: The Most Powerful Way to Turn Clicks Into Customers 132

As Siroker and Komen state:

“Part of building out your testing strategy is identifying what
constitutes — and does not constitute — a “conversion” for
your particular site. In online terms, a conversion is the point at
which a visitor takes the desired action on your website.
Pinpointing the specific actions you want people to take most
on your site and that are most critical to your business will lead
you to the tests that have an impact.” 132
Once the site’s quantifiable success metrics are agreed upon, attention can be
paid trying to discover where the bottlenecks are.132 These are the places where
users are dropping off, or the places where momentum in moving users through
the desired series of actions weakens.132

Time Series Model

A time series is an ordered sequence of values of a variable at uniformly spaced
time intervals. A Time Series model can be used to predict or forecast the future
behavior of a variable.

139
 ANDREW W. PEARSON

In his article Time Series Analysis, Muhammad Imdadullah explains that “Time
series analysis is the analysis of a series of data-points over time, allowing one to
answer question such as what is the causal effect on a variable Y of a change in
variable X over time? An important difference between time series and cross
section data is that the ordering of cases does matter in time series.”133
These models account for the fact that data points taken over time may have an
internal structure (such as autocorrelation, trend or seasonal variation) that
should be taken into account. For the cruise line industry, a Time Series Analysis
can be used to forecast sales, project yields and workloads, as well as analyze
budgets.
Time series can be broken down into two variations:
• Continuous Time Series — “A time series is said to be continuous when
observation are made continuously in time. The term continuous is
used for series of this type even when the measured variable can only
take a discrete set of values.”133
• Discrete Time Series — “A time series is said to be discrete when
observations are taken at specific times, usually equally spaced. The
term discrete is used for series of this type even when the measured
variable is a continuous variable.”133
As Sang and Dong explain in their Determining Revenue-Generating Casino
Visitors Using a Vector Autoregressive Model: The Case of the G Casino in
Korea134, time-series data was analyzed to:
“investigate the characteristics of cruise line visitors that affect
cruise lines’ revenue generation. Exchange rates — a
traditional measure relevant to tourism — and customer types
and nationalities were empirically analyzed with a vector
autoregressive model using data acquired from all branches of
Korea’s G cruise line. The results suggest that the cruise lines’
revenues were affected by the customers’ type and nationality:
VIP customers were very important factors in the cruise lines’
revenue generation; moreover, the revenue impact of Russian
visitors was quite strong despite their small numbers.”
Time series can be used to compare seasonal estimation and trend estimation in
forecasting models on both a state or a national level.

Neural Networks
Artificial Neural Networks (ANN) or just “Neural Networks” are non-linear
statistical data modeling tools that are used when the exact nature of a
relationship between input and output is unknown. In their article Neural
Networks in Data Mining135, Singh and Chauhan claim that a neural network is:

140
 THE A.I. CRUISE LINE

“A mathematical model or computational model based on

biological neural networks, in other words, is an emulation of
biological neural system. It consists of an interconnected group
of artificial neurons and processes information using a
connectionist approach to computation. In most cases an ANN
is an adaptive system that changes its structure based on
external or internal information that flows through the
network during the learning phase.”
As Jim Gao explains in his article Machine Learning Applications for Data Center
Optimization136:
“Neural networks are a class of machine learning algorithms
that mimic cognitive behavior via interactions between
artificial neurons. They are advantageous for modeling
intricate systems because neural networks do not require the
user to predefine the feature interactions in the model, which
assumes relationships within the data. Instead, the neural
network searches for patterns and interactions between
features to automatically generate a best fit model. Common
applications for this branch of machine learning include speech
recognition, image processing, and autonomous software
agents. As with most learning systems, the model accuracy
improves over time as new training data is acquired.”
Neural networks can be used to find patterns in data. A key feature of neural
networks is that they learn the relationship between inputs and output through
training.
There are three types of training in neural networks; reinforcement learning,
supervised and unsupervised training, with supervised being the most common
one. Neural Networks are data processing systems whose structure and
functioning are inspired by biological neural networks. Their fundamental
characteristics include parallel processing, distributed memory, and adaptability
to their surroundings.
For a cruise line’s marketing purposes, neural networks can be used to classify a
consumer's spending pattern, analyze a new product, identify a patron's
characteristics as well as forecast sales.135 The advantages of neural networks
include high accuracy, high noise tolerance and ease of use as they can be
updated with fresh data, which makes them useful for dynamic environments. 135
In her article How DeepMind’s AlphaGo Zero Learned all by itself to trash world
champ AI AlphaGo137, Katyanna Quach explains how neural networks can work
when training computers to play board games. According to Quach, the board
game Go is considered a “difficult game for computers to master because,
besides being complex, the number of possible moves — more than chess at

141
 ANDREW W. PEARSON

10170 — is greater than the number of atoms in the universe.” 137

“AlphaGo, the predecessor to AlphaGo Zero, crushed 18-time world champion
Lee Sedol and the reigning world number one player, Ke Jie,” explains Quach.137
The next generation of DeepMind’s technology, AlphaGo Zero, beat “AlphaGo
100-0 after training for just a fraction of the time AlphaGo needed, and it didn't
learn from observing humans playing against each other — unlike AlphaGo.
Instead, Zero's neural network relies on an old technique in reinforcement
learning: self-play.”137
As Quach notes about the process137:
“Essentially, AlphaGo Zero plays against itself. During training,
it sits on each side of the table: two instances of the same
software face off against each other. A match starts with the
game's black and white stones scattered on the board, placed
following a random set of moves from their starting positions.
The two computer players are given the list of moves that led
to the positions of the stones on the grid, and then are each
told to come up with multiple chains of next moves along with
estimates of the probability they will win by following through
each chain.
“So, the black player could come up with four chains of next
moves, and predict the third chain will be the most successful.
The white player could come up with its own chains, and think
its first choice is the strongest.
“The next move from the best possible chain is then played, and
the computer players repeat the above steps, coming up with
chains of moves ranked by strength. This repeats over and over,
with the software feeling its way through the game and
internalizing which strategies turn out to be the strongest.”
This methodology differs from the old AlphaGo, which “relied on a
computationally intensive Monte Carlo tree search to play through Go
scenarios.”137 “The nodes and branches created a much larger tree than AlphaGo
practically needed to play.”137
“A combination of reinforcement learning and human-supervised learning was
used to build ‘value’ and ‘policy’ neural networks that used the search tree to
execute gameplay strategies,” explains Quach.137 “The software learned from 30
million moves played in human-on-human games, and benefited from various
bodges and tricks to learn to win. For instance, it was trained from master-level
human players, rather than picking it up from scratch,” adds Quach. 137
“Self-play is an established technique in reinforcement learning, and has been
used to teach machines to play backgammon, chess, poker, and Scrabble,” says

142
 THE A.I. CRUISE LINE

Quach.137 David Silver, a lead researcher on AlphaGo, explains that it is an

effective technique because the opponent is always the right level of difficulty. 137
“So it starts off extremely naive," Silver said, adding that “at every step of the
learning process it has an opponent — a sparring partner if you like — that is
exactly calibrated to its current level of performance. To begin with these players
are very weak but over time they get progressively stronger.”137
Tim Salimans, a research scientist at OpenAI, explains that self-play means
“agents can learn behaviours that are not hand coded on any reinforcement
learning task, but the sophistication of the learned behavior is limited by the
sophistication of the environment. In order for an agent to learn intelligent
behavior in a particular environment, the environment has to be challenging, but
not too challenging.”137
“The competitive element makes the agent explicitly search for its own
weaknesses. Once those weaknesses are found the agent can improve them. In
self-play the difficulty of the task the agent is solving is always reasonable, but
over time it is open ended: since the opponent can always improve, the task can
always get harder," adds Salimans.
Self-play does have its limitatiosn.137 Right now, there are “problems that
AlphaGo Zero cannot solve, such as games with hidden states or imperfect
information, such as StarCraft, and it’s unlikely that self-play will be successful
tackling more advanced challenges.”137
Self-play will be worthwhile in some areas of AI, argues Salimans. 137 “As our
algorithms for reinforcement learning become more powerful the bottleneck in
developing artificial intelligence will gradually shift to developing sufficiently
sophisticated tasks and environments. Even very talented people will not
develop a great intellect if they are not exposed to the right environment,” he
warns.137
DeepMind, the company behind AlphaGo Zero and its predecessor, believes that
“the approach may be generalizable to a wider set of scenarios that share similar
properties to a game like Go.”137

Discriminant Analysis
According to Wikipedia, “Discriminant function analysis is a statistical analysis
used to predict a categorical dependent variable (called a grouping variable) by
one or more continuous or binary independent variables (called predictor
variables). The original dichotomous discriminant analysis was developed by Sir
Ronald Fisher in 1936. It differs from an ANOVA or MANOVA, which is used to
predict one (ANOVA) or multiple (MANOVA) continuous dependent variables by
one or more independent categorical variables.”138
Discriminant or discriminant function analysis is a method used to determine

143
 ANDREW W. PEARSON

which weightings of quantitative variables or predictors best discriminate

between two or more than two groups of cases and do so better than chance. It
is a method used in statistics, pattern recognition and machine learning to find
a linear combination of features that characterizes or separates two or more
classes of objects or events.
Because of its ability to classify individuals or experimental units into two or
more uniquely defined populations, discriminate analysis can be used for market
segmentation and the prediction of group membership. The discriminant score
can be the basis on which a prediction about group membership is made. For
example, the discriminant weights of each predictive variable (age, sex, income,
etc.) indicate the relative importance of each variable. In other words, if age has
a low discriminant weight then it is less important than the other variables.
For a cruise line’s marketing department, use of discriminant analysis can help
predict why a patron frequents one cruise over another. Discriminant analysis is
specifically useful in product research, perception/image research, advertising
research and direct marketing.

Survival or Duration Analysis

As per Wikipedia, “Survival analysis is a branch of statistics for analyzing the
expected duration of time until one or more events happen, such as death in
biological organisms and failure in mechanical systems. This topic is called
reliability theory or reliability analysis in engineering, duration analysis or
duration modeling in economics, and event history analysis in sociology.”139
Survival analysis attempts to answer questions such as 139:
• What is the proportion of a population which will survive past a certain
time?
• Of those that survive, at what rate will they die or fail?
• Can multiple causes of death or failure be taken into account?
• How do particular circumstances or characteristics increase or decrease
the probability of survival?
A branch of statistics that deals with death in biological organisms and failure in
mechanical systems, survival analysis involves the modeling of time to event
data; in this context, death or failure is considered an “event” in the survival
analysis literature – traditionally only a single event occurs, after which the
organism or mechanism is dead or broken. Survival analysis is the study of
lifetimes and their distributions. It usually involves one or more of the following
objectives:
1. To explore the behavior of the distribution of a lifetime.
2. To model the distribution of a lifetime.
3. To test for differences between the distributions of two or more
lifetimes.

144
 THE A.I. CRUISE LINE

4. To model the impact of one or more explanatory variables on a lifetime

distribution.

There are several other data mining techniques that can be used but the ones
listed above are the industry’s most common ones and much of what you will
need to glean from your data can be discovered by using them. Once the data
has been mined, a business intelligence solution can help visualize what's going
on with your data, while a predictive analytics program can actually analyze
current and historical trends to make predictions about future events.

Cruise line Analytical Models

The following models have been implemented in the cruise line industry, but this
should not be considered an authoritative list, as creative cruise line executives
should be able to easily come up with a lot more.

Customer Segmentation
A customer segmentation model provides a view of the cruise line from a
customer perspective, such models have many and varied applications. In this
example, customers are segmented according to what they present to the cruise
line. Views include:
• Game preference: the games offered by a cruise line are grouped
according to business needs and customer turnover by game is analyzed
to derive a preferential game for every customer above a certain
threshold turnover.
• Day of week: customer turnover is analyzed by day of the week and
clusters are derived in line with how the cruise line is visited. It is often
the case that customers group into single day segments, as well as some
longer segments, such as weekend players, midweek players, etc., etc.
With one cruise line, we found this an important metric because they
discovered that their high-rollers tended to frequent the cruise line
during the week, rather than on the weekend. Our recommendation
was to open their expensive steak restaurant during what they had
initially considered their quiet days so that their high-rollers could
spend money in the cruise line property’s most expensive restaurant.
• Time of day: split the 24 hours of the day into meaningful bins and
analyze customer turnover by time period from when a gambler’s
session began. This leads to a view of the customer according to when
they are most likely to frequent the cruise line, e.g. matinee players,
night owls, etc., etc. This can help with table games revenue
management, as well as predictive and optimizing labor needs.

145
 ANDREW W. PEARSON

• Length of session: this view clusters customers according to how long

their play session is likely to be. While this isn’t usually as pure a view
as the above models, it can give an idea of impulse players, session
players, etc., etc.
• Size of stake: the size of a customer’s stake can be an important view
for planning. Depending on how granular the venue information is,
individual stakes or the table limit or credits played per hand on a slot
machine are used to segment customers according to their stakes.
The results of this analysis presents a detailed view of how the cruise line is
populated at different times and can allow for appropriate strategic decisions to
be made. These decisions could be a function of marketing, operations or
strategy. The output is also used for the building of acquisition models as
discussed below.
Other potential for analysis would be a master segmentation model that uses
the preference results described. Customers are clustered based on their
preferences to gain a global view of the cruise line that is concise and
understandable.

Customer Acquisition Model

Just like every other business, cruise line operators are always looking for new
customers. With the cruise line market getting more and more competitive and
saturated by the day, there is always a constant need to know where to attract
customers from and what type of customer to target.
The results of the segmentation modeling previously described can be used to
build a predictive model that identifies likely characteristics of attractive
customers. Obviously, the cruise line will have no internal data available on
customers they don’t already have on their books, so the analysis becomes a
data mining exercise using publicly available input variables. Cruise lines can then
target these customers with a view to attracting those who have the traits that
they see in their already valuable customers.
The best external data to use would be population census data, linked to the
internal customers by a location identifier (such as postcode or mesh block). It is
acknowledged that in some jurisdictions robust and accurate census data may
not be available so the model would be relying on whatever information the
cruise line records on its customers from a demographic and lifestyle point of
view.
This approach becomes a classical data-mining problem, where a pool of
independent variables would be tested for the strength of association with the
response variable. Once the relevant predictors are identified and the
characteristics and traits are defined, marketing and acquisition campaigns could
be targeted at the population towards these kinds of people.

146
 THE A.I. CRUISE LINE

This would be something that looks to predict a metric derived from current/past
customers. Such a metric could come from a segmentation model that identified
the high value customers that are most attractive to the cruise line. There are
several approaches that can be used and once the target has been defined, this
allows for a parametric equation to be derived. This equation attempts to predict
the characteristics that distinguish the desirable customers from the rest.
This model can only use publicly available information (although other cruise line
information might be acceptable) as that is how a potential customer would be
identified. Current information that the company would have on hand would be
age, nationality, gender, and address.
Where available, third party data should be looked at to further enhance the
findings. This could be census data that gives an indication of further customer
demographics and this enhances the ability to home in on customer sweet spots.

Recency-Frequency-Monetary (RFM) Models

RFM is a method used for analyzing customer value. It is commonly used in
database marketing and direct marketing and has received particular attention
in the cruise line and retail industries. RFM stands for:
• Recency — How recently did the customer purchase?
• Frequency — How often do they purchase?
• Monetary Value — How much do they spend?
Most businesses will keep scores of data about a customer’s purchases. All that
is needed is a table with the customer name, date of purchase and purchase
value. One methodology is to assign a scale of 1 to 10, whereby 10 is the
maximum value and to stipulate a formula by which the data suits the scale. For
example, in a service-based business like a cruise line’s, you could have the
following:
• Recency = 10 — the number of months that have passed since the
customer last purchased.
• Frequency = number of purchases in the last 12 months (maximum of
10).
• Monetary = value of the highest order from a given customer
(benchmarked against $10k).
Alternatively, one can create categories for each attribute. For instance, the
‘Recency’ attribute might be broken into three categories: customers with
purchases within the last 90 days; purchases between 91 and 365 days; and
purchases longer than 365 days. Such categories may be arrived at by applying
business rules or using a data mining technique to find meaningful breaks.
Once each of the attributes has appropriate categories defined, segments are
created from the intersection of the values. If there were three categories for

147
 ANDREW W. PEARSON

each attribute, then the resulting matrix would have twenty-seven possible
combinations (one well-known commercial approach uses five bins per
attribute, which yields 125 segments).
Segments could also be collapsed into sub-segments if the gradations appear too
small to be useful. The resulting segments can be ordered from most valuable
(highest recency, frequency, and value) to least valuable (lowest recency,
frequency, and value). Identifying the most valuable RFM segments can
capitalize on chance relationships in the data used for this analysis. For this
reason, it is highly recommended that another set of data be used to validate
the results of the RFM segmentation process.
Advocates of this technique point out that it has the virtue of simplicity: no
specialized statistical software is required, and the results are readily understood
by businesspeople. In the absence of other targeting techniques, it can provide
a lift in response rates for promotions.
Whichever approach is adopted, profiling will be done on the results to
determine what makes up group membership. Categorical factors such as
gender, nationality/locality can be used as well as age (or, indeed, any other
demographic feature that is available) to understand the “type” of customer that
resides in each group. These factors can be used for each segment and applied
against the population metrics to determine how much more or less likely a
segment is to exhibit a feature or type of behavior when compared to the
customer base as a whole.

Propensity to Respond Model

A Propensity to Respond model is the theoretical probability that a sampled
person (or unit) will become a respondent in an offer or survey. These models
are particularly useful in the marketing field.
A response likelihood model can have substantial cost savings as it can lead to
lower mailing costs by identifying patrons who are very unlikely to respond to a
particular offer. After segmenting these people out, the cruise line can then
focus on only those most likely to take up the offer. A cruise line can identify the
likelihood of response from all eligible patrons.119 After that, it can identify the
most valuable patrons that are most likely to respond. 119 This allows the cruise
line to estimate the expected response from the most valuable patrons and
eliminate mailing(s) to the patrons that are of lower worth and/or are unlikely
to respond.119
Sutton warns that, “Occasionally, response likelihood models will lead to easy
decisions, such as cutting out low worth patrons with a low likelihood of
responding. However, more complex situations might arise since response
models are never perfect.”119 It doesn’t matter how good a model is or how
accurate the historical data is, there is always a chance that a patron identified

148
 THE A.I. CRUISE LINE

as unlikely to respond will respond.119 “Thus, when making a decision about

patrons identified as unlikely to respond to an offer, it is also important to
balance that likelihood of response with the potential return on response,”
advises Sutton.119
A propensity to respond model would be built using historical information
around marketing campaigns and it looks at predicting the likelihood a customer
will respond to a marketing communication. The advantage of this model is that
it strengthens the marketing strategy even more, beyond purely segmenting the
customer base. It can further allow for improved ROI on the marketing budget,
by identifying the likely number of respondents to be returned by a campaign.
Often a business’ marketing department will have an expected number of
respondents or an expected response rate.119 By identifying those who are most
likely to respond, the chances of meeting that expected number or rate of
response is greatly improved.119 Gone are the days of marketing to an entire
customer base. This is an unnecessary waste of the marketing budget and also
runs the risk of annoying customers by touching them too often or with the
wrong offer.119
Again, a predictive model would be built which identifies those most likely to
respond through to those least likely to respond. This would be done using
customer metrics and historical campaign/marketing information that identifies
those who responded and those who didn’t. Variables that have a significant
association with the customer action are extracted and these form part of the
prediction algorithm. Every customer is then given a score according to how
likely they are to respond to a marketing campaign.
This information can be used for strategies such as extracting the top 40% of
customers most likely to respond, or a fixed number, such as 100,000 customers.
The end result is the marketing function becomes more efficient and effective,
with better returns for the company’s marketing dollar.

Customer Conversion Model

This model would be used to score customers based on information contained
in a cruise line’s source systems as it would only be applicable for customers who
had pre-booked their cabin (as opposed to walk-in customers). Historical
information would be extracted from the cruise line’s IT systems around
desirable customers. This would include spending patterns and profitability.
To identify the relationships that may exist between how the customer comes to
the cruise line and his or her desirability metric, information would be extracted
from the cruise line’s source systems. For a cruise line, this would include
information such as source of betting, channel of betting, lead-time for betting
and the incentives offered to attract the customer. Basically, anything that can
be attributed to the initial transaction the customer has with the cruise line

149
 ANDREW W. PEARSON

would be used as a potential input.

These models might also have to be stratified by itinerary to identify the most
relevant relationships. The major advantage of a predictive model with this
intention would be that it allows the cruise line to identify customers that they
need to interact with once they step onto the ship. This would give the cruise
line hosts the potential to get the required information they need to successfully
foster a strong customer relationship.
Furthermore, if every potential customer has a score associated with him or her
as to his or her long-term likelihood of being attractive, the cruise line can further
hone in on its customers by monitoring their behavior once they are on the ship.
If customers are made to feel like they are valuable and worthwhile, the
likelihood of them returning under their own volition significantly increases.

Identify When a Patron is Likely to Return

Besides knowing which offers a patron is most likely to respond to, it would also
be good to know exactly when a patron was planning to make his or her next
trip.119 Although it might not be possible to know exactly when a patron plans to
return, the cruise line’s marketing department might be able to make an
accurate prediction around it.119
“There are a variety of methods that range in complexity that can be used to
assess when a patron will return to bet, including frequency analysis, regression,
and survival analysis. Knowing when a patron is likely to bet is beneficial as it
helps to identify patrons that haven’t made a trip in the expected amount of time
and are at risk of leaving,” advises Sutton.119 “First, the business needs to have
an idea of the average or median time between trips. This might need to be
segmented based on geography, worth, or even historical frequency,”
recommends Sutton.119 Patrons who haven’t made a bet within the set amount
of time for his or her segment will be flagged and dealt with appropriately,
perhaps marketed to more aggressively, perhaps given marketing content
referencing “We haven’t seen you in a while”, or “Last chance type of offers.” 119
This can also be done for trips, not just bets.

Patron Worth Model

As Sutton explains, “Most industry experts would agree that determining a
patron’s worth is the first and foremost responsibility of patron analytics in the
cruise line industry.”119 Of course, predicting a patron’s future behavior is not
easy and it is affected by a number of variables, “many of which are outside
factors that the business might not have insight into, including total income,
expendable income, ethnicity, reasons for a trip to the cruise line, etc., etc.”119
Even where a patron lives, or information gleaned from his or her social media
accounts could be very revealing. There is also “plenty of information to be found

150
 THE A.I. CRUISE LINE

with in-house data that can be used to build models and metrics to predict a
patron’s future worth.”119 Once patron worth has been determined, “patrons
can then be segmented into groups based on other behaviors and effective
marketing campaigns can be developed around those behaviors.”119
The first thing to do is to “determine what worth is, as the definition of worth is
critical for deciding how valuable a patron is and how much to reinvest in the
patron in the future.”119 As Sutton explains, “There are two main components of
worth — the financial sources of worth (i.e., gambling) and the unit of time to
which it refers (daily, weekly, monthly, etc.). Additionally, worth can refer to
historical worth, which is already known, or future worth, which is unknown.”119
“The definition of worth will likely depend on both the various financial sources
of revenue that affect the business directly and the exact business problems that
are being addressed. Gambling worth can also be broken down into various
sources (i.e., what types of betting does the patron like to do) depending on the
business issues being addressed,” Sutton explains.119
In the gambling industry, there are “two important measures used to assess a
patron’s gaming worth — actual and theoretical loss. Actual loss is how much
money the patron actually lost (or won), whereas theoretical loss usually refers
to the amount of money a patron is expected to lose based on the amount of
money wagered, the time spent playing, and the probability associated with type
of games played.”119 Theoretical loss or Theo loss “tends to be more heavily
relied upon for predictive analysis and is a much stronger predictor of future
behavior, as actual loss is usually used to measure campaign performance and
profitability.”119 Actual win could be heavily affected by a lucky winning streak,
while Theo win would not.
“Once patron worth has been defined, the business can then use data mining
and modeling to estimate predicted worth into the future,” states Sutton.119
“There are a variety of techniques that are used to develop models to predict
future worth, the most common being regression models. Multiple regression
models are the most common because they utilize a variety of predictors and
the relationships between those predictors to predict future worth,” adds
Sutton.119
“Regression models can also be built using such categorical variables predictors
as gender, ethnicity, age range, or other demographic variables. Developing
separate models based on categorical variables, such as separate models
predicting worth for slot and table players, might produce models with less error
and better predictions.”119 “Regression models are particularly effective because
the model can be used to score historical data to predict an unknown outcome,
which is worth in this case, within a certain degree of confidence,” adds
Sutton.119

151
 ANDREW W. PEARSON

Customer Churn Model

In his article Smart Strategies Require Smarter KPIs 140, Michael Schrage argues
that “reducing churn―the KPI that tracks customers ending their relationship
with a company over a particular time period―is a strategic priority. Even
determining that period―a week, month, quarter, or year―is itself a strategic
choice. Almost without exception, significant changes in churn rates command
immediate top management attention.”140
According to the Harvard Business Review141, the cost of acquiring new
customers can prove to be five to 25 times more expensive than keeping existing
ones. Schrage believes that, “Customer retention is key to sustaining cash flow
and profitability.”140 In today’s big data and AI environments, Schrage believes
that, “understanding ex post facto churn no longer strategically suffices;
organizations seek to predict churn to proactively prevent it.” 140 “Making churn
a more anticipatory and prescriptive KPI requires a virtuous cycle approach. In
short, ‘learning from churning’ makes the KPI smart,” argues Schrage. 140
For Schrage, data governance is key.140 He adds that, “Distinctions must be
drawn between churn presumed (a customer who simply stops engaging, that is,
no more visits, purchases, etc.) versus churn absolute (the customer who
explicitly closes an account or discontinues a service).” 140 “Similarly, differences
between reactive and prospective churn must be understood. Customers can
leave after specific bad experiences, such as poor service or unexpected charges.
Alternately, rival options might appear, or the service becomes less compelling
for other reasons. This prospective or silent churn is typically more difficult to
identify or predict,” says Schrage.140 Because it is so difficult to identify or
predict, it will probably be recognized by any patron who the cruise line reaches
out to proactively stop him or her from churning.
Schrage argues that140:
“organizations seeking to devise a smarter churn KPI need to
correlate negative customer experiences with propensity to
churn. They want to be able to capture and chart the gradual
disengagement behaviors that reliably lead to disconnection.
They’d likely analytically invest in identifying those clusters and
segments at highest risk for departure. That requires that these
companies know what data sets would make the best
resources for scoring or ranking ‘likeliest to churn’ customers
to prioritize preemptive and preventive action.”
For Schrage, even these analytics don’t go far enough, and he uses the example
of one global telecom: the company “developed and explored several churn
prediction models. It quickly discovered that its analytics lacked any meaningful
assessment of customer lifetime value (CLV)―that is, the long-term revenue and
profit potential of the customer.”140 “Potentially high-value customers were not

152
 THE A.I. CRUISE LINE

measurably differentiated from customers who reliably switched providers in

pursuit of the lowest possible price (or, almost as challenging, customers whose
constant complaining and refund requests made them money losers),” explains
Schrage.140 “Aligning the CLV KPI with the predictive churn KPI completely
transformed how the company assigned resources and designed interventions
for customer retention,” says Schrage.140
Arguably, customer retention is both one of the cornerstones of any CRM
system, as well as being the most important component of the customer lifetime
value (CLV) framework.142
In their article In pursuit of enhanced customer retention management 143,
Ascarza, Neslin, Netzer, Lemmens, & Aurélie highlight “the importance of
distinguishing between which customers are at risk and which should be
targeted — as they aren’t necessarily the same customers.”
There are indications that companies have problems managing customer
retention. According to Handley144, from the point of view of the customer, 85%
report that companies could do more to retain them. A Forbes Insight145 study
found that from the company’s point of view, a vast majority of top executives
report that customer retention is a priority for their organization, only 49%
believe their company has the ability to support their retention goals.
Ascarza et al. suggest that “an inordinate amount of effort has been devoted to
predicting customer churn”143, but less attention has been afforded to “elements
of campaign design such as whom to target, when to target, and with what
incentives, as well as the broader issues of managing multiple campaigns and
integrating retention programs with the firm’s marketing activities and
strategies.”143
Ascarza et al. propose the following definition: “First, the central idea that
customer retention is continuity — the customer continues to interact with the
firm. Second, that customer retention is a form of customer behavior — a
behavior that firms intend to manage. Accordingly, we propose that ‘Customer
retention is the customer continuing to transact with the firm.’”143
Cruise lines are unique from many other industries in that their customers are
not tied into contracts, but “many of the retention metrics relevant for
contractual firms are also relevant for non-contractual firms” state Ascarza et
al.143 A simple 0/1 indicator of transaction, and a measure of recency are
appropriate for both types of companies.143 However, on its own recency is not
a good indicator of customer retention, argue Ascarza et al. 143 They provide the
following example143:
Customers A and B last purchased 6 months ago. On the one
hand, Customer A typically purchases once a year (i.e., her
inter-purchase time is 12 months), thus a recency of 6 months

153
 ANDREW W. PEARSON

should not be taken as an indication of churn because it is well

within the customer’s purchase cycle. On the other hand,
Customer B usually purchases every month, in which case a
recency of 6 months should be worrisome for the firm. We thus
recommend calculating a recency/inter-purchase ratio, where
a ratio larger than one is an indication of a retention problem.
Regarding the examples above, Customer A’s recency/inter-
purchase times is 6/12 = 0.5, whereas Customer B’s ratio is 6/1
= 6.
Ascarza et al. warn there are two caveats that should be kept in mind when
creating the recency/inter-purchase-time ratio143; “First, it requires observing a
reasonable number of transaction to reliability calculate the average inter-
purchase time. Second, even for those customers for whom one observes a
sufficient number of transactions, the inter-purchase time measure might be
biased due to the right censoring nature of the data — the time between the last
purchase and the last observation is ignored.” 143
Ascarza et al. propose the following process to develop and evaluate a single
retention campaign143:
• Identify customers who are at risk of not being retained.
• Diagnose why each customer is at risk.
• Decide when to target these customers and with what incentive and/or
action.
• Implement the campaign and evaluate it.
For Ascarza et al., “these steps are applicable to both proactive and reactive
campaigns.”143 “Reactive campaigns are simpler because the firm doesn’t need
to identify who is at risk — the customer who calls to cancel self-identifies.
‘Rescue rates’ can readily be calculated to evaluate the program, and subsequent
behavior can be monitored.”143 The incentive should be substantial because the
company is pretty certain the customer will churn.146 Reactive campaigns,
however, can be challenging because not all customers can be rescued, and,
because we’re dealing with human nature here, customers learn that informing
the firm about their intentions to churn can be richly rewarded with valuable
incentives, which can endanger the long-run sustainability of reactive churn
management.147
“Proactive campaigns are more challenging starting from the basic task of
identifying who is at risk,” argue Ascarza et al.143 Balancing the cost of false
positives (targeting a customer who has no intention to leave) against false
negatives (failing to identify a customer who is truly at risk) requires
sophisticated analytics.148
To discover who is at risk, a predictive model must be built that identifies

154
 THE A.I. CRUISE LINE

customers at risk of not being retained, or in general of generating lower

retention metrics.143 “The dependent variable could be 0/1 churn or any
measure of retention.”143

Factors Example Method

1. Emotion in emails 1. Logistic, SVM, Random Forests
2. Customer service calls 2. SVM + ALBA
Customer
Satisfaction 3. Usage trends 3. Logistic, NN, SVM, Genetic
4. Complaints 4. Logistic, NN, SVM, Genetic
5. Previous non-renewal 5. Logistic, SVM, Random Forests
1. Usage levels 1. SVM with ALBA
Usage Behavior
2. Usage levels 2. Logistic, NN, SVM, Genetic
1. Add-on services 1. Logistic, NN, SVM, Genetic
2. Dec Tree, Naïve Bayes, Logistic,
Switching Costs
2. Pricing plan NN, SVM
3. Ease of switching 3. Graphical comparison
1. Psychographic Segment 1. Logistic, NN, SVM, Genetic
Customer
Characteristics 2. Demographics 2. Logistic, NN, SVM, Genetic
3. Customer tenure 3. Logistic, Decision Tree
1. Mail responders 1. Bagging and Boosting
2. Response to direct mail 2. Logistic, SVM, Random Forests

Marketing 3. Previous marketing

campaigns 3. Decision rules
4. Acquisition method 4. Probit
5. Acquisition channel 5. Logistic
1. Neighbor churn 1. Hazard
2. Random Forests, Bayesian
Social 2. Social network connections Networks
Connectivity
3. Social embeddedness 3. Decision rules
4. Neighbor/connections
usage 4. Logistic

Table 6: Predictors of Churn in Contractual Settings

Ascarza, Neslin, Netzer, Lemmens, Aurelie143

Table 6 summarizes variables predictor variables for several different industries,

all in contractual settings, but many will be useful for the sports betting industry.
“These include well-researched predictors like customer satisfaction, usage
behavior, switching costs, customer characteristics, and marketing efforts, as

155
 ANDREW W. PEARSON

well as more recently explored factors such as social connectivity,” explain

Ascarza et al.143
According to Ascarza et al., “Social connectivity factors can predict churn.” 143
Studies in the telco industry have shown that high ‘social embeddedness’, i.e.,
the extent to which the customer is connected to other customers within the
network, is negatively correlated with churn. 149 Additionally, “the behavior of
customer’s connections also affects her own retention.”143
Two studies150 151 found that a customer is more likely to churn from a service or
company if his or her contacts within the company has churned. Conversely, an
Ascarza, Ebbes, Netzler and Danielson study152 found that a customer is less
likely to churn if his or her contacts increase use of the service. Due to network
externalities, Ascarza et al. argue, “the service becomes less valuable to the
customer if her friends are not using it.”143 “These social-related factors are more
likely to predict churn for network oriented services such as multi-player gaming,
communications, shared services, since customers exert an externality by using
the service,” state Ascarza et al.143 Ascarza et al. conclude that, “If ones goes
beyond the predictive power of social connections and towards understanding
the social effect, it is imperative to consider the similarity among connected
customers (homophily), correlated random shocks among connected customers
(e.g., a marketing campaign that is geographically targeted and hence affects
connected customers), and true social contagious effects.” 143
Ascarza et al. also consider whether using ultra-fine grained “big data” can be
used to improve churn prediction.143 According to Ascarza et al., these include
“actions consumers take such as visiting a web page, visiting a specific location,
‘Liking’ something on Facebook, etc,”143 but the researchers have found no
evidence of this sort of data improving retention. 143 However, this type of data
has been found to improve the prediction of customer acquisition153 154 and
cross-selling155 so Ascarza et al. believe it could be considered for churn
prediction as well.
The main goal of a retention program is obviously to prevent churn, therefore
understanding the causes of such churn behavior is imperative if you are to
design an effective retention program.143 Ascarza et al. believe that, “There is a
difference between determining the best predictors of churn and understanding
why the customer is at risk of churning.”143 “For example, demographic variables
might predict churn, but these variables rarely cause customers to leave the
company.”143 “The distinction becomes less clear when we consider factors like
past consumption or related behaviors. Are heavy users more likely to be
retained because they consume more or is it that satisfaction with the service is
driving both behaviors?” ask Ascarza et al.143
Additionally, “identifying specific causes for an individual customer is quite
different from identifying general causes in a population. ”143 To identify the

156
 THE A.I. CRUISE LINE

potential causes of churn for an individual customer, the variables or

combinations of variables that are both viable causes and for which the customer
exhibits a risky behavior must be discovered.143
A competing risk hazard model could be used “to predict which of the possible
reasons of churn are most likely to cause churn at any point in time.”143 Once the
causes of churn are identified, the cruise line needs to isolate those that are
controllable and those that are not.143 Both correlates of low retention and also
the causes of it need to be identified.143
Ascarza et al. argue that it’s not always the customers who are at the highest risk
that should be targeted.143 In her article Retention Futility: Targeting High Risk
Customers Might Be Ineffective156, Eva Ascarza explains that the highest risk
customers may not be receptive to retention efforts. As Ascarza et al. put it,
“They might be so turned off by the company that nothing can retain them.” 143
Ascarza et al. advise focusing on customers who are at risk of leaving and are
likely to change their minds and stay if targeted. 157
Although research on customer response to retention programs is scarce,
Ascarza’s Retention Futility: Targeting High Risk Customers Might Be
Ineffective156, which advocates for the use of “uplift” models. “In the jargon of
customer-level decision models, uplift modeling recognizes customer
heterogeneity with respect to the incremental impact of the test,” states
Ascarza.156 Incremental impact can be modeled in many different way, including
by utilizing interactions models and machine learning methods.158 159 156
Even customers who are likely to both churn and respond still might not be worth
rescuing because they might have low lifetime values.143 Cruise lines should
always consider the potential profit that a retention action can generate, which
depends on “the likelihood of churning, the incremental likelihood or
responding, the customer’s CLV, and the incentive cost. 156 157 160
In the majority of cases, most of the customer targeted in a proactive campaign
will be those whom the company would have retained anyway. 143 In their paper
Building Data Applications for CRM161, Berson et al. discovered that “customers
targeted by a retention campaign who did not accept the retention offer ended
up with a higher churn rate than average.” Ascarza et al. speculate this is because
“the offer triggered these customers to examine whether they wanted to stay
with the company, and the answer turned out to be ‘no.’” 143 Although Ascarza
et al. found the study provocative, they did also point out that, the data did “not
permit rejection of the alternative explanation that the retention offer
bifurcated non-churners (satisfied customers who therefore accepted the offer)
and churners (dissatisfied customer who therefore churned without accepting
the offer).”143
More troublingly, in their article The Perils of Proactive Churn Prevention Using
Plan Recommendations: Evidence from a Field Experiment162, Ascarza, Iyengar

157
 ANDREW W. PEARSON

and Schleicher showed that some would-be non-churners could be provoked by

the retention effort to churn. “Non-churners may be continuing to transact with
the firm partly out of habit or inertia,”162 and the retention efforts may
inadvertently make people realize they are unhappy with the status quo, and
paradoxically cause churn.162
According to Ascarza et al., another factor to consider in deciding whom to target
is the position of the customer in the firm’s social network.”162 “Taking a social
perspective, a customer with many contacts, or highly connected with customers
who themselves are highly connected, can be very valuable because his/her
defection could cause others to churn,” warn Ascarza et al. 162 Individuals who
are central in a network might have a lower risk of churning because of the high
social cost of leaving.163 Because of the tendency of individuals to be in social
networks with others like them, high profitability customers may have a strong
effect on each other, which provides even more of an incentive to target high
CLV customers, as well as to provide channels to market virally. 164
Hogan, Lemon and Libai point out that product life cycle should also be kept in
mind as “socially related monetary loss due to customer churn would be much
higher early in the product life cycle, when social influence is critical in driving
product growth.”165
One important consideration that needs to be kept in mind is the selection of
the best action to take to prevent churn.162 Price incentives are one of the top
considerations and they can be effective in the short term, but they are easily
replicated by competitors.162 This might put pressure on margins as customers
can become overly price sensitive. Ascarza et al. argue that “non-price
incentives, such as product improvements (e.g., a gaming company adds
additional levels in a game) may work better in the long term.” 162
Another approach is to let customers pick the incentive amongst a set of
options.162 Research by Shrift and Parker166 has shown that including in that set
the option of no-choice (i.e., doing nothing) increases persistence among
customers, which would likely results in higher retention. Sports book can also
“design the retention effort in a way that it will mainly affect customers at high
risk of churn and/or in a way that all customers, targeted or not, would
appreciate it (e.g., product or service improvement).” 162
The element of surprise has proven to be quite positive for customer retention
as well. Research by Oliver, Rust and Varki167 showed that surprising positive
events attach a customer to the company. This is particularly important because
the retention campaign will most likely target many non-would-be churners and
it should enhance retention in the long run even among those uninterested in
churning.162
Ascarza et al. argue that “the best way to conceptualize when-to-target is to
consider the different type of marketing campaign throughout the customer’s

158
 THE A.I. CRUISE LINE

lifecycle: acquisition => pre-emptive => proactive => reactive => win-back => post
win-back”162 (see Figure 17). Although it won’t be necessary to target
immediately after customer acquisition, this is the time a company needs to start
thinking about retention.162 Ascarza et al. give an example from the telco
industry in which the provider would make sure the customer is on the right data
plan from the very beginning, thereby ensuring there won’t be any nasty bill
surprises.162
“Pre-emptive timing would be to target the customer before the customer
shows any sign of diminished retention,”162 while proactive timing would be the
launching of a “campaign targeted at customers who are identified as a retention
risk based on predictive models.”162 “Reactive timing is when the firm tries to
prevent the customer from churning, while that customer literally is in the act of
churning."162 Win-back is when the customer has churned and the company
attempts to re-acquire that customer.162 Post win-back actions refers to contacts
initiated after the customer has rejected a win-back offer.162

Figure 17: Alternative Timing of Retention Campaigns

Source: Ascarza, Neslin, Netzer, Lemmens, Aurelie162
Once campaigns have been initiated they need to be evaluated, if possible by
using a control group randomly selected not to be targeted. 162 “This allows top-
line results to be compiled easily without formal causal modeling,” note Ascarza
et al.162 Overall profitability metrics and various retention measures should be
calculated.162 The rescue rate should be calculated for all of the cruise line’s
retention campaigns.162 This will allow the cruise line to “undertake a meta-

159
 ANDREW W. PEARSON

analysis across multiple campaigns to understand which factors influence rescue

rates (incentive characteristics, characteristics of customers targeted, the match
between these two, etc.).”162 This meta-analysis could reveal both immediate
insights as well as long-term trends that affect the cruise line’s business.
Long-term impacts of the campaign should also be part of the evaluation. 162 As
Ascarza et al. note, ‘”Yes, the customer might have been retained this time, but
what impact did that have on the customer’s future profitability? What
happened to the customer’s retention rate after the campaign? Did it increase
because the customer was more satisfied, or decrease because now the
customer expected if not demanded incentives?” 162
Multiple campaign management is similar to single campaigns except now the
following questions need to be asked in a dynamic setting across several
campaigns:
• Who is at risk?
• Why are they at risk?
• Who should be targeted?
• When should they be targeted?
• With what efforts?
• What was gained by these actions?143
In their article Database Marketing: Analyzing and Managing Customers 168,
Blattberg, Kim, and Neslin discuss two key issues in multiple CRM campaign
management: wear-in and wear-out. As Ascarza et al. note, “A campaign may
take time before it reaches its maximal impact (wear-in) and then decline at
some rate afterwards (wear-out). These concepts have important implications
for the spacing of retention campaigns.”143
On the individual level, multiple campaign management is tricky because any
current campaign can influence what “state” a customer might be in for the next
campaign.143 A customer who receives an incentive that is clearly of lesser value
than a previous incentive might be offended and so put off by the offering that
the incentive works against the giving company. This, Ascarza et al. suggest that
dynamic optimization is required.143
In terms of integrating strategy, it is important to coordinate both the company’s
acquisition and retention as well as its retention spending with the company’s
marketing strategy and its segmentation, targeting and positioning (STP)
approach.143 Optimizing both acquisition and retention can be a tricky endeavor,
but it is important to do.143
Ascarza et al. provide an example from the financial services industry that show
how easy it is for things to get out of sync.143 They consider the STP of a financial
services company that may be trying to segment the market by customer value
and also wants to target high-value customers with premium products and

160
 THE A.I. CRUISE LINE

services.143 A retention campaign emphasizing promotional discounts would

then be “off-strategy.”143
Ascarza et al. believe that emotions and social connections are important when
it comes to identifying who is at risk of churn, but what’s more important is
understanding why the customer is at risk and whom should be targeted.143
“Social connections data could be particularly important for key ‘influencers’,
i.e., customers that provide network value to other customers, as Ascarza, Ebbes
et al. put it.152
“Emotional and other indicators gleaned from text data should provide key
insights on why customers churn,” argue Ascarza et al. 143 They add, “Textual
analysis approaches such as Linguistic Inquiry and Word Count (LIWC) and topic
modeling may be used to extract measures of emotions and other useful insights
from consumer data and leverage these for churn prediction and
management.”143
Engagement metrics do have the potential to identify what retention actions
need to be taken.143 For example, measures of how a customer plays a slot
machine could provide a hint of what incentive (monetary or not) the customer
needs to keep playing.143 Ascarza et al. conclude that, “knowledge management
will be necessary to ensure the firm’s entire experience base in retention
management is codified and accessible to planners. This is especially important
for planning multiple campaigns and integrating strategy, since these tasks
require a broad understanding of the firm’s experiences.” 143
When it comes to building the actual models, Ascarza et al. proffer that, tried
and true regression-based predictive modeling is useful for constructing the
following143:
• “Predictive models for identifying who is at risk and who will respond
to targeting.
• “Meta analyses of field data that provide insights needed for multiple
campaign planning.
• “Marketing mix model that can drive strategy integration.” 143
Deep learning has been used to uncover the probability that a customer might
defect169 and it may also help in modeling response to retention offers. 143
Ascarza et al. add that, boosted varying-coefficient regression models have been
studied for dynamic predictions and optimization in real time 170, which have
shown to offer major improvements over the classic stochastic gradient boosting
algorithm currently used for churn prediction.171 160 Ascarza et al. see these
methods as promising discoveries for retention management.
“Dimensionality reduction and variable selection techniques form another major
development in the machine learning field, and may be useful for retention
research and practice, which face an overflow of potential defection predictors,”

161
 ANDREW W. PEARSON

add Ascarza et al.143 Companies wanting to build models predicting who is at risk,
whom to target, and when to target should look at modern regularization
techniques like Lasso172, elastic net173, and adaptive regularization174. Churn
modelers should consider the Cox proportional hazard models175 176 as it has
been proven particularly effective, and it is available in the R package ‘glmnet’. 143
“State-of-the-art regularization models, which systematically control for
overfitting and thus allow modeling with larger feature sets, also offer the
potential to expand the set of predictors to include interactions between churn
predictors,” explain Ascarza et al.143 They also believe that when estimating the
effects of churn incentives it helps to include “interactions between the
treatment variable (i.e., being targeted with a retention action) and customer or
campaign design covariates.” 143
The use of analytics and data management to help detect and avoid the act of
attrition is something that can benefit all cruise lines. Churn questions that a
cruise line should be asking include:
• How is the cruise line detecting behavioral changes in its patrons?
• Does the cruise line have steps in place to identify when the customer
experience is going wrong, or when the customer is about to leave?
Cruise line operators can use Master Data Management (MDM) techniques to
communicate important customer preference information to staff who sit at
interaction points throughout the operation.
To ensure customer retention is front and center, cruise lines should be scoring
their databases on a regular basis in order to understand the likelihood of a
customer churning from their venue. This kind of modeling is prevalent in the
telecommunications, finance, and utilities industries, and should be utilized in
the gaming industry as well. While a slightly different set up due to those
industries mostly having their customers locked into contracts, gaming
companies need to stay ahead of the game in retaining their customers.
Anecdotal evidence collected in our discussions with gaming companies have
indicated a tendency to ignore customers until they have not been seen for up
to two years. At this stage, there might be a marketing activity targeted at the
customer for up to 12 months. It could be proffered that, by this stage, it is too
late to win the customer back; the customer has probably already made up his
or her mind and, once a decision like that has been made, it is almost impossible
to reverse it, no matter how attractive any competing offer might be.
One of the hardest parts for a gaming company to determine — as opposed to
commercial entities that have their customers on contract and definitely know
they are tied down — is whether the customer has categorically churned. It may
be that a change in location, circumstances, or something else has caused a
customer to disappear from the cruise line, with every intention of returning.

162
 THE A.I. CRUISE LINE

However, statistical measures could be used to identify customer’s whose

behavior has changed and the change wouldn’t be attributed to chance.
Historical internal data can be used to model the difference between a churned
customer and one who is still engaged. There would be significant metrics in the
data that identify the likelihood of churning. Similar to the acquisition model
described above, a parametric equation could be constructed that elicits the
association and relationship between the target variable and the predictors.
This model would serve as an early warning system for the cruise line. It would
also be a strategic tool useful to predict whether a customer was deemed worth
retaining or not. The model should be run on a regular basis across the entire
customer database to understand which customers have reached or are
reaching a critical value in their churn score. The theory: these customers would
then be targeted with an offer to return to the cruise line, in the process avoiding
the likelihood of them churning. Alternatively, if the customer is deemed to be
of little or no value, there would be no offer forthcoming to entice them to
return.

Optimizing Offers
As Sutton explains, “In addition to predicting the future worth of patrons, it is
important to know which marketing campaigns are the most effective for driving
response, revenue, and profit. In general, certain offers are better than others,
and specifically certain offers will be better for certain patrons.”119
“While knowing the probable future worth of a patron is critical for determining
the reinvestment level for which a patron is eligible, patrons’ behaviors and
interests can be used to identify the offer(s) that will be most appealing to each
patron as well as the ones generating the most profitable response,” Sutton
explains.119 By analyzing the likelihood that a patron will respond to a certain
offer or offers, cruise line and cruise line analysts can optimize the offer that
each patron is given in order to maximize the amount of revenue and profit
driven by the marketing campaigns as a whole.119
As previously mentioned, A/B testing is one of the best ways to identify which
offers work best. A/B testing involves “testing two different offers against one
another in order to identify the offer that drives the highest response and the
most revenue/profit,” explains Sutton.119 “More advanced statistical methods
can be used to generate likelihood of response scores and classification scores.
Some of the more common statistical approaches are logistic regression,
decision trees, and discriminant analysis,” Sutton states.119
“Essentially, these statistical methods use historical data to find the factors that
are related as to why a patron responds. Those factors can then be used to assess
the likelihood of response based on the similarity of a patron profile to that of
responders,” adds Sutton.119

163
 ANDREW W. PEARSON

“These methods have historically been used in direct marketing analysis to

identify the best types of offers and the most likely responders,” says Sutton.119
“In order to build accurate and predictive response models, historical data about
response is required. The likelihood of response might be a broad measure of
response that refers to the likelihood a patron will respond to any offer, or it
might be specific to the likelihood of response to a specific type of offer.”119
In addition, Sutton adds, “it’s a good idea to select test segments of customers
for the purpose of continually testing new offers. Doing so will help to ensure
that there is a large amount of response data that can be used to build models
and continually improve the efficacy of marketing.”119 “Effective response
models will help identify which patrons are most likely to respond to an offer,
and in turn to which offer patrons are most likely to respond,” concludes
Sutton.119

Chronological View of a Cruise Line Analytics Implementation

Data reduction via cluster analysis and segmentation is a logical starting point
and the initial work should be around identifying patron preference(s). Reducing
the customer database into more manageable and meaningful segments has
many advantages; the preferences that can be derived are dependent on the
availability of meaningful distinguishing factors.
• Segmentation models use customer metrics that help reduce and
profile the customer database and should be constructed as this
information can be the underpinning for further analyses, such as
patron acquisition worth models.
• A Propensity to Respond Model is heavily dependent on the marketing
data and the veracity and richness of it. The cruise line would need to
develop the whole view of the customer first, but, once developed, this
is one of the most powerful marketing models available.
• Customer Conversion Model could be viewed as an extension of a
number of the above models, with the idea to derive a data driven
metric that scores a customer’s likelihood of returning after his or her
first trip.
• Patron Likelihood to Return Model requires a complete view of the
customer along with considerable marketing data. This would help with
offers sent, who was sent offers, who responded to the offers, etc., etc.
The derived metric on its own would have value, but it could also be a
significant input into a two stage model to predict next trip value and
worth.
• A Patron Worth Model would identify the cruise line’s most valuable
patrons. The assumption is that a cruise line would be looking to predict
different metrics, such as worth on the next trip, worth over the next
12 months, lifetime value, etc., etc.

164
 THE A.I. CRUISE LINE

• RFM is a method used for analyzing customer value and it is commonly

utilized in database marketing and direct marketing and has received
particular attention in the cruise line industries. Cruise lines should keep
scores of data about a customer’s purchases that includes a table with
the customer name, date of purchase and purchase value. From this
data, a cruise line can score the true value of a customer and this
information can be fed to the marketing department, which can decide
to send an offer to the client if it is worth it.
• A Customer Acquisition Model would then be built by using the results
of the segmentation modeling models (or a different metric for
desirable customers). A deeper investigation of a cruise line’s source
systems is needed and this could be part of the analysis to help
understand what is available, and what might be able to be used from
external parties. Different jurisdictions would have different models.
• Customer Churn Models would require preliminary analysis to extract
only engaged customers. The cruise line would need to derive a
statsitcially dirven metric that indicated whether a customer had
churned or not. The cruise line could then build models to detect
upcoming patron attrition.
Customer analytics have evolved in six core dimensions: strategy, organization,
data, technology, analytics and measurement, and process. In terms of analytics,
cruise lines should start with segmentation to build a comprehensive view of
their customers. Segmentation provides multiple payoffs across the customer
life cycle, from acquisition through retention. Segmentation might start with
simple attributes such as bets, or even geography, but it should evolve over time
to characteristics such as lifetime value.
Table 7 shows the other types of analytics that should be used during a customer
lifecycle. Many of these models require complex data integration / data
virtualization, CRM, social, and analytics systems working in harmony with each
other.

Life-cycle Business objective Analytical method

stage
Profile customers Segmentation

Discover Evaluate prospects Lead scoring

Reach the right prospects Customer lookalike targeting

Analyze customers’ responses Offer/contact optimization

Explore
Optimize marketing mix Marketing mix modeling

165
 ANDREW W. PEARSON

Life-cycle Business objective Analytical method

stage
Delivering contextually relevant Customer location analysis
content
Test marketing inputs A/B and multivariate testing
Predict future events Propensity models
Buy Expand wallet share Cross-sell/upsell
Target accurately In-market timing models
Drive deeper product use Product & recommendation
analysis
Use
Understand use Customer device use analysis
Understand customer satisfaction Customer satisfaction analysis
Learn about drivers of Engagement analysis
engagement
Ask
Improve customer service Customer device use analysis
Identify customer pain points Voice of the customer analysis
Manage defection of customers Churn models
Personalize marketing efforts Next-best-action models
Maximize customer value Lifetime value models
Engage Increase depth of relationship Loyalty models
Optimize customer interactions Customer journey analysis
Understand customer Social network analysis
relationships

Table 7: Analytics Across the Customer Lifecycle

Source: Forrester’s How Analytics Drives Customer Life-Cycle Management177

Clickstream Analysis
When a person browses a website, he or she leaves behind a digital trail, which
is known as a clickstream. Clickstream analysis (also called clickstream analytics)
is the process of collecting, aggregating, reporting and analyzing the browsing
behavior of a web surfer to better understand the intentions of users and their
interests in specific content or products on a website. Clickstream analysis is the
process of collecting, analyzing and reporting aggregate data about which pages
a website visitor visits — and in what order. The path the visitor takes though a

166
 THE A.I. CRUISE LINE

website is, basically, the clickstream.

There are two levels of clickstream analysis: traffic analytics and e-commerce
analytics. Traffic analytics operates at the server level and tracks how many
pages are served to the user, how long it takes each page to load, how often the
user hits the browser's back or stop button and how much data is transmitted
before the user moves away from the website. E-commerce-based analysis uses
clickstream data to determine the effectiveness of a website as a channel-to-
market. It is concerned with what pages the browser lingers on, what he or she
puts in or takes out of a shopping cart, what items are purchased, whether or
not the buyer belongs to a loyalty program and uses a coupon code, as well as
his or her preferred methods of payment.
Utilizing clickstream analysis, a cruise line can help build a Master Marketing
Record for each customer in real-time. This allows the cruise line to test
scenarios and options for the website,
as well as develop personalized
responses for individuals. The system
should include a combination of social
listening, analytics, content
publication and distribution, and
tracking, as well as a strong workflow
and rules engine that is geared around
strong governance. All of these
applications are built to ultimately
feed a Master Marketing Profile — a
centralized customer record that pulls
in all data based on digital activity that
can be identified by a single customer
ID.
Figure 18 shows the customer funnel
that takes an anonymous web browser
to a known patron. Through
clickstream analytics, personalization
Figure 18: Customer funnel marketing can begin, and associating
this activity with a customer once he or
she walks through the front door should be a cruise line’s primary goal. This can
be done by enabling new users to log into his or her account via web or mobile
applications, like a cruise line’s WeChat app.
In their article Big Data and Competition Policy: Market Power, personalised
pricing and advertising 178, Marc Bourreau et al. explain that firms may collect
personal and non-personal data about users, as well as machines, in several
different ways, including:

167
 ANDREW W. PEARSON

• Publicly observed through device, operating system, IP address, etc.

• Voluntarily provided by the consumer, “either with knowledge when
registering to a website, such as name, data of birth, email or postal
address for delivery, etc., or often without knowledge when logging into
a website (login-based data) such as products the consumer is looking
for, purchases, etc.”178
• Tracking the consumer online, which can be achieved in different ways,
such as:
o “tracking cookies, which are a specific type of cookie that is
distributed, shared, and read across two or more unrelated
websites for the purpose of gathering information or
presenting customized data to a consumer”178;
o “Browser and device fingerprinting, which is a method of
tracking web browsers by the configuration and settings
information they make visible to websites”178;
o “History sniffing, which is the practice of tracking which sites a
user has and has not visited (by hacking its browser history
list)”178;
o “Cross-device tracking offers the ability to interact with the
same consumer across her desktop, laptop, tablet, wearable,
and smartphone, using both online and offline information”178;
o “Through the use of applications by the user, this information
is accessible for the Operating System owner as well as for the
developer of the application.”178
In his article Google Attribution Allows Clear, Seamless Campaign Analysis for
Marketers179, Matthew Bains explains that Google has released a new tool called
Google Attribution that “uses machine learning and data to help marketers
measure the impact of each of their marketing touch points, across multiple
channels, and across multiple devices.” “It uses data that’s already there from
Adwords and Google Analytics; it just takes that data and shows you how each
customer moved through their buyer’s journey and attributes those conversions
respectively. It provides a single view of the path to purchase to help marketers
learn what is actually working compared to what seems to be working,”179 adds
Bain.
Wanamaker would be ecstatic as “marketers can finally begin to answer the age-
old question that is typically at the forefront of their minds — is my marketing
working?”, as Bains puts it179
Moving away from the flawed last-click attribution idea, “Google Attribution
uses machine learning and data to help marketers measure the impact of each
of their marketing touch points, across multiple channels, and across multiple
devices.”179 Google Attribution shows users how each customer moves through
his or her buyer’s journey and attributes those conversions respectively. 179 “It

168
 THE A.I. CRUISE LINE

provides a single view of the path to purchase to help marketers learn what is
actually working compared to what seems to be working,” Bains explains.179
As Bains warns179:
“With last click, the reward for the conversion often went to
the last touch point that the user made, often with a sale after
a click on an ad. This could lead to false impressions about the
effectiveness of an ad campaign versus display ads, organic
search, social, email affiliates, and many other interactions
that a customer made with a business along the buyer’s
journey. Maybe organic search is actually more important than
display ads or vice versa.”
“The aim of Google Attribution is to simplify the complex problem of
multichannel, multi-device attribution by leveraging data advertisers already
have in Google Analytics, AdWords, or DoubleClick Search,” adds Kishore
Kanakemedela, director of product management at Google.179
With Attribution, users can see how effective each step of a campaign is,
whether that step is a video ad, a banner ad, a carousel ad, an email, a social
campaign, or any other quantifiable digital content.179 Attribution will show
users how these micro-moments worked together to spot leads and drive them
to conversions.179
Marketers will now have more transparency on what is actually driving their
business, which in turn, can help them better allocate their budgets between
channels179; quantifiable success on one channel leads to increased budget
spend for that channel, that is until numbers drop off, then reallocation
commences.

Location Analytics
Location analytics is a technology that enables firms to capture and analyze
location data on customers who are at a physical venue. In his article How
Location Analytics Will Transform Retail 180, Tony Costa argues that, “By
leveraging connected mobile devices such as smartphones, existing in-venue Wi-
Fi networks, low cost Bluetooth-enabled beacons, and a handful of other
technologies, location analytics vendors have made it possible to get location
analytics solutions up and running fast at a minimal cost.” “Customer tracking
data is typically sent to the location analytics vendor where it is analyzed and
accessed via online dashboards that provide actionable data tailored to the
needs of specific employees — from the store manager to the executive C-suite,”
adds Costa.180
Already, the scale of data collected by early adopters is venturing into “Big Data”

169
 ANDREW W. PEARSON

territory. Location analytics firm RetailNext currently “tracks more than 500
million shoppers per year by collecting data from more than 65,000 sensors
installed in thousands of retail stores. A single customer visit alone can result in
over 10,000 unique data points, not including the data gathered at the point of
sale.”180
RetailNext is not alone; Euclid Analytics — one of the biggest players in this space
— “collects six billion customer measurements each day across thousands of
locations, and multiple location analytics firms surveyed said they are adding
hundreds of new venues each month.”180
As Costa explains, venue owners are applying insights gathered from location
analytics to help in all aspects of their business, including180:
1. Design. “After analyzing traffic flows in their stores, a big box retailer
realized that less than 10% of customers visiting their shoe department
engaged with the self-service wall display where merchandise was
stacked. The culprit turned out to be a series of benches placed in front
of the wall, limiting customer access.”180 Simply relocating the benches
to enhance accessibility increased sales in the department by double
digits.180
2. Marketing. A restaurant chain wanted to understand whether or not
sponsoring a local music festival had a measurable impact on customer
visits. After collecting data on 15,000 visitors passing through the
festival entrances and comparing it to customers who visited their
restaurants two months before and after the festival, they concluded
that the festival resulted in 1,300 net new customer visits.180
3. Operations. “A grocery store chain used location analytics to
understand customer wait times in various departments and check-out
registers. This data not only enabled the company to hold managers
accountable for wait times, but it gave additional insight into (and
justification for) staffing needs.”180
4. Strategy. A regional clothing chain was concerned that opening an
outlet store would cannibalize customers from its main stores. “After
analyzing the customer base visiting each store, they discovered that
less than 2% of their main store customers visited their outlet. The
upside: the outlet gave them access to an entirely new customer base
with minimal impact to existing store sales.”180
Just as web analytics is an essential tool on the web, location analytics will
become a must-have for designing, managing, and measuring offline
experiences.180 Location analytics is set to have a profound impact on how
businesses operate in the very near future. Costa adds that, “Beyond creating
more efficient, effective and meaningful services, firms will begin to rethink the
notion of customer value.”180

170
 THE A.I. CRUISE LINE

Costa argues that having the ability “to identify, track, and target customers in
physical locations will enable companies to extend preferential status and
rewards to customers based on their behaviors, rewarding them on the number
and frequency of visits, where they go in venues, and their exclusive loyalty (i.e.,
not visiting competitor venues).”180

Conclusion
In this chapter, I wanted to lay out the many ways in which The A.I. Cruise Line
can track and understand its customer base on both a micro and a macro level.
Many of the analytical models I mention in this chapter have been around for
decades and every cruise line should be aware that creativity with these models
is what will separate them from their competitors.
With today’s IT budgets in the millions of dollar per year, every cruise line can
afford to buy software that segments its customers, creates marketing
campaigns and predicts customer churn, but it’s what it does with this
information that matters most. Customers want to be wowed and this is not an
easy thing to do.
Analytics can be useful for the entire customer journey process, from the initial
moment a customer is picked up in a clickstream, through the descriptive
analytics process of understanding website traffic, to data mining and diagnostic
analytics utilized to understand customer spend.
The complexity coming to this world, however, will radically alter the customer
experience and cruise lines need to prepare for these radical changes now. In
chapter four, I will delve into the future of marketing, a place where data is used
for everything from website morphing to psychometrics, to affective computing,
a place where the psychology of personalization becomes an integral part of real-
time marketing.
In this chapter, I delved into the history of analytics and it is important to
understand what today’s analytics environment was built upon to truly grasp
where it might be going. In the next chapter, I break down the various analytical
processes that are important to cruise line executives; some of these are decades
old, while others are quite new; more are surely on the way as computing power
is increasing exponentially and software is getting much more powerful and
much more sophisticated by the day. Vast sets of data can be culled through and
acted upon by cloud-based servers that can be spun up, utilized to build highly
sophisticated models and then turned off almost instantly, meaning data can be
crunched only as needed, thereby reducing unnecessary costs.
In the next couple of chapters, I will look at how these technologies can shape
the customer experience so that true personalization can be delivered to a
market of one. Capturing a first time visitor’s IP address can be an important —

171
 ANDREW W. PEARSON

and necessary — first step in the customer relation and once a user signs up for
a patron card all of his or her customer information becomes relevant, which
means personalization marketing should be in the cards, so to speak.
Analyzing clickstream data, customer card data, marketing data, as well as social
media data can help cruise lines develop three dimensional profiles on each of
their customers and, once these profiles are perfected, the behavioral marketing
work can begin to ensure that the cruise line is bringing in the customers that
will produce the highest ROI. Matching customer needs with the cruise line’s
staffing and operation requirements then becomes an added cost reduction per

108 Stodder, D. (2012). Customer Analytics in the Age of Social Media. Retrieved from
Business Times:
http://www.businesstimes.com.sg/archive/monday/sites/businesstimes.com.sg/files/Cu
stomer%20Analytics%20in%20the%20Age%20of%20Social%20Media.pdf
109 http://www.webopedia.com/TERM/C/customer_analytics.html
110 IBM. (2013). Achieving Customer Loyalty with Customer Analytics. Retrieved from
adma.com: http://www.adma.com.au/assets/Uploads/Downloads/IBM-Achieving-
Customer-Loyalty-with-Analytics.pdf
111 Davenport, T. (January 2006). Competing on Analytics. Harvard Business Review.
112 Shearer C., The CRISP-DM model: the new blueprint for data mining, J Data
Warehousing (2000); 5:13—22.
113 Harper, Gavin; Stephen D. Pickett (August 2006). "Methods for mining HTS data".
Drug Discovery Today. 11 (15–16): 694–699.
114 SAS Institute. Data Mining and the Case for Sampling.
http://sceweb.uhcl.edu/boetticher/ML_DataMining/SAS-SEMMA.pdf (Accessed
November 21, 2017).
115 https://en.wikipedia.org/wiki/Machine_learning (Accessed 20 November 2017).
116 Vogel, Dirk. (6 September 2017). 5 Ways AI Will Boost Personalization in Digital
Marketing. Selligent. https://www.selligent.com/blog/contextual-marketing/5-ways-ai-
will-boost-personalization-in-digital-marketing (Accessed 24 October 2017).
117 Johnson, Lauren. 2017. Pinterest Is Offering Brands Its Visual Search Technology To
Score Large Ad Deals. Adweek. October 2, 2017.
http://www.adweek.com/digital/pinterest-is-offering-brands-its-visual-search-
technology-to-score-large-ad-deals/ (Accessed 18 November 2017).
118 Walch, Kathleen. Forbes. The Seven Patterns of AI. 17 Septmber 2019.
https://www.forbes.com/sites/cognitiveworld/2019/09/17/the-seven-patterns-of-
ai/amp/?__twitter_impression=true (Accessed 24 October 2019).
119 Sutton, Scott. 2011. Patron Analytics in the Retailer and Gaming Industry: How the
House Always Wins. SAS. http://support.sas.com/resources/papers/proceedings11/379-
2011.pdf
120 Nyce, Charles. 2007. Predictive Analytics White Paper.
https://www.scribd.com/document/200505883/Predictive-Analytics-White-Paper
121 Eckerson, Wayne. 2007. Predictive Analytics, Extending the Value of Your Data
Warehouse Investment. TDWI Best Practices Report.
https://www.sas.com/events/cm/174390/assets/102892_0107.pdf

172
 THE A.I. CRUISE LINE

122 Profitect. Prescriptive Analytics Makes Waves with Retail & CPG.
https://bbc604576ecf2bcb1563-
2fd6af82f3627aa26d05cd090ce38fd1.ssl.cf5.rackcdn.com/profitect-1-422FA4M.PDF
(accessed 26 October 2017).
123 How Predictive Analytics is Changing the Retail Industry. International Conference on
Management and Information Systems. September 23-24, 2016.
http://www.icmis.net/icmis16/ICMIS16CD/pdf/S154.pdf
124 https://en.wikipedia.org/wiki/Decision_tree
125 https://www.lucidchart.com/pages/decision-tree (Accessed 22 January 2018).
126 Deng, L., Gao, J., Vuppalapatie, C. Building a Big Data Analytics Service Framework for
Mobile Advertising and Marketing. March 2015.
https://www.researchgate.net/profile/Jerry_Gao/publication/273635443_Building_a_Bi
g_Data_Analytics_Service_Framework_for_Mobile_Advertising_and_Marketing/links/5
508de220cf26ff55f840c31.pdf
127 Telgarsky, Matus, and Andrea Vattani. "Hartigan's Method: k-means Clustering
without Voronoi." AISTATS. 2010.
128 Hartigan, J.A., Wong, M.A. A K-Means Clustering Algorithm. Journal of the Royal
Statistical Society. Series C (Applied Statistics), Vol. 28, No. 1 (1979)
http://www.cs.otago.ac.nz/cosc430/hartigan_1979_kmeans.pdf
129 Pierson, Lillian. Solving Real-World Problems with Nearest Neighbor Algorithms.
www.dummies.com. http://www.dummies.com/programming/big-data/data-
science/solving-real-world-problems-with-nearest-neighbor-algorithms/
130 https://en.wikipedia.org/wiki/Logistic_regression
131 Karp, A. H. (2009). Using Logistic Regression To Predict Customer Retention. New
York. Sierra Information Service, Inc.
http://www.lexjansen.com/nesug/nesug98/solu/p095.pdf
132 Siroker, Dan, Koomen, Pete. A/B Testing: The Most Powerful Way to Turn Clicks Into
Customers. Google Books.
133 Imdadullah, Muhammad, December 27, 2013. Time Series Analysis and Forecasting.
http://itfeature.com/time-series-analysis-and-forecasting/time-series-analysis-
forecasting
134 Sang, Hyuck Kim, Dong Jin Kim. March 23, 2016 Determining Revenue-Generating
Casino Visitors Using a Vector Autoregressive Model: The Case of the G Casino in Korea.
Journal of Quality Assurance in Hospitality & Tourism.
135 Singh Y, Chauhan AS. Neural Networks in Data Mining. Journal of Theoretical and
Applied Information Technology. 2009; 5:37-42 http://jatit.org/volumes/research-
papers/Vol5No1/1Vol5No6.pdf
136Goa, Jim. Machine Learning Applications for Data Center Optimization.
https://static.googleusercontent.com/media/www.google.com/en//about/datacenters/
efficiency/internal/assets/machine-learning-applicationsfor-datacenter-optimization-
finalv2.pdf
137 Quach, Katyanna. (2017). How DeepMind’s AlphaGo Zero learned all by itself to trash
world champ AI AlphaGo. The Register. 18 October, 2017.
https://www.theregister.co.uk/2017/10/18/deepminds_latest_alphago_software_does
nt_need_human_data_to_win/ (Accessed 21 January 2018).
138 https://en.wikipedia.org/wiki/Discriminant_function_analysis#cite_note-cohen-1
139 https://en.wikipedia.org/wiki/Survival_analysis

173
 ANDREW W. PEARSON

140 Schrage, Michael. (2019). MIT Sloan Review. Smart Strategies Require Smarter KPIs.
September 16, 2019. https://sloanreview.mit.edu/article/smart-strategies-require-
smarter-kpis/ (accessed 25 September 2019).
141 Gallo, Amy. Harvard Business Review. The value of keeping the right customers.
October 29, 2014. https://hbr.org/2014/10/the-value-of-keeping-the-right-customers
(accessed 25 September 2019).
142 Gupta, Sunil, Lehmann, Donald R., and Stuart, Jennifer Ames. (2004), Valuing
Customers. Journal of Marketing Research, 41(1), 7-18.
143 Ascarza, Neslin, Netzer & Lemmens, A. (2017). In pursuit of enhanced customer
retention management: Review, key issues, and future directions. Customer needs and
solutions, 1-17. DOI: 10.1007%2Fs40547-017-0080-0#citeas.
144 Handley, Lucy (2013). Customer Retention: Brave New World of Consumer Dynamics.
Marketing Week Online Edition, 21. March 20, 2013.
145 Forbes Insight. (2014). Company struggling to win customers for life, says new study
by Forbes Insight and Sitecore.
https://www.forbes.com/sites/forbespr/2014/09/10/companies-struggling-to-win-
customers-for-life-says-new-study-by-forbes-insights-and-sitecore/#6c4d57741407
(Accessed 21 April 2018).
146 Springer, Tom, Kim, Charles, Azzarello, Domenico, and Melton, Jeff. (2014). Service
Quality: The Impact of Frequency, Timing, Proximity, and Sequence of Failures and
Delights. Journal of Marketing, 78(1), 41-58.
147 Lewis, Michael. (2005). Incorporating Strategic Consumer Behavior into Customer
Valuation. Journal of Marketing, 69(4), 230-238.
148 Blattberg, Robert C., Getz, Gary, and Thomas, Jacqueline S. (2001). Customer Equity:
Building and Managing Relationships as Valuable Assets, Boston, MA: Harvard Business
Press.
149 Benedek, Gabor, Lubloy, Agnes, Vastag, Yule. (2014). The Importance of Social
Embededness: Churn Models at Mobile Providers. Decision Sciences, 45(1), 175-201.
150 Nitzan, Irit, Libai, Barak. (2011). Social Effects of Customer Retention. Journal of
Marketing, 75(6), 24-38.
151 Verbeke, Wouter Martens, David, Mues, Christopher, and Baesens, Bart. (2011).
Building Comprehensive Customer Churn Predictions Models with Advanced Rule
Induction Techniques. Expert Systems with Applications, 38(3), 2354-2364.
152 Ascarza, Eva, Ebbes, Peter, Netzer, Oded, and Danielson, Matt. (2017). Beyond the
Target Customer: Social Effects of CRM Campaigns. Journal of Marketing Research: June
2017, Vol. 54, No. 3, pp. 347-363.
153 Provost, Foster, Dalessandro, Brian, Hook, Rod, Zhang, Xianhan, and Murray, Alan.
(2009). Audience Selection for On-line Brand Advertising: Privacy-friendly Social Network
Targeting. In Proceedings of the Fifteenth ACM SIGKDD International Conference on
Knowledge Discovery and Data Mining, 707-716.
154 Perlich, Claudia, Delessandro, Brian, Stitelman, Ori, Raeder, Troy, and Provost, Foster.
(2014). Machine Learning for Targeted Display Advertising: Transfer Leaning in Action.
Machine Learning 95(1), 103-127.
155 Martens, David, Provost, Foster, Clark, Jessica, de Fortuny, Enric Junque. (2016).
Mining Massive Fine-Grained Behavior Data to Improve Predictive Analytics. MIS
Quarterly, 40(4) 869-888.
156 Ascarza, Eva. (2016). Retention Futility: Targeting High Risk Customers Might Be
Ineffective. Social Sciences Research Network 2759170.
174
 THE A.I. CRUISE LINE

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2759170 (Accessed 22 April

2018).
157 Provost, Foster, and Fawcett, Tom. (2013). Data Science for Business: What You Need
To Know About Data Mining and Data Analytic Thinking. O’Reilly Media, 2013.
158 Guelman, Leo, Montserrat Guillen, and Ana M. Pérez-Marín. 2012. Random forests
for uplift modeling: An insurance customer retention case. In Modeling and Simulation in
Engineering, Economics and Management. Springer, pp. 123–33.
159 Athey, Susan and Imbens, Guido W. (2016). Recursive partitioning for heterogeneous
causal effects. PNAS 113(27), 7353-7360.
160 Lemmens, Aurélie, and Gupta, Sunil. (2013). Managing Churn to Maximize Profits.
Harvard Business School (working paper).
161 Berson, Alex, Smith, Stepthen, and Thearling, Kurt. (2000). Building Data Applications
for CRM. New York, NY. McGraw-Hill.
162 Ascarza, Eva, Iyengar, Raghuram, and Schleicher, Martin. (2016). The Perils of
Proactive Churn Prevention Using Plan Recommendations: Evidence from a Field
Experiment. Journal of Marketing Research, 52(1), 46-60.
163 Giudicati, Gianna, Massimo Riccaboni, and Anna Romiti (2013), "Experience,
Socialization and Customer Retention: Lessons from the Dance Floor," Marketing Letters,
24(4), 409–422.
164 Haelein, Michael, and Libai, Barak. (2013). Targeting Revenue Leaders for a New
Product. Journal of Marketing, 77(3), 65-80.
165 Hogan, John E., Katherine N. Lemon, and Barak Libai (2003), What Is the True Value
of a Lost Customer? Journal of Service Research, 5(3), 196–208.
166 Schrift, Rom Y., and Parker, Jeffrey R. (2014). Staying the Course: The Option of Doing
Nothing and Its Impact on Postchoice Persistence. Psychological Science, 25, 3, 772–780.
167 Oliver, Richard L., Rust, Roland T., and Vaarki, Sajeev. (1997). Customer Delight:
Foundations, Findings, and Managerial Insight. Journal of Retailing, 73(3), 311–336.
168 Blattberg, Robert C., Kim, Byung-Do, and Neslin, Scott A. (2008). Database Marketing:
Analyzing and Managing Customers, New York, NY: Springer.
169 Castanedo, Federico, Valverde, Gabriel, Zaratiegui, Jamie and Vazquez, Alfonso. 2014.
Using Deep Learning to Predict Customer Churn in a Mobile Telecommunication Network.
www.wiseathena.com/pdf/wa_dl.pdf (accessed 22 April 2018).
170 Wang, Jianqiang C. and Hastie, Trevor. 2014. Boosted Varying-Coefficient Regression
Models for Product Demand Prediction. Journal of Computational and Graphical
Statistics, 23(2), 361–382.
171 Lemmens, Aurélie, and Croux, Christophe. (2006). Bagging and Boosting Classification

Trees to Predict Churn. Journal of Marketing Research, 43(2), 276–286.

172 Tibshirani, Robert J. (1996). Regression Shrinkage and Selection via the Lasso. Journal
of the Royal Statistical Society. Series B (Methodological), 267–288.
173 Zou, Hui and Hao Helen Zhang (2009). On the Adaptive Elastic-net with a Diverging
Number of Parameters. Annals of statistics, 37(4), 1733–1751.
174 Crammer, Koby., Kulesza, Alex and Dredze, Mark. (2009). Adaptive Regularization of
Weight Vectors. Advances in Neural Information Processing Systems, 22, 414–422.
175 Zou, Hui, and Trevor Hastie (2005), “Regularization and Variable Selection via the
Elastic Net. Journal of the Royal Statistical Society: Series B (Statistical Methodology),
67(2), 301–320.
176 Simon, Noah, Friedman, Jerome, Hastie, Trevor and Tibshirani, Rob. (2011).
Regularization Paths for Cox's Proportional Hazards Model via Coordinate Descent.
Journal of Statistical Software, 39(5), 1–13.175
 ANDREW W. PEARSON

177 Forrester. 2015. How Analytics Drives Customer Life-Cycle Management. SAS.com.
https://www.sas.com/content/dam/SAS/en_us/doc/analystreport/forrester-analytics-
drives-customer-life-cycle-management-108033.pdf (Accessed April 15, 2019).
178 Borreau, Marc, de Streel, Alexandre, and Graef, Inge. (2017). Big Data and
Competition Policy: Market power, personalised pricing and advertising. Centre on
Regulation in Europe. 16 February 2017. http://www.crid.be/pdf/public/8122.pdf
(Accessed January 25, 2017).
179 Bains, Matthew. (July 28, 2017) . Search Influence.
http://www.searchinfluence.com/2017/07/google-attribution-allows-clear-seamless-
campaign-analysis-for-marketers/ (accessed 25 August 2017).
180 Costa, T. (2014, March 12). How Location Analytics Will Transform Retail. Retrieved
from Harvard Business Review: http://blogs.hbr.org/2014/03/how-location-analytics-
will-transform-retail/

176
 CHAPTER FOUR: MARKETING
“Business has only two functions - marketing and innovation.”
~ Milan Kundera

Overview
One of the recurring themes of this book is self-reliance. I’m trying to lay out a
case for cruise lines to become much more self-reliant than they probably
currently are. Software companies are providing the tools for cruise lines to
become self-sufficient in areas like AI, BI, CRM, CX, SEO, marketing, website
personalization, social media, and even analytics.
Today’s advertising environment is nothing like the advertising environment of
just a few short years ago, as Dan Woods showed in his amusing comparison of
the differing environments that marketers face today as compared to what their
1980s counterparts faced.23
Right now, there is a radical realignment going on in the advertising industry. As
Derek Thompson points out in his The Atlantic article The Media’s Post-
Advertising Future Is Also Its Past181, it might be tempting to blame media’s
advertiser problem and the current state of its demise as the inevitable end
game of the Google and Facebook’s duopoly because the two companies already
receive more than half of all the dollars spent on digital advertising, as well as
command 90 percent of the growth in digital ad sales in 2017.182 However,
Thompson argues what’s happening in media right now is more complex. 181 He
sees the convergence of the following four trends181:
1. Too many players.
2. Not enough saviors.
3. No clear playbook.
4. Patrons with varying levels of beneficence.
It isn’t just Facebook and Google, Thompson states181, “just about every big tech
company is talking about selling ads, meaning that just about every big tech
company may become another competitor in the fight for advertising
revenue.”181
“Amazon’s ad business exploded in the past year; its growth exceeded that of
every other major tech company, including the duopoly,” notes Thompson. 181
Wanting to move beyond just selling people iPhones, Apple is shifting its growth
strategy to selling services not just iPhones.183 Meanwhile, “Microsoft will make
about $4 billion in advertising revenue this year, thanks to growth from LinkedIn

177
 ANDREW W. PEARSON

and Bing.”181 “AT&T is building an ad network to go along with its investment in

Time Warner’s content, and Roku, which sells equipment for streaming
television, is building ad tech,” adds Thompson.181
As Sara Fischer explains in her article The Next Big TV Tech Platform: Roku184,
“Roku, the connected TV hardware company, is quietly building a large software
business, driven mostly by advertising revenue.” Fischer adds: “Roku typically
doesn't sell advertising through an open exchange (open bidding system), like
some of the big tech companies do, but it does use programmatic infrastructure
to digitally target those ads — a tactic commonly referred to as ‘programmatic
direct’ or ‘programmatic reserved.’”184
At the Adobe Summit in March 2019, Adobe announced 185 a new partnership
with Roku that would help advertisers engage with Roku’s 27 million OTT
viewers. These kinds of deals, which allow businesses to directly connect with
consumers are the wave of the future. According to Adweek, “marketers can
now use elements of the Adobe Advertising Cloud to match their own audience
data with Roku’s in a way that provides an unprecedented degree of targeting
granularity for those eager to engage with the streaming provider’s 27 million
viewers.”185
“Keith Eadie, VP and general manager of Adobe Advertising Cloud, said the
partnership would enable advertisers to better manage elements of their cross-
screen campaigns such as frequency capping and that it would also help them to
better measure the outcomes of their media buys on OTT — the fastest-growing
channel on the Adobe media buying platform.” 185
Meanwhile, Scott Rosenberg, general manager, business platform, Roku, added,
“Programmatic trading is already a material part of our business, but it is still a
minority but some of that is a function of the fact that TV marketers are not by
and large as yet trading programmatically.” 185
This is because the majority of TV ad space is still traded manually, rather than
programmatically. However, Rosenberg explained to Adweek that “this is about
to change. ‘Programmatic is not the predominant methodology for TV
marketers, but it’s coming in strong … this partnership is so important because
one of the friction points, that holds programmatic trading of OTT back is scale.’”
The standard ad business methodology is also rapidly evolving. As Sara Fischer
points out in her Axios article How Media Companies Lost the Advertising
Business186, the great irony is that “many of the tech companies began with an
aversion to advertising, fearing it would be a disruption towards the consumer
experience.” Fischer notes that Google initially feared that advertising-based
search engines were “inherently biased towards the advertisers and away from
the needs of consumers.”186 “Facebook CEO Mark Zuckerberg reportedly only
accepted advertising on his platform initially so he could pay the bills.” 186
“Snapchat boss Evan Spiegel initially criticized some targeted ads as ‘creepy,’ but

178
 THE A.I. CRUISE LINE

four years later, 90% of the ads sold on Snap's platform are sold in an automated
fashion.”186
Today, most tech companies are embracing the advertising model in one form
or another. Fischer adds that, “Some publishers are banding together to offer
marketers to [sic] cheaper advertising against traditional media content at
scale.” Fischer’s examples include186:
• Several digital websites, such as Quartz, New York Media, PopSugar and
Rolling Stone are all joining Concert, a digital advertising marketplace
operator whose stated goal is to combat the tech giants' ad dominance.
• News Corp launched a global digital ad network in 2018 called News IQ,
which will pull audience data from sites like The Wall Street Journal,
New York Post and Barron's, as well as give advertisers a way to reach
highly specific audiences.
• AT&T is hoping to create a similar type of ad network through its Time
Warner partnership, with plans to bring on other media and technology
partners in time.
• Disney and Verizon are looking into building their own ad networks.
According to George P. Slefo, “Quartz won't be selling the sort of junk display ads
that are scattered across the web. Instead, the company offers only a single ad
unit that's both high impact and fits with the overall scheme of the website.” 187
Joy Robins, chief revenue officer at Quartz, admitted that, "We remain in a
position where we aren't in any available open marketplaces."187 Robins added
that “the company has also started programmatic advertising through closed,
private marketplace deals.”187 "We are doing both private and very specific deals
that are able to be traded using programmatic technology, but they have to be
direct and use our custom units, which we believe provide value," he says.187
Quartz “has an in-house team of 35 employees dubbed Quartz Creative —
separate from its sales team — that works with brands and advertisers in
creating ads beyond standard display units for its website.”187
Fischer adds that, “It's not just tech firms, but retail and consumer package goods
companies, too. Ad-serving has become so democratized that any company with
an audience is now able to steal advertising dollars away from traditional media
companies. Kroger has an ad business and so does its grocery rival
Albertsons. Target has a media network and so does Walmart.”186
Thompson notes that, “These tech companies have bigger audiences and more
data than just about any media company could ever hope for. The result is that
more advertising will gravitate not only toward ‘programmatic’ artificial-
intelligence-driven ad sales but also toward companies that aren’t principally (or
even remotely) in the news-gathering business.”181
In his Where Did All the Advertising Jobs Go? 188 Derek Thompson explains that,

179
 ANDREW W. PEARSON

“The emergence of an advertising duopoly has coincided with the rise of

‘programmatic advertising,’ a torpid term that essentially means ‘companies
using algorithms to buy and place ads in those little boxes all over the internet.’”
Thompson adds that, “advertising has long been a relationship-driven business,
in which multimillion-dollar contracts are hammered out over one-on-one
meetings, countless lunches, and even more-countless drinks. With
programmatic technology, however, companies can buy access to specific
audiences across several publishing platforms at once, bypassing the work of
building relationships with each one.”188 Because advertising has become more
automated, more ads can be produced with fewer people. 188 AI needs be a part
of this programmatic advertising process because the sheer volume of work
these processes entail would be overwhelming otherwise.
In her article Experts Weight in On the Future of Advertising 189, Giselle
Abramovich quotes Keith Eadie, VP and GM of Adobe Advertising Cloud, who
argues that, “Programmatic advertising is no longer a silo or a distinct media
channel — it’s simply how brands are buying ads.” “As a result,” Eadie says, “the
focus is shifting from execution to strategy and better connecting marketing and
advertising.”189
“With programmatic, the big lure for advertisers is its efficiency, according to
Amy Avery, Droga5’s chief intelligence officer.”189 “But it also needs to be about
effectiveness, too,” Avery argues.189 “I don’t think we will ever go to 100%
[programmatic ad buying]. But I do think it can increase to much more than it is
now once the effectiveness variables come into play.”189 Avery believes this will
happen once AI is integrated to help understand context and it uses this
information to inform messaging.189
Eadie says the $70 billion TV advertising market is a great example of progress
on this front.189 Specifically, NBCUniversal recently “made its full portfolio of
broadcast and cable television available to advertisers through a DSP, essentially
automating ad buying for its TV market.”189
“The old story about programmatic advertising was that both marketers and
digital publishers — think AOL, or any news site — embraced the technology, as
it allowed companies to cheaply target specific audiences on a budget,” explains
Thompson.188 However, Thompson notes that, “the new reality is that
programmatic advertising has placed many advertisements in controversial
places, next to low-quality news sources or outright offensive content.”188 This
has caused marketers to both cut back on running programmatic ad campaigns,
as well as bringing their operations in-house, where they have more control over
both who sees their ads and where they are seen. 188 “The upshot,” Thompson
contends is that, “Programmatic ads have been a double blow to media agencies,
first automating their function and then encouraging companies to insource the
work.”188

180
 THE A.I. CRUISE LINE

Becoming more self-sufficient may not be a bad thing for a cruise line marketers.
Software that does everything from automating marketing campaigns to
inexpensively segmenting customers, to simplifying the mundane and repetitive
processes of producing and categorizing content can help marketers speed up
the creative process enormously.
Thompson argues that currently there is a “merging of the advertising and
entertainment businesses.”188 “As smartphone screens have edged out TV as the
most important real estate for media, companies have invested more in
‘branded content’ — corporate-sponsored media, such as an article or video,
that resembles traditional entertainment more than it does traditional
advertising."188 Thompson concludes that, “In short, the future of the advertising
business is being moved to technology companies managing ad networks and
media companies making branded content — that is, away from the ad
agencies.”188 These are cross-currents that cruise line brands need to be aware
of because they are not just radically changing the marketing landscape but also
offering huge marketing opportunities for cruise lines willing to embrace and
exploit them.
If software solutions can alleviate the mundane and repetitive tasks humans are
currently toiling away at — and they most definitely can — then businesses can
redeploy their staff to handle the more interesting and probably more profitable
work, like programmatic functions.

Data & Marketing

The history of the methodical use of data in marketing begins in 1910, with the
work of Charles Coolidge Parlin for the Curtis Publishing Company in Boston. 190
As Wedel & Kannan explain in their article Marketing Analytics for Data-Rich
Environments191, “Parlin gathered information on markets to guide advertising
and other business practices, prompting several major U.S. companies to
establish commercial research departments.”191
Questionnaire survey research, which Gallup popularized in the 1820s with its
opinion polling, became increasingly common in the 1920s.192 At about the same
time, “concepts from psychology were being brought into marketing to foster
greater understanding of the consumer,” explain Wedel and Kannan. 191 Today,
psychology has taken an oversized role in the marketing process, as will be
shown throughout this chapter and book. Starch’s attention, interest, desire,
action (AIDA) model193 is an example of this. Starch is widely considered to be
one of the pioneers of marketing and consumer research. This is also the era
when eye-tracking technology debuted, including the ability to collect data that
followed the movements of the eye. 194 It is also a technology that is making a
strong comeback almost a century later.

181
 ANDREW W. PEARSON

“In 1923, A.C. Nielsen founded one of the first market research companies.
Nielsen started by measuring product sales in stores, and in the 1930s and 1950s,
he began assessing radio and television audiences. ”191 Today, Nielsen is a
household name in the United States and it dominates TV ratings in the US and
several other countries. Recognizing that social is becoming an important
channel, Nielsen has also moved into social media measurement now.
Beginning in the late 1970s, geo-demographic data was collected from
government databases and credit agencies by the market research firm
Claritas.191 “The introduction of the Universal Product Code and IBM’s
computerized point-of-sale scanning devices in food retailing in 1972 marked the
first automated capture of data by cruise lines.”191
Companies like Nielsen “quickly recognized the promise of using point-of-sale
scanner data for research purposes and replaced bimonthly store audits with
more granular scanner data,” notes Wedel and Kannan.191 Shortly after the start
of the data collection process, individual customers could be traced through
their loyalty cards use, which led to the emergence of scanner panel data. 195
The introduction of IBM’s personal computer in 1981 enabled the collection of
customer data on a massive scale.191 Personal computers allowed marketers to
store data on current and potential customers 191, contributing to the emergence
of database marketing, which was pioneered by Robert and Kate Kestenbaum
and Robert Shaw.196
“In 1990, CRM software emerged, for which earlier work on sales force
automation at Siebel Systems paved the way.”191 Personal computers simplified
survey research through personal and telephone interviewing. 191
In 1995, after more than two decades of development at the Defense Advanced
Research Projects Agency and several American universities, the internet was
born, and this meant large volumes of marketing data were suddenly
accessible.191
Clickstream data extracted from server logs allowed businesses to track page
views and website clicks using cookies.191 Click-through data revealed the true
effectiveness of online advertising.191 “The Internet stimulated the development
of CRM systems by firms such as Oracle, and in 1999 Salesforce was the first
company to deliver CRM systems through cloud computing,” state Wedel and
Kannan.191
Founded in 1998, Google championed keyword search and the capture of search
data.191 Google emerged from the highly competitive 1990s search environment,
beating out the likes of Alta Vista, Yahoo!, Infoseek, and Lycos.
The launch of Facebook in 2004 opened up an era of social network data and it
quickly eclipsed MySpace as the dominant social network. 191 The arrival of user-
generated content (UGC), including pictures, online product reviews, blogs, and

182
 THE A.I. CRUISE LINE

videos, resulted in an explosion in the volume and variety of data. 191

“With the advent of YouTube in 2005, vast amounts of data in the form of user-
uploaded text and video became the raw material for behavioral targeting,”
explains Wedel and Kannan.191 Twitter, with its much simpler 140-character
messages, appeared in 2006. While the social network, blogging, and micro-
blogging scene solidified in the early 2000s, another important step for
marketing measurement appeared in 2007 when Apple introduced the iPhone.
With its global positioning system (GPS) capabilities, the first iPhone meant one
could capture consumer location data at an unprecedented rate. 191

Analytics
The initiative of the Ford Foundation and the Harvard Institute of Basic
Mathematics for Applications in Business in late 1950s and early 1960s is widely
credited for providing the catalyst that introduced analytics into marketing. 197 By
then, statistical methods, such as analysis of variance, had been utilized in
marketing research for more than a decade198, but the development of statistical
and econometric models tailored to specific marketing problems only took off
“when marketing was recognized as a field of decision making through the
Ford/Harvard initiative.”199
The development of Bayesian decision theory at the Harvard Institute 200 also
played a key role, demonstrated by its successful application to, among other
things, pricing decisions.201 Academic research in marketing then started
focusing more on the development of statistical models and predictive
analytics.191
New product diffusion models202 involved applications of differential equations
from epidemiology. Stochastic models of buyer behavior 203 were “rooted in
statistics and involved distributional assumptions on measures of consumers’
purchase behavior,” argue Wedel and Kannan.191
The application of decision calculus204 205 to optimize spending on advertising
and the sales force became popular after its introduction to marketing by John
Little in his Models and Managers: The Concept of a Decision Calculus 206.
Nakanishi and Cooper introduced market share and demand models for store-
level scanner data in 1974, which were derived from econometric models of
demand.207
According to Wedel and Kannan, multidimensional scaling and unfolding
techniques, founded in psychometrics208, also became an active area of
research.191 “These techniques paved the way for market structure and product
positioning research by deriving spatial maps from proximity and preference
judgments and choice,” contend Wedel and Kannan.191
Conjoint analysis209 and, later, conjoint choice analysis210 are unique

183
 ANDREW W. PEARSON

contributions that evolved from work in psychometrics by Luce on the

quantification of psychological attributes.211 Also, “The nested logit model that
captures hierarchical consumer decision making, i.e., understanding the factors
that influence the way a consumer shops, was introduced in marketing 212, and it
recognized that models of multiple aspects of consumer behavior (e.g.,
incidence, choice, timing, quantity) could be integrated 213 into the marketing
mix. This proved to be a powerful insight for models of recency, frequency, and
monetary (RFM) metrics214, which is the method for analyzing customer value by
looking at how recently someone has purchased an item, how often they
purchase and how much they spend. Time-series methods215 can help cruise
lines forecast sales, project yields and workloads, analyze budgets, as well as
enable researchers to test whether marketing instruments result in permanent
or transient changes in sales.
In their paper A Probabilistic Choice Model for Market Segmentation and
Elasticity Structure216, Kamakura and Russell state that heterogeneity in the
behaviors of individual consumers becomes a core premise on which marketing
strategy should be based, and the mixture choice model is the first to enable
managers to identify response-based consumer segments from scanner data.216
Wedel and DeSarbo expounded upon this, arguing that the model should be
generalized to accommodate a wide range of models of consumer behavior. 217
Rossi, McCulloch and Allenby concluded that consumer heterogeneity was
represented in a continuous fashion in hierarchical Bayes models. 218
Although scholars have hotly debated which of these two approaches best
represents heterogeneity, research has revealed that the two different
approaches each match specific types of marketing problems, with few
differences between them. 219 Today, it can safely be stated that the Bayesian
approach is one of the most dominant modeling approaches in marketing,
offering a powerful framework to develop integrated models of consumer
behavior.220 Bayesian models have been successfully applied to advertisement
eye tracking221, e-mail marketing222, web browsing223, social networks224, and
paid search advertising.225
Data-driven analytics in marketing has progressed from its inception around
1900 up to the introduction of the Internet in 1995 through approximately three
stages:
1. “The description of observable market conditions through simple
statistical approaches.
2. “The development of models to provide insights and diagnostics using
theories from economics and psychology.
3. “The evaluation of marketing policies, in which their effects are
predicted and marketing decision making is supported using statistical,
econometric, and OR approaches.”191

184
 THE A.I. CRUISE LINE

In many cases, throughout the history of marketing analytics, once new sources
of data get introduced, methods to analyze them are immediately developed.
Figure 19 contains an outline of the history of data and analytical methods.
“Many of the methods developed by marketing academics since the 1960s have
now found their way into practice and support decision making in areas such as
CRM, marketing mix, and personalization and have increased the financial
performance of the firms deploying them,” note Wedel and Kannan.191

Figure 19: Timeline of Marketing Data and Analytics

Source: Marketing Analytics for Data-Rich Environments191

“Since 2000, the automated capture of online clickstream, messaging, word-of-

mouth (WOM), transaction, and location data has greatly reduced the variable
cost of data collection and has resulted in unprecedented volumes of data that
provide insights on consumer behavior at exceptional levels of depth and
granularity,” explain Wedel and Kannan.191

Although academics have risen to the challenge of developing diagnostic and

predictive models for the variety and velocity of data, we’ve seen over the last
decade, these developments are admittedly still in their infancy. 191
On the one hand, descriptive metrics displayed on dashboards are popular in
practice.191 Perhaps because of “constraints on computing power, a need for
rapid real-time insights, a lack of trained analysts, and/or the presence of
organizational barriers to implementing advanced analytics.” 191 In particular,
unstructured data in the form of blogs, reviews, and tweets offer opportunities
for deep insights into the economics and psychology of consumer behavior,
which could usher in the second stage in digital marketing analytics once
appropriate models are developed and applied,” argue Wedel and Kannan. 191
On the other hand, machine learning methods have become popular in practice,

185
 ANDREW W. PEARSON

but have been infrequently researched in marketing academia. 191 “It is

reasonable to expect that the third step in the evolution of analytics in the digital
economy — the development of models to generate diagnostic insights and
support real-time decisions from big data — is imminent,” contend Wedel and
Kannan.191

Mobile Marketing
If an advertising executive had set about to create the perfect marketing and
advertising tool, she could hardly have created something more superior to the
mobile phone. Not only is the mobile phone within reach of its owner almost
every single hour of every single day but, because it can connect to a marketer
in a highly personalized way with the simple touch of a button, it has the
potential to become not only more effective than television or radio advertising
but, just as importantly, more analyzable.
As the authors of Mobile Advertising226 point out that, “With respect to targeting,
no other medium can provide the accurate and rich user profile, psychographic,
social engagement and demographic data available from mobile. No other
medium has the viral capability that mobile possesses — within seconds
following a simple click, a unit of advertisement can spread like wildfire.”
No other media comes even remotely close to the data measurement capacity
that mobile offers either, which begins with exposure to the advertisement,
followed by the persuasive effect of the advertisement and, finally, to the actual
purchase of a product.226 Just about every link in the marketer’s chain is touched
by mobile.
In 1996, the Internet advertising landscape changed forever when Procter &
Gamble convinced Yahoo! that it would only pay for ads on a cost-per-click basis,
rather than for banner ads226 Procter & Gamble realized the importance of
gaining truthful user metrics for internet advertising and this move ushered in
the world of internet analytics; eyeballs were no longer the goal, click-thrus that
showed actual product interest became paramount.
As Sharma et al. state in their book Mobile Advertising226 the time is right for
mobile marketing because “the heavy lifting of measurements and metrics; of
banner ad standards; of search keyword auctions; of advertising cost models and
the new, digital ad networks that support them have been built. The groundwork
for digital advertising in mobile is largely in place.” However, because there are
so many players involved, the mobile advertising value chain is incredibly
complicated.226
As Sharma et al. explain, “the mobile value chain comprises advertisers,
agencies, solution providers and enablers, content publishers, operators and
consumers. Phone manufacturers or original equipment manufacturers (OEMs)

186
 THE A.I. CRUISE LINE

are enablers in this value chain rather than active participants.” 226 The
bottleneck in the chain arises because, even though there are only a limited
number of mobile operators, the number of vendors in the value chain is
exceedingly high.226 Although this was written almost a decade ago, the
complexity of the advertising environment still remains and it is something that
must be kept in mind when developing mobile marketing campaigns.
In their article The Typological Classification of the Participants’ Subjectivity to
Plan the Policy and Strategy for the Smart Mobile Market227, Kim et al. argue that
the core technologies of cloud computing can greatly enhance mobile marketing
efforts. Without cloud computing, it would be impossible to successfully produce
targeting context-aware ads, real-time LBS ads, interactive-rich media ads,
mobile semantic webs or in-app ads, advanced banner ads or incentive-based
coupon ads, AR or QR codes, social network ads, and n-screen ads.227 It would be
especially difficult integrating and converging multifunctional mash-up ads
involving a mix of the aforementioned.227 “Smart mobile advertising products
continuously derive combined services where two or more advertising
techniques integrate and interlock due to innovative hardware or software
technologies.”227
Mobile advertising has the potential to give cruise lines the best bang for their
marketing buck, but a mobile marketing campaign should not simply be viewed
as an extension of a company’s internet marketing brought to the mobile phone.
In Mobile Advertising, the authors state that the three basic types of mobile
advertisements are226:
• Broad-based brand advertising: broad-based campaigns that take
advantage of user filtering and targeting. These can include subsidized
premium content, sponsorships, video pre-rolls or intromercials, post-
roll video, on-demand mobile media and contextual or behavioral
advertising.
• Interactive, direct response campaigns: these are opt-in campaigns in
which the mobile user usually exchanges some personal information for
some type of content. TXT short codes, mobile subscription portals, and
user registration campaigns are all examples of this type of campaign.
• Highly targeted search advertising: mobile’s ability to inform advertiser
of the user’s basic age, sex, and address information is far better than
any other form of advertising around. These campaigns include content
targeted search advertising and paid placement or paid inclusion
search.
Although there were hints that a marketing revolution was underway at the
beginning of the 21st Century, few people would have predicted the radical
changes that have transformed the industry today. In their article Interactivitys
Unanticipated Consequences for Marketers and Marketing 228, Deighton and
Kornfeld argue that:

187
 ANDREW W. PEARSON

“Mass communication technology empowered marketers with

marketer-to-consumer tools such as radio, television and
database-driven direct marketing. The digital innovations of the
last decade made it effortless, indeed second nature, for
audiences to talk back and talk to each other. They gave us peer-
to-peer tools like Napster, eBay, TiVo, MySpace, YouTube,
Facebook, Craigslist and blogs, and information search tools like
Google and Wikipedia. Mobile platforms have given us
ubiquitous connectivity, context-aware search, and the ability to
tag and annotate physical spaces with digital information that
can be retrieved by others. In sum, new traffic lanes were being
built, not for the convenience of marketers, but for consumers.”
Successful marketing is about reaching a consumer with an interesting offer
when he or she is primed to accept it. Knowing what might interest the consumer
is half the battle to making the sale and this is where customer analytics comes
in.
Howard Luck Gossage was probably onto a marketing truism when he stated,
“The real fact of the matter is that nobody reads ads. People read what interests
them, and sometimes it’s an ad.”229 The great advertising maven David Ogilvy
would agree – “What really decides consumers to buy or not to buy is the
content of your advertising, not its form.”230 It is the cruise line’s marketing
department’s duty to use customer analytics to glean as much information as it
can about its customers and then use this information about the customer’s
interests and behavior to devise a marketing message that captures the
imagination of its customers. If they can do this skillfully, an increase in sales
should quickly follow.
Customer analytics have evolved from simply reporting customer behavior to
segmenting customers based on their profitability, to predicting that
profitability, to improving those predictions (because of the inclusion of new
data), to actually manipulating customer behavior with target-specific
promotional offers and marketing campaigns.
Data must be gathered from disparate sources and seamlessly integrated into a
data warehouse that can then cleanse it and make it ready for consumption. 231
Trends that surface from the data mining process can help in monetization, as
well as in future advertising and service planning. 226 As the authors’ state in
Mobile Advertising226:
“The analytical system must have the capability to digest all the
user data, summarize it, and update the master user profile.
This functionality is essential to provide the rich user
segmentation that is at the heart of recommendations,
campaign and offer management, and advertisements. The

188
 THE A.I. CRUISE LINE

segmentation engine can cluster users into affinities and

different groups based on geographic, demographic or socio-
economic, psychographic, and behavioral characteristics.”
Of course, with all of this data collection comes justified privacy concerns and
the most important aspect of mobile marketing is ensuring the consumer has
control of the advertising.226 Without this, it is doubtful mobile marketing will
reach its true potential.226 If mobile advertisers do allow users to configure and
control the ads depending on where they are, what mood they are in, who they
are with, and what their current needs and desires happen to be, mobile
marketing could prove to be one of the most successful forms of advertising
available to cruise line marketers ever devised.226
The potential to market to an individual when she is primed to accept the
advertising is advantageous for both parties involved. Cruise line marketers
won’t waste time advertising to consumers when they aren’t primed to accept
the advertisements but do market to consumers when and where they might
want to use the advertisements.

Digital Interactive Marketing: The Five Paradigms

In their article Interactivity’s Unanticpated Consequences for Marketers and
Marketing228, Deighton and Kornfeld write that in this new media environment,
there are five emerging marketing paradigms that are responses to the decrease
of marketing’s power relative to the consumer. Digital interactive marketing has
little use for words such as “viewer” and “listener”.228 Even the label “consumer”
is of limited value because today’s interactions with a person will include
encounters that have nothing to do with consuming or being part of a “target
market.” Deighton and Kornfeld see this new digital interactive marketing
breaking down into five different paradigms228, as per Table 8.
Today, when a user searches for information or entertainment on sites such as
Google, she leaves a trail (also known as a “clickstream”) that reveals what is on
her mind.228 This information, which Deighton and Kornfeld refer to as “thought
tracing”, may be “available to marketers in exactly the sense that it is available
to marketers through Google, as a clue to our thoughts, goals and feelings.” 228
Mobile and social media alter the marketing landscape because the ubiquitous
nature of computing makes it an “always on” proposition; both the thought and
the activity are being traced.228 “The argument is that when a person is always
connected to the Internet, the person is always in the market, always available
to be communicated with, and always an audience” contend Deighton and
Kornfeld.228
Of course, most people don’t like to be marketed to continuously throughout
the day so technology that allows people to filter out messages that don’t
interest them needs to be developed.228 However, customized marketing

189
 ANDREW W. PEARSON

messages will be allowed to get through. Just as television demands its audience
to sit through commercials in order to enjoy free programming, Deighton and
Kornfeld contend that, “we will enjoy ubiquitous computer connectivity for the
price of voluntary exposure to context-specific persuasion efforts.”228

Interactive How people use How firms interpose

Resulting digital
marketing interactive themselves to pursue
media markets
paradigm technology marketing goals
Firms infer states of
People search the web
mind from search terms A market in
Thought for information and
and Web page content search terms
tracing browse for
and serve relevant develops.
entertainment.
advertising.
Firms exploit A market in
People integrate
Activity information on access and
always-on computing
tracing proximity and identity
into everyday life.
pertinence to intrude. develops.
Firms compete with A market in
People participate in these exchanges, service and
Property
anonymous exchanges rather than reputation and
exchanges
of goods and services. participating with reliability
them. develops.
A market in
community
People build identities
Social Firms sponsor or co-opt develops,
within virtual
exchanges communities. competing on
communities.
functionality and
status.
People observe and
Firms offer cultural
Cultural participate in cultural Firms compete in
products or sponsor
exchanges production and buzz markets.
their production.
exchange.

Table 8: Digital Interactive Marketing: Five Paradigms

Source: Journal of Interactive Marketing 228, 23 pg. 4-10

If businesses want to succeed in this new marketing environment they must

become an ally to the marketed individual, someone who is actually sought out
as a person with cultural capital.228 “Property exchanges”, “social exchanges”
and “cultural exchanges” are all paradigms that are “built on peer-to-peer
interactivity motivated by the desire to exchange, to share information, or to
express one’s self,” state Deighton and Kornfeld.228
Arguably, internet property exchanges were introduced on a mass scale by
Napster, which was the first company to allow users to share and exchange files

190
 THE A.I. CRUISE LINE

in an anonymous way.228 Unsurprisingly, Napster ran into trouble with copyright

holders and quickly left the content exchange business, but sites such as eBay,
Flicker and YouTube allow users to share and even sell their property over the
Internet. This is a trend that is probably never going away.
While the property exchange deals in things, the social exchange deals in
identities and reputations.228 In general, social networking sites let a person
present a face to the world, “including information about whereabouts and
action and a ‘wall’ on which friends can post short, often time-sensitive notes,
allows people to exchange digital gifts, provides a marketplace for buying and
selling, and allows posting of photographs and video clips.” 228
These sites allow for contextually relevant advertising because friends can share
information amongst each other and some of this information can include a
marketer’s message. Since this messaging is coming from a trusted source, the
message is considered much more trustworthy and enticing and, therefore,
much more likely to be acted upon. For example, “a recent Nielsen analysis of 79
campaigns on Facebook over six months showed that, on average, social ads —
those that are served to users who have friends that are fans of or have
interacted with the advertised brand and prominently call the relationship out
— generate a 55 percent greater lift in ad recall than non-social ads.”232
One of the key criteria of mobile marketing is that a consumer must opt-in to the
service. Mobile marketing is primarily a “pull” media model, meaning a
consumer must sign up for the service rather than the traditional “push” media
model, which gives the consumer no choice in whether they want to be
marketed to or not.
Cruise line mobile marketers must spend money to get users to sign up, but, if
they do, the potential market for mobile marketing is huge. It is also a market
that is rapidly evolving, and its advantages include:
• Ubiquity: mobile devices and their users are everywhere.
• Effective: over 90% of received text messages are read by the recipient.
• Powerful two-way dialogue: an instantaneous link between the
business and its customer is created.
• Economical: compared to other marketing channels, mobile marketing
is incredibly cheap per marketed individual.
• Spam-free: in the U.S. (but not in many other parts of the world) it is
illegal to send a text message to someone who hasn’t opted-in to a
marketing campaign.
In her book, The Mobile Marketing Handbook233, Kim Dushinski lists eight types
of advertising campaigns that a mobile marketer can engage in:
1. Voice: this includes text-to-call messages in which users are sent a link
that, when clicked upon, initiates a phone call to the company sending

191
 ANDREW W. PEARSON

out the message. These days, Apple’s SIRI, Microsoft’s Tellme and
Google’s Now are adding a whole new dimension to voice.
2. Text messaging: this used to be the “now” marketing tool of mobile,
and it is still one of the most important tools available. Text messaging
includes both SMS and Common Short Codes (CSC), which are
abbreviated phone numbers. Text messages are sent to mobile users,
the content of which are limited only by SMS character limitations and
the marketer’s overall imagination.
3. Mobile web: most smart phones have the ability to connect to the web
and many of them have graphic capabilities that rival computer screens.
4. Mobile search: as previously discussed, a mobile user can search
company listings through his or her mobile phone, just as he or she can
find this information on the Internet.
5. Mobile advertising: placing banner ads and text ads on mobile websites
can build brand awareness.
6. Mobile publicity: presenting a company’s executives as experts in his or
her field can be useful to members of the media who need instant
information for fast approaching deadlines.
7. Social networking: done right, this can help marketers tap into word-of-
mouth campaigns, which will, hopefully, have their marketing messages
lighting up social media websites.
8. Proximity marketing: Bluetooth and geofencing campaigns that invite
users to accept a multimedia message can deliver unique and location-
specific marketing messages.
To these eight, I would add another two — OTT and mobile apps marketing —
and I will break each of these campaigns down throughout the rest of this book.

Website Morphing
It is all well and good to offer personalized service to customers face-to-face, but
what happens when a customer visits a brand’s website for the first time, or even
the hundredth time? Today, personalized web pages can be rendered during the
web page load and elements of the page can take into account past purchase
history, clickstream behavior, as well as a whole host of other data points. For a
marketer, their website can really be a powerful customer center.
In her article The Art and Science Behind Every “Add to Cart” 234, Christie Chew
argues that, “Neuroscience and the way people make decisions impact what
compels people to click and buy. Together, these considerations and best
practices can work together to drive customers to take action.”
Guliz Sicotte, head of product design and content for Magento, says to prompt
a customer purchase, brands must create online experiences that focus on four
principal characteristics — they must be personalized, reflective, transparent,

192
 THE A.I. CRUISE LINE

and use pleasing aesthetics.234

Morphing is one of the ways a brand can hyper-personalize the customer
shopping experience. So, what exactly is morphing? In their article Website
Morphing235, Hauser et al. state that, “’Morphing’ involves automatically
matching the basic ‘look and feel’ of a website, not just the content, to cognitive
styles.” Hauser et al. use Bayesian updating to “infer cognitive styles from
clickstream data.”235 Then they “balance exploration (learning how morphing
affects purchase probabilities) with exploitation (maximizing short-term sales)
by solving a dynamic program (partially observable Markov decision process).”235
In a world of deep personalization, website design becomes a major profit
driver.235 As Hauser et al. see it, “Websites that match the preferences and
information needs of visitors are efficient; those that do not forego potential
profit and may be driven from the market.” 235 The authors believe that
businesses “might serve their customers better and sell more products and
services if their websites matched the cognitive styles of their visitors.”235
Keeping with the themes of simplicity and seamlessness, Hauser et al. do not
believe personal self-selection — the process in which a customer is given many
options and allowed to select how to navigate and interact with the site — is
viable.235 “As the customer’s options grow, this strategy leads to sites that are
complex, confusing, and difficult to use,” they argue.235 The second option,
which requires “visitors to complete a set of cognitive style tasks and then select
a website from a predetermined set of websites”235 is just as problematic.
Website visitors probably won’t see value in taking the time to answer these
questions and there is always the problem of self-bias hindering any potential
results.235
Hauser et al. propose another approach: “’morphing’ the website automatically
by matching website characteristics to customers’ cognitive styles.” 235 A
cognitive style is “a person’s preferred way of gathering, processing, and
evaluating information.”236 It can be identified as “individual differences in how
we perceive, think, solve problems, learn and relate to others.”237 “A person’s
cognitive style is fixed early on in life and is thought to be deeply pervasive [and
is] a relatively fixed aspect of learning performance.” 238
The “goal is to morph the website’s basic structure (site backbone) and other
functional characteristics in real time.”235 “Website morphing complements self-
selected branching (as in http://www.Dell.com), recommendations (as in
http://www.Amazon.com), factorial experiments (Google’s Website Optimizer),
or customized content.239 240”235
For Hauser et al., cognitive styles dimensions “might include impulsive (makes
decisions quickly) versus deliberative (explores options in depth before making
a decision), visual (prefers images) versus verbal (prefers text and numbers), or
analytic (wants all details) versus holistic (just the bottom line).”235 For example,

193
 ANDREW W. PEARSON

“a website might morph by changing the ratio of graphs and pictures to text, by
reducing a display to just a few options (broadband service plans), or by carefully
selecting the amount of information presented about each plan. A website might
also morph by adding or deleting functional characteristics such as column
headings, links, tools, persona, and dialogue boxes.” 235 There are, literally,
hundreds of thousands or even millions of ways a website can morph to better
serve its customers.
Because of its real-time nature, website morphing is not easy. It presents at least
the following four technical challenges235:
1. The customer acquisition problem, i.e., the website must morph based
on relatively few clicks of a first-time visitor; otherwise, the customer
sees little benefit.
2. Even knowing a customer’s cognitive style is not enough, the website
must learn which characteristics are best for which customers (in terms
of sales or profit).
3. To be practical, a system needs prior distributions on parameters.
4. Implementation requires a real-time working system, which is one of
the most complex systems to set up, run, and maintain.
For their website morphing, Hauser et al. used:235
“a Bayesian learning system to address the rapid assessment
of cognitive styles and a dynamic program to optimally
manage the tension between exploitation (serving the morph
most likely to be best for a customer) and exploration (serving
alternative morphs to learn which morph is best). Uncertainty
in customer styles implies a partially observable Markov
decision process (POMDP), which we address with fast
heuristics that are close to optimal. Surveys, using both
conjoint analysis and experimentation, provide priors and
‘prime’ the Bayesian and dynamic programming engines. We
demonstrate feasibility and potential profit increases with an
experimental website developed for the BT Group to sell
broadband service in Great Britain.”
Hauser et al. expect different morphs to appeal differentially depending on the
visitors’ cognitive style.235 “For example, impulsive visitors might prefer less-
detailed information, whereas deliberative visitors might prefer more
information. Similarly, the more focused of the two morphs might appeal to
visitors who are holistic, while the ability to compare many plans in a table might
appeal to analytic visitors.”235 If preferences match behavior, then, by matching
a website’s characteristics to a customer’s cognitive style, the morphing website
should be able to sell more effectively, thereby producing greater profits for the
brand.

194
 THE A.I. CRUISE LINE

Hauser et al. applied a “Bayesian updating and dynamic programming to an

experimental BT Group (formerly British Telecom) website using data from 835
priming respondents.”235 The challenge was to infer the cognitive-segment to
which each visitor belonged, “while simultaneously learning how to maximize
profit by assigning morphs to cognitive-style segments.”235
Web visitor cognitive style segments are inferred from their clickstreams.235 This
was possible “because each visitor’s click is a decision point that reveals the
visitor’s cognitive-style preferences.”235 Hauser et al. believe that with enough
observed clicks, they could have been able to identify a visitor’s cognitive-style
segment quite conclusively.235 However, in any real application, the number of
clicks observed before morphing would be quite small, yielding at best a noisy
indicator of segment membership.235
Hauser et al. observed about ten clicks, inferred probabilities for the visitor’s
cognitive-style segment, then morphed the website based on their inference of
the visitor’s segment.235 “The visitor continued until he or she purchased a BT
broadband service or left the website without purchasing.” 235
In most cases, cognitive styles are measured with methods that “include direct
classification, neuro-fuzzy logic, decision trees, multilayer perceptrons, Bayesian
networks, and judgment.”235 Hauser et al. acknowledge that, while most authors
match the learning or search environment based on judgment by an expert
pedagogue or based on predefined distance measures, they inferred cognitive
styles from a relatively small set of clicks, then automatically balanced
exploration and exploitation to select the most appropriate morph. 235
To set a baseline cognitive style standard Hauser et al. used “a professional
market research company (Applied Marketing Science, Inc.) and a respected
British online panel (Research Now).”235 They invited “current and potential
broadband users to complete an online questionnaire that combined BT’s
experimental website with a series of preference and cognitive style
questions.”235
835 respondents completed the questionnaire, which contained the following
sequential sections235:
• Identify whether respondent was in the target market.
• Identify which of 16 broadband providers they might be considering,
along with purchase-intention probabilities.
• Eight randomly assigned potential morphs of the BT website. Each
respondent was encouraged to spend at least five minutes on BT’s
experimental website.
• Post-visit response consideration and purchase-intention probabilities.
• Identify their preferences between eight pairs of websites, with a
choice-based conjoint analysis-like exercise. These data augment
clickstream data when estimating.

195
 ANDREW W. PEARSON

• A cognitive style measure.

Hauser et al. expected “these scales to identify whether the respondent was
analytic or holistic, impulsive or deliberative, visual or verbal, and a leader or a
follower.”235 “The analytic versus holistic dimension is widely studied in
psychology and viewed as being a major differentiator of how individuals
organize and process information,”235 including by Riding and Rayner241, Allinson
and Hayes242, Kirton243, and Riding and Cheema.244 According to Hauser et al.,
“Researchers in both psychology and marketing suggest that cognitive styles can
be further differentiated as either impulsive or deliberative. 245 246”235
In summary, Hauser et al. identified the following four empirical constructs to
measure respondents’ cognitive styles235:
• Leader versus follower.
• Analytic/visual versus holistic/verbal.
• Impulsive versus deliberative.
• (Active) reader versus (passive) listener.
In conclusion, Hauser et al. “used segments of cognitive styles rather than
continuously defined cognitive styles because the dynamic program requires
finitely many ‘arms.’”235 Websites were morphed once per visit, in part, because
Hauser et al. observed a single subscription decision per customer.235
In her article The Art and Science Behind Every “Add to Cart”, Christie Chew notes
that the central question driving most purchases is, “What’s in it for me?” 234
“Customers should feel that products are relevant to their intentions,” adds
Guliz.234 “This sense of relevance can be traced back to what Carmen Simon,
Ph.D., cognitive neuroscientist at Memzy, describes as habitual decision-making
— habits are conscious at first but eventually become subconscious,” adds
Guliz.234
“Link your techniques, content, value proposition, or whatever you’re offering,
to something that feels familiar to the customer’s brain,” says Simon. 234 This
increases a person’s comfort level, which makes them more likely to take a
favorable action for your brand because what you’re asking them will feel
easy.234
Guliz concurs: “In the end, customers are faced with a barrage of e-commerce
opportunities. Expedite the shopping experience and increase conversions by
identifying products that ‘people like me’ have purchased. Once I can vet a
product based on people who closely match my profile, I am that much closer to
feeling comfortable in making the purchase.”234
“It’s also essential that each step in your e-commerce experience reflects
intention,” says Chew.234 “For example, the category page should include
curiosity-triggering components,” adds Guliz.234 “If you’re displaying an array of
products online, make it easy to determine the sentiment around each, without

196
 THE A.I. CRUISE LINE

customers needing to invest time to dive into each.”234 “From here, be sure to
show customers the most important details and features relevant to them and
their purchase experience,” recommends Chew.234
“This process can help create a series of clicks that drive those customers closer
to making a purchase,” says Chew.234 “Don’t pitch immediately. Don’t make
people think too hard. Work toward a series of smaller, more habitual ‘yeses,’”
advises Simon.234 “This creates more momentum and a pattern of ‘yes,’ which
can make a customer more comfortable with a bigger, riskier purchase decision,”
she says.234
For example, if you’re promoting high-end travel, query users about their overall
travel experience, rather than asking them outright if they are booking travel for
a vacation or business.234 These kinds of questions will likely elicit positive
responses, whereas direct questions can often be off-putting.234
Posing questions about a customer’s travel habits and preferences can lead
potential customers to the critical “yes” — and they might at least consider
booking a luxury vacation.234 By that point, Chew argues, “they’ve been
habituated toward a positive response and will be more open to bigger
considerations — and bigger purchases.”234
According to Carmen, “the brain makes decisions in a reflexive, habitual, and/or
goal-oriented way.”234 “The mistake some businesses make is asking people to
tap into their goals too quickly, at the expense of tapping into reflexes and habits
first,” argues Chew.234 She recommends brands, “Create opportunities for the
buyer to take small steps first, toward a larger goal or purchase.” 234
Transparency is essential to ensure positive customer experiences that will drive
customers toward a purchase.234 “It’s important to bring high visibility to the
critical decision-making factors like return policies and shipping times, by writing
them in clear, simple ways,” Guliz says.234 “If an array of products is displayed,
make it easy to determine the sentiment around these products without needing
to invest time to dive into each product offering.”234
“This type of layout will play well to a customer’s need to feel like they’re in
control of their environment,” says Chew.234 Guliz adds that brands should
ensure users can “easily navigate to different aspects of the product page.” 234
“Take them to reviews when they click on star ratings. Let them filter product
reviews.”234
“Retailers are making people think way too much,” Simon says. 234 He adds that,
“If you start with something that feels familiar and habitual, you’ll have an easier
time when it comes to persuasion. Show customers something that doesn’t
require a lot of cognitive energy to process.”234 What is true for retailers is
certainly true for cruise lines; when it comes to simplifying the buying process
probably more so for cruise lines.

197
 ANDREW W. PEARSON

“Humans are innately emotional — we react to everything from people to

environments to colors and sounds, based on our existing and real-time
experiences. This insight can help e-commerce brands better structure their
retail experiences,” argues Guliz.234 The same is true for cruise lines. “The right
aesthetics are major elements of trustworthiness,” explains Guliz.234 “Lots of
detailed photos of key features is crucial to a good experience.” 234
Simon adds that “there are a series of innate behaviors in which you already
know what to do next. In the buying process, that includes your reflex toward
something beautiful.”234
Marcia Flicker, Ph.D., associate professor at Fordham University’s Gabelli School
of Business, argues that, “Creating this aesthetic experience requires having the
right visuals.”234 E-commerce brands need bigger and more detailed photos,
especially apparel brands.234 “It can be hard for customers to buy apparel online
because they want to try it on, feel the fabric,” contends Flicker.234 “Retailers
need to reproduce that experience of being able to see the actual item.
Customers need to see large images from a variety of angles — or even video,”
advises Flicker.234
“This aesthetics-focused notion can, then, be woven into an e-commerce brand’s
UX design to help pave a customer’s path to purchase,” says Guliz.234 “When you
design an interface, you’re more likely to have people use it if it’s aesthetically
appealing,” Simon says.234 “Principles like proximity, balance, unity, and contrast
are important to this notion of aesthetics.”234 Cruise lines would also profit from
a liberal use of aesthetically pleasing visuals.
“Understanding why customers buy and designing experiences to match their
patterns is just the beginning. Going forward, brands will continue to fine-tune
these strategies, layering in more future-forward technologies,” predicts
Guliz.234 “Augmented reality could be a game-changer,” says Flicker.234
However, Guliz concludes that, “even with the most cutting-edge understanding
of what makes us tick — and click — and stunning aesthetics and powerful UX
design, none of it matters if the experience doesn’t fill a need in the customer’s
purchase path.”234 Experience is both the fallback and the reason for purchases.

Psychology of Personalization
So, what does “personalization” really mean? It is a word that has been kicking
around the marketing community for at least a decade or two now. The
underlying psychology of the individual being marketed to is one of the key
elements of personalization marketing. In his Buffer article 15 Psychological
Studies That Will Boost Your Social Media Marketing 247, Kevan Lee lists several
psychological techniques that marketers should be using to reach today’s
audience. Lee’s list is as follows247:

198
 THE A.I. CRUISE LINE

1. The endowment effect — the hypothesis that people ascribe more

value to things merely because they own them.248
2. Reciprocity — in social psychology, reciprocity is a social norm of
responding to a positive action with another positive action, rewarding
kind actions.249
3. Consistency principle — People like to be consistent with the things
they have previously said or done.250
4. Foot-in-the-door technique — “a strategy used to persuade people to
agree to a particular action, based on the idea that if a respondent will
comply with a small initial request then they will be more likely to agree
to a later, more significant, request, which they would not have agreed
to had they been asked it outright.”251
5. Framing effect — a cognitive bias where people decide on options
based on if the options are presented with positive or
negative semantics; e.g. as a personal loss or gain.252
6. Loss aversion — the disutility of giving up an object is greater than the
utility associated with acquiring it, i.e., you’re leery of giving something
up once you have it, as compared to seeing that something as gain if
you don’t have it in the first place. 253
7. Conformity and social influence — the theory that people will conform
their ideas to the ideas of a group under social pressure. 254
8. Acquiescence effect — a tendency to respond in the affirmative to
survey items irrespective of substantive content.255
9. Mere exposure effect — the more often a person sees something new,
the more positive meaning they will give it. 247
10. Informational social influence — social influence occurs when a
person's emotions, opinions or behaviors are affected by others
intentionally or unintentionally.
11. The decoy effect — the phenomenon whereby consumers will tend to
have a specific change in preference between two options when also
presented with a third option that is asymmetrically dominated.256
12. Buffer effect or social support — the process in which a psychosocial
resource reduces the impact of life stress on psychological well-
being.257
13. Propinquity effect — the tendency for people to form friendships or
romantic relationships with those whom they encounter often, forming
a bond between subject and friend.258
14. Availability heuristic — “a mental shortcut that relies on immediate
examples that come to a given person's mind when evaluating a specific
topic, concept, method or decision.”259
15. Scarcity principle — “economic theory in which a limited supply of a
good, coupled with a high demand for that good, results in a mismatch
between the desired supply and demand equilibrium.” 260

199
 ANDREW W. PEARSON

Throughout the rest of this section, I will break down the 15 principles that can
be utilized by cruise line marketers. In the ensuing chapters, I will delve deeper
into how technology and psychology can be used together to increase
personalization.
The endowment effect was revealed in a famous study from Duke University,
which discovered that students who had won some coveted basketball tickets in
a raffle valued the tickets at $2,400, while those who had not won the tickets
would only agree to pay $170 for them. 247
The marketing takeaway here is that a brand’s customers will attribute a higher
value to things they already own.247 Cruise lines should try to increase their
customer’s ownership in their products by encouraging feedback and making it
easier to upload suggestions and comments through social media. 247 Scott Cook
believes that, “A brand is no longer what we tell the consumer it is — it is what
consumers tell each other it is.”261 This is a sentiment echoed in Jonah Sachs
statement that, “Your brand is a story unfolding across all customer touch
points.”262
In terms of reciprocity, a 2002 research found that “waiters could increase tips
with a tiny bit of reciprocity.”247 Tips rose by 3 percent when diners were given
an after-dinner mint, but went up to 20 percent, when the server delivered the
mint while looking the customer in the eye and telling them the mint was
specifically for them.247
In another example, “BYU sociologist Phillip Kunz sent Christmas cards to 600
completely random strangers. He received 200 Christmas cards back in
response.”247
The consistency principle was displayed in a study where “Princeton
researchers asked people if they would volunteer to help with the American
Cancer Society. Of those who received a cold call, 4 percent agreed. A second
group was called a few days prior and asked if they would hypothetically
volunteer for the American Cancer Society. When the actual request came later,
31 percent agreed.”247
The marketing takeaway here is for cruise lines to “help current customers and
potential users create an expectation of what they may say or do. For instance,
get users to opt-in to a marketing course and offer tools at the end that are used
by expert marketers. Subscribers may wish to stay consistent with their stated
goal of improving their marketing, and signing up for recommended tools will fall
right in line with this expectation.”247
According to Lee, “The first study on the foot-in-the-door method was
performed in the 1960s by Jonathan Freedman and Scott Faser.” 247 Researchers
called several homemakers to inquire about the household products they
used.247 Three days later, the researchers called again, this time asking to send a

200
 THE A.I. CRUISE LINE

group of workers to the house to manually note the cleaning products in the
home. The research found that “the women who responded to the first phone
interview were two times more likely to respond to the second request.”247
The marketing takeaway here: provide strong enough content that customers
will be motivated to frequently open your brand emails, as well as download
your content or generally go along with your requests.247 The more little things
they do, the more likely they are to comply with a larger request, like sharing
your content and inviting their friends to join in the brand conversation.247
Researchers Amos Tverksy and Nobel prize winning Daniel Kahneman found the
way they framed a question was more important than the question itself.247 The
researchers “polled two different groups of participants on which of two
treatments they would choose for people infected with a deadly disease.
• Treatment A: ‘200 people will be saved.’
• Treatment B: ‘a one-third probability of saving all 600 lives, and a two-
thirds probability of saving no one.’247
The majority of participants picked Treatment A because of the clear and simple
gain in saving lives. However, in Group 2, participants were told the following:
• Treatment A: ‘400 people will die.’
• Treatment B: ‘a one-third probability that no one will die, and a two-
thirds probability that 600 people will die.’ 247
According to Lee, “The majority of participants picked Treatment B because of
the clear negative effect of Treatment A.”247
The marketing takeaway here is that the “words you use and the way you frame
your content has a direct impact on how your readers will react.” 247 Lee
recommends that, whenever possible, brands “frame things in a positive light so
that readers can see a clear gain.”247 Word use is imperative as well. As Mark
Twain put it so succinctly, “The difference between the almost right word and
the right word is the difference between the lightning bug and the lightning.” 263
According to Decision Lab, “The framing effect has consistently proven to be one
of the strongest biases in decision making. The ways in which framing can be
used are nearly unlimited; from emotional appeals to social pressure to
priming.”264
When a positive frame is presented people are more likely to avoid risks but will
be risk-seeking when a negative frame is presented. The effect does seem to
increase with age, which could be highly important when designing health and
financial policies, as well as marketing to an older audience.264
In a famous loss aversion study, several Chicago Heights teachers were split into
two groups.247 “One group of teachers stood to receive bonuses based on the
performance of their students on standardized testing. Another group received

201
 ANDREW W. PEARSON

their bonus at the beginning of the year and stood to either keep it or lose it
based on the results of their students’ tests,” explains Lee.247 The results showed
that “the prepaid bonuses — the ones that could have been lost — had a bigger
impact on teachers.”247
The marketing takeaway here is that brands need to discover their customer’s
challenges and reservations, and then try to alleviate those concerns up front. 247
“Risk-free trials and money-back guarantees are one way to deal with loss
aversion,” argues Lee, since it removes the fear of loss from the equation. 247
In 1951, social psychologist Solomon Asch conducted an experiment to
investigate whether an individual would conform under social pressure. 254 As
detailed in Saul McLeod Solomon Asch — Conformity Experiment254, Solomon
Asch experimented on 50 male students from Swarthmore College to study
whether they would allow peer pressure to affect their judgment. “Using a line
judgment task, Asch put a naive participant in a room with seven
confederates/stooges. The confederates had agreed in advance what their
responses would be when presented with the line task.”254
According to McLeod, “The real participant did not know this and was led to
believe that the other seven confederates/stooges were also real participants
like themselves.”254 “Each person in the room had to state aloud which
comparison line (A, B or C) was most like the target line. The answer was always
obvious. The real participant sat at the end of the row and gave his or her answer
last.”254
In 12 of the 18 trials, the confederates gave the wrong answer. 254 “On average,
about one third (32%) of the participants who were placed in this situation went
along and conformed with the clearly incorrect majority on the critical trials,”
says McLeod.254
After the test, the subjects were asked why they conformed and most of them
“said that they did not really believe their conforming answers, but had gone
along with the group for fear of being ridiculed or thought ‘peculiar.’”254 Asch
concluded that, “Apparently, people conform for two main reasons: because
they want to fit in with the group (normative influence) and because they believe
the group is better informed than they are (informational influence).” 254 The key
takeaway for marketers here is that, “influencers and industry leaders can help
your product appear more valuable to others.” 247
According to the psychology website “Changing Minds”265 there are three
scenarios in which we are most likely to acquiesce to the request of others:
1. They seem to be a superior in some way.
2. They have a need whereby we can easily help them.
3. Answering the question fully seems like hard work.
Lee states that, “Leading questions are one way that the acquiescence effect

202
 THE A.I. CRUISE LINE

impacts the answers that one gives.”247

The marketing takeaway here is that cruise lines should be aware of the leading
questions they may be asking in customer development calls, surveys, or
questionnaires.247 “People can be easily swayed to answer in a certain way if the
question seems tilted in a certain direction,” warns — and recommends —
Lee.247
Robert Zajonc’s Chinese character study showed how the mere exposure to
something could increase positive feelings about it. Zajonc showed several
Chinese characters to non-Chinese-speaking participants, either once or up to 25
times, then asked the participants to guess the meaning of the characters.247 The
study revealed that the “more often a participant saw a character, the more
positive meaning they gave.”247
The marketing takeaway here is brands shouldn’t be afraid to repeat their
messaging.247 Social media is the perfect channel for brands to share their
content, as reposting helpful content can have a direct impact on an audience. 247
The repetition seen here probably goes unnoticed because customers can easily
surf away from a brand’s messaging by visiting other social media pages and/or
websites.
In an effort to curtail energy usage, Alex Lasky of Opower ran an experiment to
study how messaging could best encourage others to save energy. 266 Opower
sent customers one of the following four messages247:
• You can save $54 this month.
• You can save the planet.
• You can be a good citizen.
• Your neighbors are doing better than you.
Only the fourth message worked, leading to a 2 percent reduction in household
energy usage.247 The study showed that brands should use the experience of
others to help people see the benefits of their product or services. 247 There’s a
close correlation between informational social influence and social proof.247
The decoy effect can be seen in an old subscription advertisement for The
Economist, which stated247:
• Web Subscription – $59
• Print Subscription – $125
• Web and Print Subscription – $125
When Professor Dan Ariely tested this model with his students at MIT, he asked
them to choose a subscription option among the three choices.247 The results
were as follows247:
• Web Subscription – $59 (16 students)
• Print Subscription – $125 (0 students)

203
 ANDREW W. PEARSON

• Web and Print Subscription – $125 (84 students)

• Total revenue: $11,444
When the print subscription was removed, the results looked like this 247:
• Web Subscription – $59 (68 students)
• Web and Print Subscription – $125 (32 students)
• Total revenue: $8,012
Obviously, adding the decoy increases sales and the marketing takeaway is for
brands to add a decoy in their pricing.247 Lee concludes that, “The inclusion of an
option that is ‘asymmetrically dominated’ (a plan that seems out of whack or a
feature list that doesn’t quite add up) will make the other options more
appealing.”247
In terms of the buffer effect, in a study of pregnant women, “researchers
found that 91 percent of those with high stress and low social support suffered
complications whereas only 33 percent of pregnant women with high stress and
high social support suffered complications.”
The marketing takeaway here is for brands to be consistent with availability and
support for their customers. 247 “Constant support — in the form of email
communication, blogging, in-app messages etc. — may help others feel more
comfortable and less stressed,” advises Lee.247
For the propinquity effect, researchers discovered that “tenants in a small two-
floor apartment had closer friendships with their immediate neighbors. Least
likely friendships were between those on separate floors. And tenants who lived
near staircases and mailboxes had friendships on both floors.” 247 The marketing
takeaway here is for brands to be a constant presence on social media, as well
as in the inbox of its customers and subscribers. 247
In the late 1960s, Amos Tversky and Daniel Kahneman began their work on
“heuristic and biases.” 259 They discovered “that judgment under uncertainty
often relies on a limited number of simplifying heuristics rather than extensive
algorithmic processing.”259 Tversky and Kahneman coined the term “availability
heuristic” to explain these biases.
According to Wikipedia, an “availability heuristic is a mental shortcut that relies
on immediate examples that come to a given person's mind when evaluating a
specific topic, concept, method or decision. As follows, people tend to use a
readily available fact to base their beliefs about a comparably distant
concept.”259
In their New Yorker article, The Two Friends Who Changed How We Think About
How We Think267, Cass Sunstein and Richard Thaler explain that there were two
distinct themes in the work of Tverksy and Kahneman – judgment and decision-
making. “Judgment is about estimating (or guessing) magnitudes and

204
 THE A.I. CRUISE LINE

probabilities. How likely is it that a billionaire businessman from New York with
no experience in government gets elected President? Decision-making is about
how we choose, especially when there is uncertainty (meaning almost all the
time). What should we do now?” say Sunstein and Thaler.267
“Kahneman and Tversky showed that, in both of these domains, human beings
hardly behave as if they were trained or intuitive statisticians. Rather, their
judgments and decisions deviate in identifiable ways from idealized economic
models,” explain Sunstein and Thaler.267 “Most of the importance of Kahneman
and Tversky’s work lies in the claim that departures from perfect rationality can
be anticipated and specified. In other words, errors are not only common but
also predictable,” they say.267
Sunstein and Thaler explain the heuristic principle as such267:
For instance: ask people what they think is the ratio of gun
homicides to gun suicides in the United States. Most of them
will guess that gun homicides are much more common, but the
truth is that gun suicides happen about twice as often. The
explanation that Kahneman and Tversky offered for this type
of judgment error is based on the concept of “availability.” That
is, the easier it is for us to recall instances in which something
has happened, the more likely we will assume it is. This rule of
thumb works pretty well most of the time, but it can lead to big
mistakes when frequency and ease of recall diverge. Since gun
homicides get more media coverage than gun suicides, people
wrongly think they are more likely. The availability heuristic, as
Kahneman and Tversky called it, leads people to both excessive
fear and unjustified complacency — and it can lead
governments astray as well.
“The influence of their work has been immense — not only in psychology and
economics, where it has become part of the normal conversation, but in every
other field of social science, as well as medicine, law, and, increasingly, business
and public policy,” note Sunstein and Thaler.267
The marketing takeaway here is for brands to make their products or services
easy to grasp by providing examples of the actions users should take.247
Also known as the ‘Fear of missing out’ syndrome, the scarcity principle plays
upon the idea that people covet things that are scarce. As Investopedia explains
it, “Consumers place a higher value on goods that are scarce than on goods that
are abundant. Psychologists note that when a good or service is perceived to be
scarce, people want it more. Consider how many times you’ve seen an
advertisement stating something like: limited time offer, limited quantities,
while supplies last, liquidation sale, only a few items left in stock, etc. The feigned
scarcity causes a surge in the demand for the commodity.” 260

205
 ANDREW W. PEARSON

“Marketers use the scarcity principle as a sales tactic to drive up demand and
sales,” says Investopedia.260 The psychology behind the scarcity principle
dovetails well with the concepts of social proof and commitment.260 “Social proof
is consistent with the belief that people judge a product as high quality if it is
scarce or if people appear to be buying it. On the principle of commitment,
someone who has committed himself to acquiring something will want it more
if he finds out he cannot have it,” argues Investopedia. 260
The FYRE festival played up the fear of missing out principle as well as any
promotional event ever, promising concertgoers the experience of a lifetime in
the Bahamas. Having now been dubbed ‘the best festival that never was’, Fyre
Festival was then touted by hip hop mogul JaRule as being the ‘cultural
experience of the decade’. It has now become both legendary and the most
talked about festival flop ever.268
As explained in The Tonic Communications Fyre Festival: How Millennial FOMO
Enabled High-end Fraud, “A promotional video was produced with the specific
intent of giving audiences FOMO (Fear of Missing Out), a form of social anxiety
rooted in the concern that others might be having rewarding experiences that
the individual is not a part of.”268 “The video combined persuasive messaging
such as ‘immersive’, ‘transformative’, ‘remote and private island’ with imagery
of supermodels living their best lives — a carefully crafted illusion of what was
in store for attendees, should they be willing to spend thousands of dollars to
partake.”268
Billy McFarland, the CEO of the festival’s production company, “commented that
the video’s release would be known as the ‘Best coordinated social influencer
campaign ever’. 400 of the ‘hottest’ celebrities around the world including
artists, comedians, influencers and models posted an ambiguous burnt orange
‘Fyre tile’ across their Instagram accounts using the #FyreFestival and each
inviting their followers to ‘join me’. That was it.” 268 The campaign amazingly
“garnered over 300 million impressions within 24 hours.” 268 The event
immediately “sold out and rival festival organisers were stunned as investors
tried to pull money out of their events to put into Fyre.” 268
As two documentaries of the event have shown, it was all a scam. McFarland
defrauded investors to the tune of $27.4M and he is currently serving six years
in prison for these and other crimes.268 Thanks to the fear of missing out and a
brilliant social media marketing campaign, thousands of unwitting concert-goers
descended upon a little known island in the Bahamas for what turned out to be
the experience of a lifetime all right, just not quite the one they were expecting.
As the world becomes numb to advertising, marketers need to find a way to
connect with an audience on a visceral and emotional level and utilizing the
above psychological methods could be a good first step in the long process of
customer personalization.

206
 THE A.I. CRUISE LINE

Social Proof
Besides the 15 psychological methodologies described above, there are a
few others to consider, including social proof as well as the principle of
authority. “Think of it as building the foundation for massively scalable
word-of-mouth” — these are the words of venture capitalist and blogger Aileen
Lee describing the concept of social proof in her article Social Proof Is The New
Marketing.269 Lee believes that the best way to market a product or service “is
by harnessing a concept called social proof, a relatively untapped gold mine in
the age of the social web.”269 Lee contends social proof can generate sharing on
a viral level through social channels that can multiply the discovery of a brand
and add to its influence.269
Wikipedia describes social proof as “a psychological phenomenon where people
assume the actions of others reflect the correct behavior for a given situation…
driven by the assumption that the surrounding people possess more information
about the situation.”270 In other words, “people are wired to learn from the
actions of others, and this can be a huge driver of consumer behavior.”269
Eric Hoffer’s quote that, “when people are free to do as they please, they usually
imitate each other”271 is quite amusing and, unquestionably, true. It speaks
volumes about the herd mentality humans seem to succumb to as they
individually take cues for proper behavior in most situations from the behavior
of others. Psychologists call it the “conformity bias” and it is something that
politicians and marketers have tapped into to enormous effect for centuries.
Oscar Wilde’s quip that, “Most people are other people. Their thoughts are
someone else's opinions, their lives a mimicry, their passions a quotation”272
strikes a similar chord and it’s an idea that brands should keep in mind as they
devise marketing plans aimed at the market of one.
According to Robert Cialdini, who studied the principle of social proof in-depth
in his book Influence: The Psychology of Persuasion 273, “we view a behavior as
more correct in a given situation to the degree that we see others performing
it.”
In his article The Psychology of Marketing: 18 Ways Social Proof Can Boost Your
Results274, Alfred Lua concurs, stating, “So often in situations where we are
uncertain about what to do, we would assume that the people around us
(experts, celebrities, friends, etc.) have more knowledge about what’s going on
and what should be done.” Besides that, “we often make judgments based on
our overall impression of someone — A.K.A. the halo effect (named by
psychologist Edward Thorndike).”274
In general, Lua claims there are six types of social proof, including 274:
1. Expert: an expert in one’s industry recommends your products and/or
services or is associated with your brand.

207
 ANDREW W. PEARSON

2. Celebrity: a celebrity endorses your products.

3. User: current users recommend your products and/or services based on
personal experiences with your brand.
4. The wisdom of the crowd: a large group of people endorse your brand
for a myriad of reasons.
5. The wisdom of your friends: people see their friends approve of a
product or service.
In his influential Harvard Business Review paper Harnessing the Science of
Persuasion275, Robert B. Cialdini looked at the science behind the power of
persuasion and, since advertising is little more than trying to persuade a person
to choose one’s product and/or service over another, I think it is important to
explore persuasion through the lens of social media. Cialdini contends that275:
“For the past five decades, behavioral scientists have
conducted experiments that shed considerable light on the way
certain interactions lead people to concede, comply or change.
This research shows that persuasion works by appealing to a
limited set of deeply rooted human drives and needs, and it
does so in predictable ways. Persuasion, in other words, is
governed by basic principles that can be taught, learned and
applied.”
Cialdini’s six principles are275:
1. Like: People like those who like them.
2. Reciprocity: People repay in kind.
3. Social proof: People follow the lead of similar others.
4. Consistency: People align with their clear commitments.
5. Authority: People defer to experts.
6. Scarcity: People want more of what they can have less of.
For the reciprocity principle, people tend to give what they want to receive. 275
Praise is likely to have a warming and softening effect on people because there
is a human tendency to treat people the way they are themselves treated. 275 All
kinds of companies use this concept in their marketing to customers and brands
should emulate these offerings.
For the principle of social proof, people tend to follow the lead of similar
others.275 People use peer power whenever it’s available.275 Cialdini adds that,
“Social creatures that they are, human beings rely heavily on the people around
them for cues on how to think, feel, and act.”275 We know this intuitively, Cialdini
says “because intuition has also been confirmed by experiments, such as the one
first described in 1982 in the Journal of Applied Psychology.”275 In that study, “A
group of researchers went door-to-door in Columbia, South Carolina, soliciting
donations for a charity campaign and displaying a list of neighborhood residents
who had already donated to the cause. The researcher found that the longer the

208
 THE A.I. CRUISE LINE

donor list was, the more likely those solicited would be to donate as well.” 275
“To the people being solicited, the friends’ and neighbors’ names on the list were
a form of social evidence about how they should respond. But the evidence
would not have been nearly as compelling had the names been those of random
strangers,” explains Cialdini275 The lesson here is that “persuasion can be
extremely effective when it comes from peers.”275 Cialdini argues that, “The
science supports what most sales professionals already know: Testimonials from
satisfied customers work best when the satisfied customer and the prospective
customer share similar circumstances.”275
For the principle of consistency, brands should make their commitments active,
public, and voluntary. Cialdini states that, “Liking is a powerful force, but the
work of persuasion involves more than simply making people feel warmly toward
you, your idea, or your product. People need not only to like you but to feel
committed to what you want them to do. Good turns are one reliable way to
make people feel obligated to you. Another is to win a public commitment from
them.”275
For the principle of authority, people defer to experts, so brands should project
their expertise and not assume things are self-evident.275 As Lee explains,
“Approval from a credible expert, like a magazine or blogger, can have incredible
digital influence.”269 Her examples include the following269:
• “Visitors referred by a fashion magazine or blogger to designer fashion
rentals online at Rent the Runway drive a 200% higher conversion rate
than visitors driven by paid search.”269
• “Klout identifies people who are topical experts on the social web. Klout
invited 217 influencers with high Klout scores in design, luxury, tech and
autos to test-drive the new Audi A8. These influencers sparked 3,500
tweets, reaching over 3.1 million people in less than 30 days — a
multiplier effect of over 14,000x.”269
• “Mom-commerce daily offer site Plum District also reached mom
influencers thru Klout, and found customers referred by influential
digital moms shop at 2x the rate of customers from all other marketing
channels.”269
However, Lee warns that269:
“I don’t think a social proof strategy will be effective if you
don’t start with a great product that delights customers, and
that people like well enough to recommend. How do you know
if you have a great product? Track organic traffic growth,
reviews, ratings and repeat rates. And measure your viral
coefficient — if your site includes the ability to share, what
percentage of your daily visitors and users share with others?
How is the good word about your product being shared outside

209
 ANDREW W. PEARSON

your site on the social web? Do you know your Net Promoter
Score, and your Klout score?”
In his Fast Company article How to use the psychology of social proof to your
advantage276, Ed Hallin argues that, “A lot of things go into a person’s decision
to purchase a product, and social proof is certainly one of those important
factors. Studies show that 70% of consumers say they look at product reviews
before making a purchase, and product reviews are 12x more trusted than
product descriptions from manufacturers.”276 This isn’t really that surprising.
One subset of social proof is celebrity social proof. This is, of course, “celebrity
approval of your product or endorsements from celebrities.” 276 However, Hallin
warns that, “Celebrity endorsement is always a double-edged sword. If the
celebrity is properly matched to the brand, it can do wonders for the company.
If it’s a mismatch, it may produce a bad image of the company and its brand.” 276
Celebrities are also human beings and there can be a flavor-of-the-month aspect
to them, especially amongst athletes, but, for every Aaron Hernandez disaster
there might be a William Shatner Priceline endorsement that strikes internet and
financial gold, for both parties involved.
As Hallin explains, “To understand why celebrity endorsements work from a
psychological perspective, it’s important to familiarize yourself with the concept
of the extended self.”276 “The extended self,” Hallin contends, “is made of up the
self (me) and possessions (mine). It suggests that intentionally or unintentionally
we view our possessions as a reflection of ourselves. This is why consumers look
for products that signify group membership and mark their position in
society.”276 Apple is the perfect example of a company that produces goods that
people gladly overpay for in order to signify membership within an exalted – or
they think is an exalted – group.
“User social proof is approval from current users of a product or service,”
explains Hallin.276 This includes customer testimonials, case studies, and online
reviews and it is particularly effective when storytelling is involved. 276
Hallin believes that “We tend to imagine ourselves in other people’s shoes when
we read or hear a story. This is why stories are so persuasive and often more
trustworthy than statistics or general trends. Individual examples stick with us
because we can relate to them. Although statistics can be effective, it can be
tougher to really see yourself in the aggregate the way you can with a personal
account.”276
‘Wisdom of the Crowds’ social proof is “approval from large groups of other
people. It’s showing evidence that thousands, millions, or even billions have
taken the action that the company wants you to take — making a purchase,
subscribing, etc.”276
Hallin argues, “We kind of joke about FOMO in pop culture, but actually the Fear

210
 THE A.I. CRUISE LINE

of Missing Out is a real thing. It’s a form of social anxiety, and it’s a compulsive
concern that one might miss out on an opportunity. This anxiety is especially
relevant for social media, as the sharing of what’s going on in our daily lives
means you can constantly compare your status to others on these platforms.”276
Unsurprisingly, Hallin contends, “Social media has sparked dozens of different
ways to provide this kind of social proof. Facebook widgets that show other
Facebook friends that ‘like’ a brand, Twitter’s display of people you follow that
also follow another person, and the various ways that company offer rewards
for referring others to the brand are all examples of this.”276
Social proof is a powerful marketing tool and one that brands of all kinds need
to exploit. “One study of 10,000 accounts at a German bank revealed that
customers who came from customer referrals had 16% higher lifetime value than
those who came from other acquisition sources. Additionally, the customers
churned 18% less,” says Hallin.276
“The concept of implicit egotism is that most people subconsciously like things
that ‘resemble’ them in some way,” explains Hallin. 276 He adds that, “Studies
show that we value the opinions of people we perceive as most like us. We tend
to become friends with people that we have a lot in common with, so it makes
sense that social triggers like Facebook’s Like Box or referral programs are
successful.”276
Aileen Lee concludes that, “In the age of the social web, social proof is the new
marketing. If you have a great product waiting to be discovered, figure out how
to build social proof around it by putting it in front of the right early
influencers. And, engineer your product to share the love. Social proof is the
best way for new users to learn why your product is great, and to remind existing
users why they made a smart choice.”276
One word of caution when it comes to influencers and this is a rather sad story
in a few ways. In my former career/other life, I was a fledgling screenwriter and
film producer and I worked with a partner who production managed a movie
directed by her son that featured controversial YouTube personality Logan Paul.
The film was Airplane Mode and it contained a cast of social media influencers,
with Paul as the lead actor. Just as the film was about to be released, it was pulled
because Logan Paul filmed his infamous and controversial Suicide Forest video277,
which was uploaded to YouTube. The film quickly received worldwide
condemnation and was taken down. YouTube removed Paul from Google
Preferred, its preferred ad program. Suddenly, Paul became toxic.
Airplane Mode’s distributors cancelled its theatrical run, claiming Paul had
violated the moral clause in his contract. Ultimately, the movie went straight to
video and/or streaming and must have lost considerable money for its backers.
In many cases, when dealing with influencers, you’re not dealing with seasoned
professionals, but rather a flavor of the month celebrity. These are generally

211
 ANDREW W. PEARSON

younger people who are just getting their feet wet in a new marketing channel,
so caution should be taken. Sometimes you’re also dealing with professionals
who like to push the envelope because that’s what gets them noticed and flocks
of followers. It’s a tricky balance, but one that can be quite profitable if proper
caution is taken.

Psychometrics
According to Amit Paul Chowdhury, Psychometrics “is a field of study concerned
with the theory and technique involved behind psychological measurement. This
field is primarily concerned with testing, measurement, assessment, and related
activities. The field entails two key aspects for research purposes — a)
construction of instruments, b) revolves around the development of procedures
for measurement.”278
Social media can also be a wonderful place to capture a customer’s personality
traits. As Hannes Grassegger and Mikael Krogerus explain in their Das Magazin
article I Just Showed That the Bomb Was There 279, “Psychologist Michal Kosinski
developed a method of analyzing people’s behavior down to the minutest detail
by looking at their Facebook activity.”279 According to Grassegger and
Krogerus279:
“Psychometrics, sometimes also known as psychography, is a
scientific attempt to ‘measure’ the personality of a person. The
so-called Ocean Method has become the standard approach.
Two psychologists were able to demonstrate in the 1980s that
the character profile of a person can be measured and
expressed in five dimensions, the Big Five: Openness (how open
are you to new experiences?), Conscientiousness (how much of
a perfectionist are you?), Extroversion (how sociable are you?),
Agreeableness (how considerate and cooperative are you?),
and Neuroticism (how sensitive/vulnerable are you?). With
these five dimensions (O.C.E.A.N.), you can determine fairly
precisely what kind of person you are dealing with — her needs
and fears as well as how she will generally behave. For a long
time, however, the problem was data collection, because to
produce such a character profile meant asking subjects to fill
out a complicated survey asking quite personal questions. Then
came the internet. And Facebook. And Kosinski.”
In 2008, with a fellow Cambridge student, Kosinski created a small app for
Facebook called MyPersonality that asked users a handful of questions from the
Ocean survey and they would receive a rating, or a “Personality Profile”,
consisting of traits defined by the OCEAN method.279 The researchers, in turn,
got the users’ personal data, which soon amounted to millions and millions of

212
 THE A.I. CRUISE LINE

reviews.279 “It was, literally, the then-largest psychological data set ever
produced,” state Grassegger and Krogerus.279
In the ensuing years, Kosinski and his colleagues continued the research; “first
surveys are distributed to test subjects — this is the online quiz. From the
subjects’ responses, their personal Ocean traits are calculated. Then Kosinski’s
team would compile every other possible online data point of a test subject —
what they’ve liked, shared, or posted on Facebook; gender, age, and location.”279
Once the researchers dug into the data, they discovered that amazingly reliable
conclusions could be drawn about a person by observing their online behavior. 279
For example, “men who ‘like’ the cosmetics brand MAC are, to a high degree of
probability, gay,” which isn’t that surprising. However, there were other, more
interesting findings; for example, one of the best indicators of heterosexuality is
liking Wu-Tang Clan.279 Also, followers of Lady Gaga are most probably
extroverts, while someone who likes philosophy is probably an introvert. 279
Kosinski and his team continued their work, tirelessly refining their models. “In
2012, Kosinski demonstrated that from a mere 68 Facebook likes, a lot about a
user could be reliably predicted: skin color (95% certainty), sexual orientation
(88% certainty), Democrat or Republican (85%).”279 Level of intellect, religious
affiliation, alcohol, cigarette, and drug use could all be calculated as well. 279 For
businesses, employee Facebook pages could be scanned by HR to screen out
potentially problematic candidates.
As Kosinski continued refining his model, he discovered that, with a mere ten
likes as input, his model could appraise a person’s character better than an
average coworker.279 With seventy, “it could ‘know’ a subject better than a
friend; with 150 likes, better than their parents. With 300 likes, Kosinski’s
machine could predict a subject’s behavior better than their partner. With even
more likes it could exceed what a person thinks they know about themselves,” 279
which is a pretty frightening thought in-and-of-itself.
The day Kosinski published his findings, he received two phone calls, both from
Facebook; one a threat to sue, the other a job offer.279
Since the publication of Kosinski’s article, Facebook has introduced a
differentiation between public and private posts so the data isn’t as easily
accessible now.279 In “private” mode, “only one’s own friends can see what one
likes. This is still no obstacle for data-collectors: while Kosinski always requests
the consent of the Facebook users he tests, many online quizzes these days
demand access to private information as a precondition to taking a personality
test.”279
Kosinski and his team are now adding variables beyond Facebook Likes.279 Offline
activity is now traceable and “motion sensors can show, for example, how fast
we are moving a smartphone around or how far we are traveling (correlates with

213
 ANDREW W. PEARSON

emotional instability).”279
Flipping this idea on its head, Kosinski speculated his research could become a
search engine for people.279 By using all of this data, psychological profiles could
not only be constructed, but they could also be sought and found. 279 For
example, if a company, or a politician, wants to find worried fathers, or angry
introverts, or undecided Democrats, these profiles could be uncovered in the
data.279
To Kosinski’s chagrin, one company he had been partnered with — Cambridge
Analytica — was involved with Donald Trump’s 2016 presidential election. 279
Cambridge Analytica has now become infamous and was shut down in 2018
because of its questionable activities during the U.S. 2016 presidential election.
It had bought extensive personal data on American voters — “What car you
drive, what products you purchase in shops, what magazines you read, what
clubs you belong to” — and used the data in highly unethical ways to help elect
Donald Trump.279
In America, detailed personal consumer data is available for a price and
Cambridge Analytica snapped it up and the company crosschecked these data
sets with Republican Party voter rolls and online data, such as Facebook likes.279
OCEAN personality profiles were built from this data and, from a selection of
digital signatures, there suddenly emerged real individual people with real fears,
needs, and interests — and home addresses.279 By the time of the 2016
presidential election, Cambridge Analytica had assembled psychograms for all
adult US citizens — 220 million people — and they used this data to influence
electoral outcomes.279
Chowdhury puts it succinctly when he says, the success of the Cambridge
Analytica work “can be attributed to the combination of three core
techniques, behavioral science using the OCEAN Model, Big Data analysis, and ad
targeting.”278 Cambridge Analytica bought “personal data from a range of
different sources, like land registries, automotive data, shopping data, bonus
cards, club memberships, and more.” It aggregated “this data with the electoral
rolls of the Republican party and online data, to calculate a Big Five personality
profile.”278
According to Chowdhury, Nix showed “how psychographically categorized voters
can be differently addressed. The messages differed for the most part only in
microscopic details, to target the recipients in the optimal psychological way by
including different headings, colors, captions, with a photo or video.” 278 Digging
down into such granular detail helped Trump reach down to the most granular
group-level of a customer. “Pretty much every message that Trump put out was
data-driven,” states Nix.278
The return on investment was extraordinary. “The embedded Cambridge
Analytica team comprised of only a dozen people. The firm received $100,000

214
 THE A.I. CRUISE LINE

from Trump in July, $250,000 in August, and $5 million in September. According

to Nix, the company earned over $15 million overall.”278
Most importantly, “The decision to focus on Michigan and Wisconsin in the final
weeks of the campaign was made on the basis of data analysis done by the
organization.”278
“Trump’s conspicuous contradictions and his oft-criticized habit of staking out
multiple positions on a single-issue result in a gigantic number of resulting
messaging options that creates a huge advantage for a firm like Cambridge
Analytica: for every voter, a different message,” explains Grassegger and
Krogerus.279
Mathematician Cathy O’Neil notes that Trump is like a machine learning
algorithm that adjusts to public reactions.279 On the day of the third 2016
presidential debate, “Trump’s team blasted out 175,000 distinct variations on his
arguments, mostly via Facebook,”279 which is an astounding number of unique
ads. “The messages varied mostly in their microscopic details, in order to
communicate optimally with their recipients: different titles, colors, subtitles,
with different images or videos” were utilized, explains Grassegger and
Krogerus.279 This is personalization marketing at its finest.
Small towns, city districts, apartment buildings, and even individual people could
be targeted, explains Grassegger and Krogerus.279 Blanket advertising — the idea
that a hundred million people will be sent the same piece of marketing collateral,
the same television advert, the same digital advert — is over, forever, note
Grassegger and Krogerus.279 Micro and personalization targeting has reached the
point where politicians — and regular companies — can advertise highly detailed
and personalized messages to a market of one.
Cambridge Analytica separated the entire US population into 32 different
personality types, and focused their efforts on only seventeen states. 279 “Just as
Kosinski had determined that men who like MAC cosmetics on Facebook are
probably gay, Cambridge Analytica found that a predilection for American-
produced cars is the best predictor of a possible Trump voter.” 279 Among other
things, this kind of information helped the Trump campaign focus in on what
messages to use, and where to use them, perhaps even what channel to use
them on.279 In effect, the candidate himself became an implementation
instrument of the model.279
As Grassegger and Krogerus note, the first results seen by Das Magazin were
amazing: psychological targeting increased the clickthru rate on Facebook ads by
more than sixty percent. And the so-called conversion rate (the term for how
likely a person is to act upon a personally-tailored ad, i.e., whether they buy a
product or, yes, go vote) increases by a staggering 1,400 percent.” 279
Now, what does all of this mean for a marketer? How can a marketer use

215
 ANDREW W. PEARSON

Facebook Likes to gain a deeper understanding of its customers? Well,

potentially, by analyzing these ‘Likes’, a marketer could predict how open,
conscientious, outgoing and neurotic an individual user and/or customer is. It
could be as simple as doing a Facebook graph search of “Pictures liked” or
“Videos liked” and/or “Stories Liked” with the customer’s name. In addition to
predicting a user's personality, these tests could estimate a user’s/customer's
age, relationship status, intelligence level, life satisfaction, political and religious
beliefs, and education.
A brand’s HR department would also find these personality test results
interesting as matching a candidate with jobs based on their personality might
make more sense than the current scattershot approach HR often takes in hiring
— and firing. These personality tests could also reveal troubling traits, like
alcohol and/or drug use that should give pause to the hiring of a potential
prospect.

Figure 20: IBM Watson’s Personality Insights Sunbrust Chart Visualization on

Author’s Twitter Feed.
Source: https://personality-insights-demo.ng.bluemix.net/
In its Artificial Intelligence in Logistics280, DHL Customer Solutions & Innovation
describe the IBM Watson Personality Insights tool, which allows users to develop
a highly specific understanding of a person’s character, as seen in Figure 20. As

216
 THE A.I. CRUISE LINE

per the DHL Customer & Innovation team, “The tool can be used for the creation
of novel and personalized services. For example, in the wealth management
industry, IBM Watson Investment Advisor can draw correlations between a
customer’s personality, life situation, and the vast ocean of financial market
data.”280
According to the DHL team, “These inputs can be matched with various
investment alternatives to recommend an optimal personalized wealth
management strategy.”280 In addition, the system “uses deep learning to provide
financial advisors with a highly efficient and personalized way to serve clients,
while indicating how to deepen relationships through other channels in their
firms, such as lending solutions.”280

Facial Recognition
Facial recognition technology is the capability to identify or verify a person from
a digital image or a video frame from a video source by comparing the actual
facial features of someone on camera against a database of facial images, or
faceprints, as they are also known.
As patrons enter a cruise ship, “security cameras feed video to computers that
pick out every face in the crowd and rapidly take many measurements of each
one’s features, using algorithms to encode the data in strings of numbers,”281 as
explained in the Consumer Reports article Facial Recognition: Who’s Tracking
Who in Public.281 These are called faceprints or templates.281 The faceprints are
compared against a database, and when there’s a match, the system alerts hosts,
sales people or security guards, should anyone have been caught cheating,
stealing or shoplifting in the past.
Cruise line personnel can receive alerts through a mobile app or an SMS when a
member of a VIP loyalty program enters the cruise line property. A screen can
display the patron’s name, or a photo just taken from the video feed. Shopping
preferences and other details, like a customer’s average daily Theo or ADT can
also be displayed.
Currently, facial recognition technology can be more useful for security
departments than customer service.281 At the 2014 Golden Globe Awards, facial
recognition technology was used to scan for known celebrity stalkers. 281 The
technology has also been used to bar known criminals from soccer matches in
Europe and Latin America.281 “Police forces and national security agencies in the
U.S., the UK, Singapore, South Korea, and elsewhere are experimenting with
facial recognition to combat violent crime and tighten border security.” 281
Facial recognition technology is becoming second nature to consumers, who are
used to tagging themselves in photos on Facebook, Snapchat, Picasa, and/or
WeChat. In 2015, Google launched a photo app that helped users organize their

217
 ANDREW W. PEARSON

pictures by automatically identifying family members and friends. 281 Google,

however, suffered a public relations and social media disaster when its system
labeled a photo of two black people as gorillas.281 The search giant quickly
apologized profusely and promised to fix its algorithms 281, but this does show
that the technology isn’t foolproof and sensitivity is imperative.
Currently, MasterCard is “experimenting with a system that lets users validate
purchases by snapping a selfie. Like fingerprint scanners and other biometric
technologies, facial recognition has the potential to offer alternatives to
passwords and PINs.”281
This technology is moving so fast, privacy advocates are having trouble keeping
up with it all. In this regard, today’s facial recognition technology is reminiscent
of the World Wide Web of the mid-1990s.281 Back then, few people would have
anticipated that every detail about what we read, watched, and bought online
would become a commodity traded and used by big business and sometimes,
more sinisterly, hacked and used by nefarious individuals to perpetrate
crimes.281
Facial recognition technology “has the potential to move Web-style tracking into
the real world, and can erode that sense of control.” 281 Experts such as Alvaro
Bedoya, the executive director of Georgetown Law’s Center on Privacy &
Technology, and the former chief counsel to the Senate’s subcommittee on
privacy, technology, and the law finds this attack on privacy alarming. 281
“People would be outraged if they knew how facial recognition” was being
developed and promoted, Bedoya states.281 “Not only because they weren’t told
about it, but because there’s nothing they can do about it. When you’re online,
everyone has the idea that they’re being tracked. And they also know that there
are steps they can take to counter that, like clearing their cookies or installing an
ad blocker. But with facial recognition, the tracker is your face. There’s no way
to easily block the technology,” warns Bedoya.281
Right now, facial recognition is largely unregulated and few consumers seem to
even be aware of its use. “Companies aren’t barred from using the technology
to track individuals the moment we set foot outside. No laws prevent marketers
from using faceprints to target consumers with ads. And no regulations require
faceprint data to be encrypted to prevent hackers from selling it to stalkers or
other criminals,” Bedoya warns.281 This is true for both the United States, Asia,
and Europe.
Users might be happy to tag their face and the faces of their friends and
acquaintances on a Facebook wall, but they might shudder if every mall worker
was jacked into a system that used security-cam footage to access their family’s
shopping habits.281 This could, however, be the future of retail, according to Kelly
Gates, associate professor in communication and science studies at the
University of California, San Diego. 282

218
 THE A.I. CRUISE LINE

In her article Our Biometric Future: Facial Recognition Technology and the
Culture of Surveillance282, Gates argues that “Regardless of whether you want to
be recognized, you can be sure that you have no right of refusal in public, nor in
the myriad private spaces that you enter on a daily basis that are owned by
someone other than yourself.” Gates concludes that by entering a retail
establishment filled with facial recognition technology, you are tacitly giving your
consent to the cruise line to use it, even if you are unaware of its use.282
Facial recognition technology in the offline world is now becoming more and
more prevalent, particularly in the hospitality industry. “On Disney’s four cruise
ships, photographers roam the decks and dining rooms taking pictures of
passengers. The images are sorted using facial recognition software so that
photos of people registered to the same set of staterooms are grouped together.
Passengers can later swipe their Disney ID at an onboard kiosk to easily call up
every shot taken of their families throughout the trip.” 281
Starting in 2010, the 1,200-room Hilton Americas-Houston in Texas used a facial
recognition system that was mainly designed as a security tool to identify VIP
guests so the hotel staff could greet them by name. 281 The hotel won’t confirm
if the system is still active, but similar technology is being rolled out at hotels and
cruise lines worldwide.281
“In a recent study of 1,085 U.S. consumers by research firm First Insight, 75
percent of respondents said they would not shop in a store that used the
technology for marketing purposes. Notably, the number dropped to 55 percent
if it was used to offer good discounts.”281
However, consumers may warm to facial recognition technology once it
becomes more widespread, especially if cruise lines offer enough incentives to
make it worthwhile. In some cases, full facial recognition isn’t needed, some
marketers just want to determine the age, sex, and race of shoppers.
In Germany, the Astra beer brand recently created an automated billboard
directed solely at women, even to the point of shooing men away. 281 The
billboard approximates the women’s age, then plays one of 80 pre-recorded ads
to match.281 For a cruise line, this could help if they want to direct specific
advertising towards women, or to men, or to a certain age group.
Cruise lines can also utilize “facial recognition systems to see how long people of
a particular race or gender remain in the shop, and adjust displays and the store
layout to try to enhance sales.”281 Using related technology, some high-end
cruise lines in the U.S. have experimented with “memory mirrors” that perform
tricks such as storing images of what shoppers tried on so that they can be
revisited, or emailed directly to friends for feedback.
In 2014, Facebook announced a project it calls DeepFace, “a system said to be
97.35 percent accurate in comparing two photos and deciding whether they

219
 ANDREW W. PEARSON

depicted the same person — even in varied lighting conditions and from
different camera angles. In fact, the company’s algorithms are now almost as
adept as a human being at recognizing people based just on their silhouette and
stance.”281
“Entities like Facebook hold vast collections of facial images,” says Gates, the UC,
San Diego professor.281 “People have voluntarily uploaded millions of images,
but for their own personal photo-sharing activities, not for Facebook to develop
its facial recognition algorithms on a mass scale.”281
Potentially Facebook, Instagram, WeChat, Pinterest, Snapchat, Google, or a
whole host of other social media companies could use their vast databases of
faceprints to power real-world facial recognition.281 “Hypothetically, a tech giant
wouldn’t need to share the faceprints themselves. It could simply ingest video
feeds from a store and let salespeople know when any well-heeled consumer
walked through the door.”281 It could also, potentially, do this for a cruise
operator as well, to prevent money laundering, Know Your Customer (KYC), or
AML activities.

Affective Computing
In his article We Know How You Feel283, Raffi Khatchadourian profiles Rana el
Kaliouby, co-founder and CEO of Affectiva, a startup that specializes in AI
systems that sense and understand human emotions. Affectiva develops
“cutting-edge AI technologies that apply machine learning, deep learning, and
data science to bring new levels of emotional intelligence to AI.” 284 It has been
ranked by the business press as one of the United States’ fastest-growing
startups.283 Affectiva is the most visible among a host of competing startups that
are building emotionally responsive machines.283 Its competitors include
Emotient, Realeyes, and Sension.283
Khatchadourian explains that, “Our faces are organs of emotional
communication; by some estimates, we transmit more data with our expressions
than with what we say, and a few pioneers dedicated to decoding this
information have made tremendous progress.”283 Arguably, The most successful
of these pioneers is Rana el Kaliouby.283
“Since the nineteen-nineties a small number of researchers have been working
to give computers the capacity to read our feelings and react, in ways that have
come to seem startlingly human,” explains Khatchadourian. 283 Researchers
“have trained computers to identify deep patterns in vocal pitch, rhythm, and
intensity; their software can scan a conversation between a woman and a child
and determine if the woman is a mother, whether she is looking the child in the
eye, whether she is angry or frustrated or joyful.” 283 “Other machines can
measure sentiment by assessing the arrangement of our words, or by reading

220
 THE A.I. CRUISE LINE

our gestures. Still others can do so from facial expressions,” says

Khatchadourian.283
In his book Architects of Intelligence284, Martin Ford interviews the CEO of
Affectiva, Rana el Kaliouby, and she explains her work in the following way:
“If you think about a lot of people who are building these
devices, right now, they’re focused on the cognitive intelligence
aspect of these devices, and they’re not paying much attention
to the emotional intelligence. But if you look at humans, it’s not
just your IQ that matters in how successful you are in your
professional and personal life; it’s often really about your
emotional and social intelligence. Are you able to understand
the mental states of people around you? Are you able to adapt
your behavior to take that into consideration and then
motivate them to change their behavior, or persuade them to
take action? All of these situations, where we are asking people
to take action, we all need to be emotionally intelligent to get
to that point. I think that this is equally true for technology that
is going to be interfacing with you on a day-to-day basis and
potentially asking you to do things.”
Kaliouby’s thesis “is that this kind of interface between humans and machines is
going to become ubiquitous, that it will just be ingrained in the future human-
machine interfaces, whether it’s our car, our phone or smart devices at our home
or in the office.”284 Kaliouby sees a world where, “We will just be coexisting and
collaborating with these new devices, and new kinds of interfaces.” 284 “I think
that, ten years down the line, we won’t remember what it was like when we
couldn’t just frown at our device, and our device would say, ‘Oh, you didn’t like
that, did you?’” says Kaliouby.283
Afectiva’s signature software, Affdex, tracks four emotional “classifiers” —
happy, confused, surprised, and disgusted. 283 “The software scans for a face; if
there are multiple faces, it isolates each one. It then identifies the face’s main
region — mouth, nose, eyes, eyebrows — and it ascribes points to each,
rendering the features in simple geometries,” explains Khatchadourian. 283
“Affdex also scans for the shifting texture of skin — the distribution of wrinkles
around an eye, or the furrow of a brow — and combines that information with
the deformable points to build detailed models of the face as it reacts,” says
Khatchadourian.283 The algorithm identifies an emotional expression by
comparing it with countless others that it has previously analyzed. “If you smile,
for example, it recognizes that you are smiling in real time,” says Kaliouby.283
Like every company working in the emotional intelligence field, “Affectiva relies
on the work of Paul Ekman, a research psychologist who, beginning in the sixties,
built a convincing body of evidence that there are at least six universal human

221
 ANDREW W. PEARSON

emotions, expressed by everyone’s face identically, regardless of gender, age, or

cultural upbringing.”283 Classifying human expressions into combinations of
forty-six individual movements called “action units”, Ekman compiled the Facial
Action Coding System, or FACS — a five-hundred-page taxonomy of facial
movements.283 FACS “has been in use for decades by academics and
professionals, from computer animators to police officers interested in the
subtleties of deception.”283
Although widely used, Ekman and FACS has its critics, “among them social
scientists who argue that context plays a far greater role in reading emotions
than his theory allows.”283 However, context-blind computers appear to support
Ekman’s conclusions.283 “By scanning facial action units, computers can now
outperform most people in distinguishing social smiles from those triggered by
spontaneous joy, and in differentiating between faked pain and genuine pain,”
says Khatchadourian.283 “Operating with unflagging attention, they can register
expressions so fleeting that they are unknown even to the person making them,”
notes Khatchadourian.283
“The human face is a moving landscape of tremendous nuance and complexity.
It is a marvel of computation that people so often effortlessly interpret
expressions, regardless of the particularities of the face they are looking at, the
setting, the light, or the angle,” notes Khatchadourian.283 He adds that, “A
programmer trying to teach a computer to do the same thing must contend with
nearly infinite contingencies. The process requires machine learning, in which
computers find patterns in large tranches of data, and then use those patterns
to interpret new data.”283
In March, 2011, Kaliouby and her team were invited to demonstrate an early
version of their technology to executives from Millward Brown, a global market-
research company.283 “Kaliouby was frank about the system’s limitations — the
software still was having trouble distinguishing a smile from a grimace — but the
executives were impressed,” reports Khatchadourian.283 Ad testing relies heavily
on subjective surveys, which is easily tainted by human bias.283 Spontaneous,
even unconscious, sentiment is what really interests marketers and Kaliouby’s
technology promised that, along with better results.283
“A year earlier, Millward Brown had formed a neuroscience unit, which
attempted to bring EEG technology into the work, and it had hired experts in
Ekman’s system to study video of interviews,” explains Khatchadourian.283
However, these ideas had proved impossible to scale up.283 The Millward Brown
executives proposed a test to Kaliouby: “If Affdex could successfully measure
people’s emotional responses to four ads that they had already studied,
Millward Brown would become not just a client but also an investor.”283
Millward Brown chose the Dove TV commercial “Onslaught,” which begins with
an image of a young girl, then “shifts to her perspective as she is bombarded by

222
 THE A.I. CRUISE LINE

a montage of video clips — a lifetime of female stereotypes compressed into

thirty-two seconds — before the ad ends with the girl, all innocence, and the
tagline ‘Talk to your daughter before the beauty industry does.’” 283
Although the ad was critically acclaimed, surveys revealed that many people
considered it emotionally difficult to watch.283 In its study, Affdex scanned more
than a hundred respondents watching the ad, and discovered more complicated
responses. Although respondents were uncomfortable during the ad, at the
moment of resolution this uneasiness vanished.283 “The software was telling us
something we were potentially not seeing,” Graham Page, a Millward Brown
executive, explained.283 Recognizing the power of Affectiva’s technology,
Millward Brown’s parent company, WPP, invested $4.5 million in the
company.283 Affdex was soon being used to test thousands of ads per year. 283
Kaliouby claims that her company has analyzed more than two million videos, of
respondents from over eighty countries.283 “This is data we have never had
before,” says Kaliouby.283 When Affectiva began, she had trained the software
on just a few hundred expressions, but, once she started working with Millward
Brown, hundreds of thousands of people on six continents began turning on web
cams to watch ads for testing, and all their emotional responses — natural
reactions, in relatively uncontrolled settings — flowed back to Kaliouby’s
team.283
Affdex can now read the nuances of smiles better than most people can. As the
company’s database of emotional reactions grows, the software is getting better
at reading other expressions, including furrowed eyebrows. “A brow furrow is a
very important indicator of confusion or concentration, and it can be a negative
facial expression,” explains Kaliouby.283
Today, Kaliouby says companies pay millions of dollars to create funny and
emotionally compelling ads, but the advertisers and the brands have no idea if
they are striking the right emotional chord with their audience.284 The only way
to find out, before emotional response technology existed, was to ask people. 284
So, if you are the person watching the ad, you’d get a survey with some basic
questions, but the answers wouldn’t be very reliable because it is biased, or so
believes Kaliouby.284
With Affdex, however, Kaliouby explains that, “as you’re watching the ad, with
your consent it will analyze on a moment-by-moment basis all your facial
expressions and aggregate that over the thousands of people who watched that
same ad.”284 The result: “an unbiased, objective set of data around how people
respond emotionally to the advertising.”284 Affectiva can “then correlate that
data with things like customer purchase intent, or even actual sales data and
virality.”284
“People are pretty good at monitoring the mental states of the people around
them,” says Kaliouby.284 “We know that about 55% of the signals we use are in

223
 ANDREW W. PEARSON

facial expression and your gestures, while about 38% of the signal we respond
to is from tone of voice. So how fast someone is speaking, the pitch, and how
much energy is in the voice. Only 7% of the signal is in the text and the actual
choice of words that someone uses!”284
A multi-billion-dollar industry that tracks people’s sentiments about this product
or that service has been built within just a couple of years, which is amazing
when you think that all of these tweets, likes and posts only account for about
7% of how humans communicate overall. 284 “What I like to think about what
we’re doing here, is trying to capture the other 93% of non-verbal
communication,” contends Kaliouby.284
According to Kaliouby, Affectiva looks “at the tone of voice and the occurrence
of speech events, such as how many times you say ‘um’ or how many times you
laughed. All of these speech events are independent of the actual words that
we’re saying.”284 Affectiva combines “these things and takes what we call a
multimodal approach, where different modalities are combined, to truly
understand a person’s cognitive, social or emotional state,” explains Kaliouby.284
“If you take facial expressions or even the tone of a person’s voice, the
underlying expressions are universal,” says Khatchadourian.284 A smile is a smile
no matter where in the world it breaks across a face. “However, we are seeing
this additional layer of cultural display norms, or rules, that depict when people
portray their emotions, or how often, or how intensely they show their
emotion,” says Khatchadourian.284 “We see examples of people amplifying their
emotions, dampening their emotions, or even masking their emotions
altogether.”284 Masking often occurs in Asian markets, where Asian populations
are less likely to show negative emotions.284 In Asia, there is an increased
incidence of what’s known as a “social smile”, or a “politeness smile.”284 These
are not expressions of joy, but rather expressions that say, “I acknowledge you,”
and, in that sense, they are very social signals.284
Affdex is sold “as a tool that can make reliable inferences about people’s
emotions — a tap into the unconscious,”283 if you will. Clients like CBS use it to
tests new TV shows.283 Affectiva is also working with Oovoo, an instant
messaging service, to integrate the technology into video calls.283 “People are
doing more and more videoconferencing, but all this data is not captured in an
analytic way,” says Kaliouby.283 “Capturing analytics, it turns out, means using
the software — say, during a business negotiation — to determine what the
person on the other end of the call is not telling you,” writes Khatchadourian.
“The technology will say, ‘O.K., Mr. Whatever is showing signs of engagement
— or he just smirked, and that means he was not persuaded,’” says Kaliouby.283
Kaliouby believes Affectiva’s technology has the potential to monetize what she
calls an ‘Emotion Economy’.283 “Tech gurus have for some time been predicting
the Internet of Things, the wiring together of all our devices to create ‘ambient

224
 THE A.I. CRUISE LINE

intelligence’ — an unseen fog of digital knowingness,” explains

Khatchadourian.283 Emotion could be a part of this IoT.283
Kaliouby predicts that, in the coming years, mobile devices will contain an
“emotion chip,” which constantly runs in the background, the way geolocation
currently works on phones now.283 “Every time you pick up your phone, it gets
an emotion pulse, if you like, on how you’re feeling,” Kaliouby says.283 “In our
research, we found that people check their phones ten to twelve times an hour
— and so that gives this many data points of the person’s experience,” she
explains.283
The free economy is, in fact, an economy of the bartered self, but attention can
never be limitless.283 Thales Teixeira, a business professor who collaborated with
Kaliouby on her technology, explains that, “There are three major fungible
resources that we as individuals have. The first is money, the second is time, and
the third is attention. Attention is the least explored.” 283 Teixeira calculated the
value of attention, and found that, like the dollar, its price fluctuates. 283
Using Super Bowl ads as a rough indicator of the high end of the market, Teixeira
“determined that in 2010 the price of an American’s attention was six cents per
minute. By 2014, the rate had increased by twenty per cent — more than double
inflation.”283 The jump was attributed to the fact that attention, at least, the kind
worth selling, is becoming increasingly scarce as people spend their free time
distracted by a growing array of devices and services. 283 “What people in the
industry are saying is ‘I need to get people’s attention in a shorter period of time,’
so they are trying to focus on capturing the intensity of it,” explains Teixeira. 283
“People who are emotional are much more engaged. And because emotions are
‘memory markers’ they remember more. So the idea now is shifting to: how do
we get people who are feeling these emotions?” says Teixeira. 283
Affectiva filed a patent for “a system that could dynamically price advertising
depending on how people responded to it.” 283 However, they soon discovered
that they were not alone; more than a hundred similar patents for emotion-
sensing technology existed, many of them, unsurprisingly, also focused on
advertising.283
Companies like AOL, Hitachi, eBay, IBM, Yahoo!, and Motorola are also
developing technology in this space.283 Sony had filed several patents; “its
researchers anticipated games that build emotional maps of players, combining
data from sensors and from social media to create ‘almost dangerous kinds of
interactivity,’” notes Khatchadourian.283 There are “patents for emotion-sensing
vending machines, and for A.T.M.s that would understand if users were ‘in a
relaxed mood,’ and receptive to advertising,” claims Khatchadourian. 283
Incredibly, Verizon had drafted a plan for “a media console packed with sensors,
including a thermographic camera (to measure body temperature), an infrared
laser (to gauge depth), and a multi-array microphone. By scanning a room, the

225
 ANDREW W. PEARSON

system could determine the occupants’ age, gender, weight, height, skin color,
hair length, facial features, mannerisms, what language they spoke, and whether
they had an accent.”283
According to Khatchadourian, “the console could identify pets, furniture,
paintings, even a bag of chips.”283 It could track “ambient actions,” such as
“eating, exercising, reading, sleeping, cuddling, cleaning, playing a musical
instrument.”283 It could even probe other devices, learning what a person might
be browsing on the web, or writing in an e-mail.283 The console could scan for
affect, tracking moments of laughter or the raised voice of an argument. 283 All of
this data tracking would then shape the console’s choice of TV ads. 283 “A marital
fight might prompt an ad for a counsellor. Signs of stress might prompt ads for
aromatherapy candles. Upbeat humming might prompt ads ‘configured to target
happy people,’” which is a pretty scary idea.283 Verizon’s plan was for the system
to then broadcast the ads on every device in the room. 283
Although Verizon’s system seems very Big Brotheresque, it was not an anomaly,
explains Khatchadourian.283 Microsoft’s Xbox One system already contains many
of these features, including “a high-definition camera that can monitor players
at thirty frames per second.”283 “Using a technology called Time of Flight, it can
track the movements of individual photons, picking up minute alterations in a
viewer’s skin color to measure blood flow, then calculate changes in heart
rate.”283 “The software can monitor six people simultaneously, in visible or
infrared light, charting their gaze and their basic emotional states, using
technology similar to Affectiva’s.”283 As Khatchadourian sees it, “the system has
tremendous potential for making digital games more immersive.”283 Microsoft
isn’t stopping at game development either, they envision TV ads that target a
viewer’s emotions, and program priced according to how many people are
watching in the room.283 Google, Comcast, and Intel are charting a similar
path.283
Wearables like Nike’s FuelBand and particularly Fitbit collect a tremendous
amount of health data on a person.283 Apple’s Health app, a fitness app pre-
installed on new iPhones “can track weight, respiratory rate, sleep, even blood-
oxygen saturation.”283 This information could be used to build emotional
profiles, says Khatchadourian.283 Researchers at Dartmouth have already
“demonstrated that smartphones can be configured to detect stress, loneliness,
depression, and productivity, and to predict G.P.A.s.” 283
For Affectiva, there is now plenty of interest in its Affdex solution.283 The
company has conducted research for Facebook, experimenting with video ads.283
Samsung has licensed it and a company in San Francisco wants to give its digital
nurses the ability to read faces.283 A Belfast entrepreneur is interested in its use
at night clubs.283 A state initiative in Dubai, the Happiness Index, wants to
measure social contentment.283 “Dubai is known to have one of the world’s
tightest CCTV networks, so the infrastructure to acquire video footage to be

226
 THE A.I. CRUISE LINE

analyzed by Affdex already exists,” explains Kaliouby.283

All-in-all, Affectiva could be revealing the future of customer engagement.
Although somewhat Big Brotheresque, all this data collection is incredibly
seamless, which means it will probably be popping up in all kinds of technology
in the coming years. For that reason alone, it is important for cruise lines to keep
an eye on this potentially revolutionary technology.
Affectiva’s other use cases include:
• Improve story flow — Moment-by-moment emotion data can pinpoint
viewer confusion and lack of engagement. This insight helps improve
the story arch of an animatic or ad.
• Create cut-downs — Identify the most emotionally engaging moments
in longer TV ads so you can retain the most impactful parts when cutting
down to shorter online ads.
• TV show character analysis — Assess audience engagement with
characters on TV shows and the interplay of characters — especially
important when introducing new actors to a show.
• Movie trailer creation – Movie studios use emotion analytics to test
different versions of a movie trailer and use the most engaging
moments in each to create the final cut.
• Determine media spend — Test final ads for emotional engagement to
identify potential wear out. Direct your advertising dollars to the ads
with the best emotional impact on repeat view.
• Test voice overs and brand reveal — Use emotion data to see if your
audiences are emotionally engaged at the moment of brand reveal in
an ad and test the effectiveness of taglines and voice-overs.

SEO
A web search engine is a software system designed to search for information on
the web and the search results are generally presented in a line of results often
referred to as search engine results page (SERPs).285 “The information may be a
mix of web pages, images, and other types of files. Some search engines also
mine data available in databases or open directories. Unlike web directories,
which are maintained only by human editors, search engines also maintain real-
time information by running an algorithm on a web crawler.” 286
In the US, Google is, by far, the biggest search engine around.287 Outside the U.S.,
Google’s main competitors are “Baidu and Soso.com in China; Naver.com and
Daum Communications in South Korea; Yandex in Russia; Seznam.cz in Czech
Republic; Yahoo! in Japan, Taiwan [sic].”288
Bit players like Bing compete with Google on standard search, but today Apple

227
 ANDREW W. PEARSON

and Amazon are making big inroads into Google’s dominance, with Facebook set
to be a challenger in the not-too-distant future as well. With those latter two,
search is organically included within their platforms, i.e., when someone
searches for an item to buy on Amazon, it gets included in the overall search
rankings, ergo, an ecommerce site has become an important search engine.
Why is search so influential? Because users flock to search engines to organize
the vast amounts of information most buyers need to make purchase decisions.
“The main purpose of Google Search is to hunt for text in publicly accessible
documents offered by web servers, as opposed to other data, such as with
Google Image Search.”288 “The order of search on Google's search-results pages
is based, in part, on a priority rank called a ‘PageRank.’”288 As Sharma et al. state
in their book Mobile Marketing, “Search is one of the best ways to find content
and the absolute best way for a marketer to determine consumer intent.”226
Google Search “provides at least 22 special features beyond the original word-
search capability, and language translation of displayed pages.” 288 “In June 2011,
Google introduced ‘Google Voice Search’ and ‘Search by Image’ features for
allowing the users to search words by speaking and by giving images. In May
2012, Google introduced a new Knowledge Graph semantic search feature to
customers in the U.S.”288
“When Google was a Stanford research project, it was nicknamed BackRub
because the technology checks backlinks to determine a site's importance.”288
Backlinks — and the quality of them — are very important for search engine
optimization (SEO). The higher the quality of backlinks, the higher a website’s
ranking.
Even today, backlinks count, and they likely count prominently for SEO and,
although backlinks are not always within a company’s control, they are highly
important due to their stature as the earliest persisting Google ranking factor.
According to the Moz 2015 Ranking Survey289, “the data continues to show some
of the highest correlations between Google rankings and the number of links to
a given page.” Today, quality backlinks are of the utmost importance and Google
is the one who decides the quality of those backlinks; links from known spammy
sites or sites associated with them, or merely hosted on servers that also host
spammy content negatively affect rankings.289
In the early days of the battle for internet search supremacy, “previous keyword-
based methods of ranking search results, used by many search engines that were
once more popular than Google, would rank pages by how often the search
terms occurred in the page, or how strongly associated the search terms were
within each resulting page.”290 Google’s PageRank algorithm instead “analyzes
human-generated links assuming that web pages linked from many important
pages are themselves likely to be important. The algorithm computes a recursive
score for pages, based on the weighted sum of the PageRanks of the pages

228
 THE A.I. CRUISE LINE

linking to them.”290 As a result, PageRank is thought to correlate well with human

concepts of importance.290
Google wants site owners to focus on developing great content — clear,
accurate, highly-readable content that other site owners want to link to. 289 The
way modern engines make this determination is by using advanced natural
language processing, artificial intelligence and machine learning.289 These
evolving technologies enable the search engines to understand content without
relying on a small set of specific keywords and phrases. 289 Google has invested
heavily in this area, as evidenced by the plethora of white papers and research
posted on its ‘Machine Intelligence’ website.289
As Brian Alpert argues in his article Search engine optimization in 2017: A new
world where old rules still matter291, “One aspect of today’s search engines that
makes them very different from their predecessors is that advances in artificial
intelligence and machine learning have enabled them to understand content and
its underlying concepts independently of specific keywords.” 291 Alpert adds:
“This renders null and void the old concept that one must focus on keywords
specific to a certain kind of content in order to be found via search engines. In
today’s landscape, exact keyword matches are less influential than ever before
as engines can understand the relationships between words that are
semantically related.”291
With the introduction of its Knowledge Graph, Google is attempting to give users
answers instead of just links.288 “If you want to compare the nutritional value of
olive oil to butter, for example, Google Search will now give you a comparison
chart with lots of details. The same holds true for other things, including dog
breed and celestial objects. Google says it plans to expand this feature to more
things over time.”288
Knowledge Graph also allows users to filter results.288 “Say you ask Google: ‘Tell
me about Impressionist artists.’ Now, you’ll see who these artists are, and a new
bar on top of the results will allow you to dive in to learn more about them and
to switch to learn more about abstract art, for example,” explains Lardinois. 288
Search advertising falls into two main types — natural search results, and paid
sponsorship based on keywords.292 “Natural search requires high-quality,
constantly updated content and search engine optimization (SEO). Paid search
requires work to optimize keyword choice and messaging, but can be
phenomenally expensive.”292 When not optimized for conversion, this can be a
very pricey channel to use, with low conversion rates as well.292
Cruise Line companies should constantly be testing their web pages for the most
searched for and/or visited pages. As Siroker and Komen explain, A/B testing is
particularly good for website marketing, especially for uncovering a website’s
best landing pages.132 “Defining success in the context of A/B testing involves
taking the answer to the question of your site’s ultimate purpose and turning it

229
 ANDREW W. PEARSON

into something more precise: quantifiable success metrics. Your success metrics
are the specific numbers you hope will be improved by your tests,” argue Siroker
and Komen.132 An e-commerce website could easily define its success metrics in
terms of revenue per visitor132, but it is still important to understand such things
as traffic sources, bounce rate, top pages, conversion rates, conversion by traffic
source, amongst other things.
In a Google Analytics enabled website, every user visiting a webpage is tracked
from its immediately preceding website293 294, which are divided into six main
channels – search, referral, display advertising, email, social and direct. These
analytics allow businesses to assess the contribution of each channel to their
overall traffic volume, as well as evaluate investment in traffic driving measures.
According to Baye et al.295, Search is traffic delivered by internet search engines
and it is the largest channel. The source of search traffic can either be sponsored
(paid) or organic, which is highly dependent on SEO effectiveness.296 Search
traffic has been extensively studied by Berman and Katona297, Jerath et al298, Yao
and Mela299; unsurprisingly, the general consensus is that the top position on the
search result is not necessarily the most profitable, as discovered in studies by
Agarwal et al300, and Jeziorski and Segal301. Liu and Toubia302 theorize that this
might be because users conducting “informational” search queries are unlikely
to have much awareness of the brand they ultimately visit and are therefore only
on the first step of their customer journey.
As per Strzelecki303, Referral traffic originates from domains other than search
engines or social media, its where users find links of interest and click from.
Referral traffic may thus consist of users who have acquired awareness and
enough interest in the linked brand from elsewhere to visit its site304.
Google states that Display traffic is traffic that “found your site by clicking on an
ad that you ran on another website. Banner ads on blogs and image ads on news
sites are some common generators of display traffic.”47
Email traffic “clicked on links from email campaigns, follow up emails, and even
email signatures,” says Google.47 It is one of the best traffic tools, thought to
generate website traffic at rates as high as 16%305. Chittenden and Rettie306 claim
that it is often used to maintain transactional performance of existing customers
by presenting discounts or offers.
Social traffic will be counted from people who find a company’s page through an
associated social media account.47 Businesses can check in on users who are
landing on their page as a result of social media accounts like Facebook, LinkedIn,
or Twitter.47 This channel is hugely important to marketers as the reach,
influence and viral potential of social makes it a highly cost effective way to reach
a motivated audience. Bickart & Schindler307, Kumar & Benbasat308, Zhang,
Craciun, & Shin309 found that electronic word of mouth (e-WOM) marketed
through social media is an effective brand communication tool, which has a

230
 THE A.I. CRUISE LINE

strong influence on purchase intention, as per Chan & Ngai310, Copley311, and
See-To & Ho312.
According to Google, Direct traffic occurs when browsers come to a site by
entering the URL directly into the browser’s address bar.47 Google recommends
businesses “keep an eye on this one if you've been running offline or traditional
media ads like print, TV, or radio, because they require audiences to remember
and type out your web address.”47
Ultimately, Analytics channels are all sources of traffic, says Google.47 “As such,
the same traffic metrics can be applied to all channels. The best part about being
able to apply the same metrics to each channel is that you now have the ability
to perform direct comparisons in order to diagnose issues and optimize
performance.”47
In his article Supercharging Your SEO with AI Insights, Automation, and
Personalization313, Jim Yu believes that, AI “is making search more human.
Although search does not yet ‘speak’ to users in the same way the Google Duplex
demo could, its objective is very similar.” He adds, “Google’s RankBrain
technology uses machine learning to understand the meaning of the content it
crawls; it infers intent from ambiguous search queries; and it uses feedback data
to improve the accuracy of its results. In other words, it listens and it learns.” 313
Research by BrightEdge “into a dataset of over 50 million keywords revealed that
84.4 percent of queries return universal search results. This occurs as Google
uses AI to match the layout of search results pages to the user’s intent.”313
According to BrightEdge, “There are now 37 different search engine result page
(SERP) categories, a number that will only increase over the coming months and
years.”313 These are:
Standard Category Weather
Taller Organic Cards Images Game scores
Local 3-pack Video / Trailers Twitter Tweets
Quick answers Live Discover more places
Shopping/PLA Top sights Send to Google home
Rich snippets Reviews People also search for
Site carousel Blogs See results about
Site links Knowledge panel Widgets
Site image carousel Carousel Found in related search
Top stories/News Apps Quotes
AMP Google for jobs Events

231
 ANDREW W. PEARSON

Google flights Recipes People also ask

Scholarly research

Table 9: Search Engine keywords

Source: BridgeEdge313

“The potential for personalization has not yet been truly tapped, but Google’s
Sundar Pichai recently made public its goal to be an ‘AI-first’ company.”313 This
means, “we should all expect the search landscape to change dramatically as AI
takes center stage in the way it has already done in products like Google Photos
and Google Lens.313 As co-founder Sergey Brin put it: “AI touches every single
one of our main projects, ranging from search to photos to ads.” 313
The pace of development on this front is accelerating, as everything at Google
seems to have something to do with AI.313 “Google is all too aware that AI can
simply deliver better, more personalized experiences for consumers,” says Yu. 313
“However, search marketers need to pay close attention to these technological
advancements if they are to avail themselves of these opportunities for SEO,”
claims Yu.313
There are three key areas in which AI can improve SEO performance313:
1. Insights
2. Automation
3. Personalization
AI can process and analyze data at a scale simply not possible for humans. 313
“This makes it an essential complement to any search strategist, as AI can deliver
the information we need to make informed decisions out of noisy, unstructured
data,” claims Yu.313
AI can be used to glean SEO insights in the following ways313:
• Understand underlying need in a customer journey.
• Identify content opportunities.
• Define opportunity space in the competitive context.
• Map intent to content.
• Use structured data and markup.
• Invest in more long-tail content.
• Ensure content can be crawled and surfaced easily by all user-agents.
• Automation.
“SEO is a labor-intensive industry that requires a huge amount of attention over
the long term. Where we can automate tasks to receive the same output, we
could produce ourselves, we should make this a top priority. The time saved
through automation can be applied to the areas that require our skills, like

232
 THE A.I. CRUISE LINE

strategy and creative content,” advises Yu.313

Ultimately, AI can be used for SEO personalization in the following ways313:
• Create content by persona, customer journey stage and delivery
mechanism.
• Enhance user experience and conversion through personalization.
• Use semantically specific pages to associate query and intent.
• Use personalization and audience lists to nurture leads across search
and social.
• Use AI to help publish content at the right times on the right networks.
As reported by Tom Simonite in his Wired article Google Search Now Reads at a
Higher Level314, on 25 October 2019, Google announced that was enhancing “its
search-ranking system with software called BERT, or Bidirectional Encoder
Representations from Transformers to its friends.” “It was developed in the
company’s artificial intelligence labs and announced last fall, breaking records
on reading comprehension questions that researchers use to test AI software,”
says Simonite.314
Pandu Nayak, Google’s vice president of search, said at a briefing Thursday that
“the muppet-monickered software has made Google’s search algorithm much
better at handling long queries, or ones where the relationships between words
are crucial. You’re now less likely to get frustrating responses to queries
dependent on prepositions like ‘for’ and ‘to,’ or negations such as ‘not’ or
‘no.’”314
“This is the single biggest positive change we’ve had in the last five years,” said
Nayak.314
“One illustration of BERT’s power offered up by Google is how it helped its search
engine interpret the query ‘Parking on hill with no curb.’ The current version of
its search algorithm responded to that as if it referred to a hill that did have a
curb. The BERT-powered version highlights a page advising drivers to point their
wheels toward the side of the road,” explains Simonite.
“Another was the query ‘2019 brazil traveler to usa need a visa.’ To a human,
that’s a clear attempt to discover the requirements for Brazilians heading to the
US, but pre-BERT Google misunderstood the crucial ‘to’ and returned an article
about US citizens traveling to Brazil as the top result. With BERT, the search
engine correctly serves up a page about requirements for Brazilian citizens
heading north,” notes Simonite.314
“Google says it receives billions of searches per day and that the BERT upgrade
will affect rankings on one out of every 10. But Nayak says most users probably
won’t notice,” says Simonite.314 “People outside the US turning to Google for
help will see some of the most significant changes. Nayak said that the BERT

233
 ANDREW W. PEARSON

upgrade helped its system get much better at identifying so-called featured
snippets, particularly in languages other than English,” adds Simonite. 314
“Google’s upgrade is a notable example of recent progress in software that
attempts to understand language. It has made machine learning algorithms
much better at decoding the subtleties of language by attending to the context
around a particular word,” says Simonite.314
Although machine learning has proven to be a powerful way to teach software
to sort or interpret data such as images or text, each program typically has to be
“trained” using example data, which can be a long, involved and expensive
proposition.314 “That’s often been tricky to come by for text documents.”314
Projects would depend on paying people to label specific examples, such as good
and bad restaurant reviews,” notes Simonite.314
As Simonite explains it314:
“In the spring and summer of 2018, OpenAI and the Allen
Institute for AI showed a simpler and more powerful method.
They taught machine learning programs the differences
between words — even homonyms like May the month, may
the verb, and May the name — by looking at other words in the
text, even if they’re in a different sentence. Models trained that
way on very large collections of text picked up a kind of general
sense for language and could then be specialized to particular
tasks using relatively small collections of labeled data.”
“Allen AI christened its system ELMo, for Embeddings from Language Models,”
says Simonite.314 “That caused Google’s researchers to think of Sesame Street in
October 2018 when they announced their own still-more-powerful take on the
new way for machine learning to learn language, BERT,” notes Simonite. 314 “Like
the systems from OpenAI and Allen AI, Google’s software set new records on AI
language tests, such as answering questions.”314
“People are very excited, because progress is so quick,” says Jeff Wu, a research
engineer who has worked on OpenAI’s language projects.314 “One side effect:
Researchers have had to invent new and more difficult tests for software on
tasks such as basic reading comprehension,” explains Simonite.314
“That doesn’t mean BERT is ready to critique your college essay,” warns
Simonite.314 “Language is incredibly subtle and nuanced,” Nayak says.314 “Each
time Google improves the search box’s facility with language, he says, people
submit more complex and challenging queries, effectively raising the bar for
Google’s reading robots,” offers Simonite.314

234
 THE A.I. CRUISE LINE

Content Intelligence
In its article The Magic of AI in a content-driven world. Using AI to create content
faster315, the Adobe Enterprise Content Team argues that we're currently in the
midst of a content explosion. Perhaps because of this, it is also a time when
“Consumers expect to have personalized, relevant experiences at all times, in all
places, and on all platforms.”315
“An IDC survey cites that 85 percent of marketing professionals feel under
pressure to create assets and deliver more campaigns, more quickly. In fact, over
two-thirds of respondents are creating over ten times more assets to support
additional channels. This increased level of complexity is driving volume and
associated costs.”315
When thinking about what is needed to create this kind of content for thousands
or even millions of customers at the near real-time speed that is necessary, doing
it manually is impossible.315 Adobe's State of Creativity in Business 2017 survey
found that “40 percent of creatives are using AI in photo and design
retouching,”315 so it’s already happening.
Currently, it can take hours for a designer to find just the right image to use in a
piece of marketing collateral, and that’s not counting the time required to
manipulate the image, to crop it, to find the right layout scenario, and then to
publish it to an online catalog and/or social media channel. 315 Serving the right
content to the right person at the right time adds more time. 315 The cost for all
this work adds up, as does the cost of photo shoots to create new assets. 315 AI
and machine learning can help marketers find and reuse assets more efficiently,
as well as deliver new and personalized content at scale, thereby helping a brand
get a better return on its marketing investments.315
According to the Adobe Sensei Team, “AI can help you create more relevant
content and more engaging experiences across the customer journey at the
speed your customers expect.
On the creative side, AI can speed up all kinds of tedious tasks, from identifying
and organizing assets to adjusting and refining for specific channels.” 26 “On the
audience level, AI can help you better understand which audiences respond to
which content, or how often people prefer to receive emails, so you can deliver
the experiences your customers want while respecting their preferences and
privacy,” says the Sensei team.26
Klein concurs, stating “because the technology becomes smarter and more
intuitive as it ingests more data, AI also can play a valuable role in automating
the content creation process.”63 AI “offers capabilities for marketers that range
from choosing the best image for a campaign or optimizing the content in a
creative based on real-time user interactions,” says Klein.63 “For example, from

235
 ANDREW W. PEARSON

a content creation perspective, this allows the ability to understand the focal —
or sellable — point of hero images, and then to auto-crop them for best
performance based on an understanding of millions of assets with similar meta-
data,” explains Klein.63 “In this way, AI enhances creativity and enables a level of
responsiveness and efficiency that until very recently was unachievable for
marketers,” contends Klein.63
The advertising maverick Dave Trott was probably correct when he said that
“Creativity may well be the last legal unfair competitive advantage we can take
to run over the competition”316, so anything that enhances creativity should be
pursued. However, one should remember the award–winning copywriter Luke
Sullivan’s warning that, “Creativity is like washing a pig. It’s messy. It has no rules.
No clear beginning, middle or end. It’s kind of a pain in the ass, and when you’re
done, you’re not sure if the pig is really clean or even why you were washing a
pig in the first place.”317
“Designers simply don't have time to tag the hundreds of images uploaded from
every photo shoot. Even if they did, the list of keywords probably wouldn't be as
exhaustive as it should be. But when a photo isn't tagged, it's virtually impossible
to find by searching in an image bank of thousands,” argue the Adobe Experience
Cloud team.315 “According to IDC, marketers report that one-third of marketing
assets go unused or underutilized with the average organization creating
hundreds of new marketing assets each year.”315 Repurposing images is unlikely,
which means ROI suffers.
To try to tackle this issue, Adobe has created “Auto Tag”, an Adobe Sensei
capability that automatically tags images with key words. 315 For example, a
marketer might have a picture of a young girl on a beach under a clear blue sky,
which could be tagged with keywords like “beach”, “girl”, “dancing”, “sundress”,
“blue sky”, “white sand”, or even a place like “Aruba.”
“The Auto Tag service is used to power the Smart Tags features in Adobe
Experience Manager, Photo Search in Adobe Lightroom, and Visual Search in
Adobe Stock,” explains the Enterprise Content Team. 315 "It’s exciting to see the
capabilities of auto-tagging,” says Jonas Dahl, product manager for Adobe
Experience Manager.315 “We did several manual search queries against a
customer's repository and showed the assets we were able to find. Then we
applied Smart Tagging and did the same searches. This time the results were
significantly better and much more comprehensive. And in a fraction of the
time,” explains Dahl.315
“Adobe Sensei uses a unified AI and machine-learning framework, along with
Adobe's deep domain expertise in the creative, marketing, and document
segments, to harness the company's massive volume of content and data assets
— from high-resolution images to customer clicks,” says the Enterprise Content
Team.315

236
 THE A.I. CRUISE LINE

“Adobe Sensei technology has learned to automatically identify what is in a

photo. And not just an object like a car or a girl, but the concept of the photo,
including context, quality, and style.”315 Theoretically, someone could search for
an image with the words “walking” and “slow” and “the search might result in
an image of an elderly man using a walker, because the technology made the
connection between slow and walker.”315 Sensei’s auto-phrasing service could
tell you how each tag scored for prominence because the machine differentiates
between the primary and secondary objects.315 “This enables the technology to
build a simple sentence or caption that more accurately describes the photo,
such as ‘An elderly man walking with a walker in a park.’” 315
Using the Sensei framework, cruise lines could train the AI and machine learning
models to create their own unique auto tags.315 Identifying brand characteristics
like the company logo could help the designers adhere to specific brand
standards, or training it to identify a company’s products so that they can be
tagged in pictures on social media, which helps identify true reach. 315
Custom auto tagging not only has the potential to increase a marketing team's
efficiency, but it could lead to image-based shopping.315 “Auto tagging identifies
what is in the photo and finds the best matches for the customer. Auto tagging
also allows brands to gain a deeper understanding of their audience and can help
uncover market trends on social media, without the brand having to rely on tags
and text.”315 “If you run a social media feed through Adobe Sensei, it will tag
places your brand is pictured — even if it's not mentioned or tagged — allowing
you to see what is trending,” explains the Enterprise Content Team.315
As any marketer can tell you, locating an image is simply step one in a multi-step
process.315 Unless, that is, you're using AI.315 With products like Adobe’s Deep
Cutout, designers “can automatically remove an image’s background and replace
it with one that fits the brand guidelines.”315 Soon, designers will “be able to
mask out an area such as a highway, and in just a few clicks, see what it looks
like with a river, neighborhood, or other background — completely reinventing
the photo in seconds.”315
Auto Crop is another Adobe tool that can automate the cropping and sizing of
images for different aspect ratios. The Enterprise Content Team gives the
example of a shoe manufacturer who “may have guidelines that require only the
shoe be shown, so they can automatically have all photos cropped
accordingly.”315
The AI can be trained on image aesthetics as well, so it automatically selects the
best image and rejects anything that is below a certain quality standard and/or
criteria.315
Time savings in any one of these areas could be quite substantial, but combined
together, the velocity of creating and delivering content gets faster and faster.315
Adobe argues that the real power of this technology comes from creating custom

237
 ANDREW W. PEARSON

workflows that allow brands to search, mask, crop, and publish in a fraction of
the time it took in the past.315
For brands creating international marketing campaigns, this type of custom
workflow can eliminate or reduce “the tedious, manual work involved in creating
all of the different assets.”315 It allows brands to scale their campaigns to as many
countries as needed.315 When a “designer uploads a file to Adobe Creative Cloud,
a custom workflow kicks off a series of Adobe Sensei Content AI Services that
expedites the entire process from tagging, to cropping, to delivery of your
production-ready asset to Adobe Experience Manager.” 315 Adobe claims all of
this can happen in a matter of hours instead of days.315 Once again, creatives are
allowed to focus on being creative, a place they would, undoubtedly, prefer to
be anyway.
When this type of AI is coupled with a brand’s content and audience data, its
value increases exponentially.315 “When you can combine what you know about
the image with what you know about the customer from online and offline
behaviors, you can micro-target customers with content that is truly relevant,”
says Richard Curtis, principal solutions consultant for Adobe.315 “Furthermore,
the machine will continue to learn customer patterns that help you fine-tune
your personalization even further.”315 As Richard notes, “More personalization
leads to more clicks,”315 to say nothing of stronger brand loyalty bonds.315
According to Adobe’s Indelible content, incredible experiences75 article, the
Adobe Enterprise Content team says that, “marketers are competing with brands
that lure their customers not only with products and services, but also with
individual experiences. And they’re setting some healthy expectations, from
recommending a film that customers will love, including a personal treat in their
orders.”
“To meet those expectations, marketers need to develop a steady flow of
compelling content. You start the content journey with ideas and concepts, then
create and manage assets, deliver and personalise experiences and finally
analyse performance,” says the Adobe Enterprise Content team.75 “And you
need to do all of this fast enough for the experiences to adapt instantly to every
channel and screen your customer may use. The goal is achieving what McKinsey
calls marketing’s holy grail: digital personalization at scale.”75
Machine learning is quickly becoming the “go-to tool to help marketers connect
content with data and analytics, everywhere from lead scoring and retargeting
to personalization and segmentation.” According to the Adobe Enterprise
Content team, the following three ways can make content more potent 75:
1. Automate tedious tasks: “The complex, data-driven tasks that once
only humans could perform are now in the realm of machines. They
can easily and accurately handle repetitive tasks in specific contexts,

238
 THE A.I. CRUISE LINE

like categorizing or scheduling, and can free humans for more value-
added activities.”75
2. Gain insights from big data: “Humans can’t readily process massive
amounts of data. Computers can. They can analyze big data — even
unstructured data — to discover patterns, trends, and associations,
and then offer actionable insights.”75
3. Improve prediction accuracy: “Not only can computers analyze data,
they can learn from it. And the more data they have, the savvier they
become at making on-target recommendations and predictions.”75
“Creating authentic one-to-one experiences requires extensive resources and an
investment that your budget may not support. Even if you’re flush with cash, you
can’t scale manually — you simply cannot hire that many people or analyse such
vast datasets,” warns the Adobe Enterprise Content team. 75 The solution lies at
the intersection of content marketing and AI — content intelligence.”75
With machine learning, software can analyse images imported into an editor “to
detect facial features, similar images and even which way the subject is looking,”
claims the Adobe Enterprise Content team.75 A designer can swap out images in
real time to quickly preview as many options as a client might want to see.75 If it
doesn’t look quite right, no problem, the editor can go back to any point in the
process and see how a different decision — perhaps a young couple in a
mortgage ad look excitedly at each other rather than at their new home — and
then judge how it impacts the emotional experience of the ad. 75
“AI can serve as your creative assistant, quickly assembling suggested content
for audiences at every touchpoint and even optimizing it, so the burden’s not on
you. If your AI application supports voice recognition, you can even tell your
assistant what you want — like making the mountains disappear in a climbing
shot to focus on the gear,” explains the Adobe Enterprise Content team.75
“You want one place where everyone — marketers, creatives and outside
agencies — can find approved images and video to ensure experiences will
remain consistent across channels,” advises the Adobe Enterprise Content
team.75 “But manually tagging images with descriptive and contextual metadata
is tedious, inconsistent and often incomplete. It’s the type of job where
machines excel. AI-powered smart tags automatically provide consistent,
content-based metadata in seconds — saving you hours,” they claim.75 As
Adobe’s Senior Product Marketing Manager Elliot Sedegah succinctly puts it,
“Computers will not complain about having to add metadata, they will not try to
avoid it and they will work just as hard on the hundred thousandth image as on
the first.”75
Adobe allows for personalised customer treatment as well. “You can introduce
as many experience variations as you choose for your digital properties to
personalise customer experiences. By evaluating all behavioural and contextual

239
 ANDREW W. PEARSON

variables, machine learning can determine the best experience for each
consumer — regardless of channel, device or screen. As machine learning learns
what works with each customer, predictive analytics can tell you what each one
wants to see and buy — so you’ll know whether they’ll be excited by the image
of the hotel on the island beach in Phuket or the snowy slopes at Whistler.”75
“Just as a doctor must address each patient’s issues and concerns, you must
appreciate each customer’s needs and desires,” argues the Adobe Enterprise
Content team.75 Meeting the expectations of each customer calls for new tools,
“You can’t win in the digital era with industrial-age technology,” Adobe claims.75
Integrating AI will help brands “deliver the truly surprising and delightful
experiences that keep customers feeling on top of the world,” promises the
Adobe Enterprise Content team.75

Measurement
In her article Future of Advertising: Automated, Personalized, and Measurable 318,
Giselle Abramovich details an Adobe Think Tank panel discussion at Advertising
Week 2017, in which “Phil Gaughran, U.S. chief integration officer at agency
McGarryBowen, made a bold prediction: By 2022, he said, 80% of the advertising
process will be automated, ‘a threshold that will never be surpassed.’” The
remaining 20%, Gaughran claims, “will comprise such elements as brand value,
storytelling, and other more experiential tactics that will always need a human
driver.”318
According to Gaughran, this means a “changing job description in terms of what
it means to work in advertising, unlocking a huge well of opportunity for
advertisers.”318 “He reminded the audience that data doesn’t deliver insights —
people do.”318 “The more automated we become, the more we need humanity,”
he said.318 Keith Eadie, VP of Adobe Advertising Cloud, agreed, “adding that as
automation becomes mainstream, the big differentiator for brands will be
human insight and creativity.”318 “Brands will always need human capital to
innovate,” Eadie argued.318
“Measurement, the panelists agreed, is a huge topic in advertising today, and
one that’s growing in importance as more advertising becomes measurable,”
says Abramovich.318 One big hurdle to measurement is the concept of the
“walled gardens” and the dark social of Facebook, Instagram, WeChat, and
Google, which, according to Eadie, “have scaled media properties and tons of
data, but the data and its ability to be activated stays within these platforms.” 318
However, Eadie believes these walls “will start to come down in as little as five
years, as Facebook and Google gear up to compete against newer entrants.” 318
Time to plan for this change is now.
“Amazon is a rising walled garden, and this rise will mean a new set of

240
 THE A.I. CRUISE LINE

competition to the media landscape,” says Eadie.318 “If companies can’t get a
sense of which garden is most impactful, they will move dollars to the platforms
that do provide understanding of impact.”318 Facebook has proven to be very
accommodating in this area any time advertisers are willing to put money on the
table. This is a scenario that won’t be ending any time soon, no matter how much
privacy trouble Facebook gets into.
Will Warren, EVP, digital investment, at Zenith Optimedia, argues that
measurement has improved with the onset of automation, because it allows
companies to have a single view of their media, not just a single view of their
customer.318 “Further digitization will allow more user level data, and we can tie
that to an outcome,” he says.318 “[Automation essentially brings] multitouch
attribution across the digital landscape. Consolidated ad buying means better
measurement.”318
Jill Cress, National Geographic CMO, “believes the current state of measurement
is more about ‘measure what you can’ than ‘measure what you need to
measure.’”318 “Today, [advertisers] are focused a lot on the vanity metrics, like
views, impressions, and clicks. But we need to figure out how far down the funnel
these things are taking people,” she says. “We feel like we are at a moment
where we will see an ambition and a shift to emotional connection and the
psychology of the consumer. That’s how brands will differentiate.” 318
Another panelist, Aubrey Flynn, SVP and chief digital officer of REVOLT TV &
Media, believes Millennials and Gen Z not only want purpose in their lives, but
they want the brands they use to share that purpose. 318 “To understand each
person’s individual purpose, brands need to move away from demographics and
get closer to psychographics,” says Flynn.318 “In order to know people on an
intimate level, companies will likely start investing in the study of human
behavior to find authentic ways of personalizing experiences,” explains
Abramovich.318
Today, Facebook is far, far down the psychographics road. Although Facebook
and the now defunct Cambridge Analytica got into a lot of trouble by harvesting
Facebook data to sway political elections, the lessons and tactics learnt during
the turbulent year of 2016 are far too powerful to be ignored by future
advertisers. There is a way to utilize Facebook data that is either considerate of
privacy concerns or anonymized all-together and advertisers are currently
salivating at the prospect of getting their hands on all that incredibly important
psychographic detail.
Flynn believes that authenticity also is key when it comes to advertising to the
Millennial and Gen Z demographic.318 “We market a lifestyle, and bringing that
to life means different things to different people,” she says. 318 “Telling people
about your company is one thing, she adds, but empowering audiences to
successfully pursue the purposes that are important to them is a totally different

241
 ANDREW W. PEARSON

type of engagement.”318 As companies learn about the drives and motivations of

their customers, personalization will be key.318
“Most of the solutions to measure emotions are in beta, so it’s still the early
stages,” says Abramovich. However, “the ability to understand not only how long
someone engaged with an ad for but also how it made them feel is going to give
advertisers an unprecedented understanding of the effectiveness of their ads,”
concludes Abramovich.318

Geofencing Applications
Today, most smart phones have geofencing capabilities that tap into GPS or RFID
technology to define geographical boundaries. Basically, geofencing programs
allow an administrator to set up triggers — usually SMS push notifications or
email alerts — so when a device crosses a “geofence” and enters or exits a set
area, a user is notified. Applications such as Facebook, Foursquare and China’s
WeChat and Jiepang use geofencing to locate users, as well as help users find
their friends and/or check into physical places.
As TechTarget explains, geofencing has many uses, including319:
• Mobile Device Management — When a host’s tablet PC leaves the
cruise line property an administrator receives a notification so the
device can be disabled.
• Fleet management — When a truck driver breaks from his route, the
dispatcher receives an alert.
• Human resource management — An employee smart card will send an
alert to security if an employee attempts to enter an unauthorized area.
• Compliance management — Network logs record geofence crossings to
document the proper use of devices and their compliance with
established rules.
• Marketing — A retail business can trigger a text message to an opted-
in customer when the customer enters a defined geographical area.
• Asset management — An RFID tag on a pallet can send an alert if the
pallet is removed from the warehouse without authorization.
With geofencing applications, “users can also offer peer reviews of locations,
which add a layer of user-generated content. In exchange for loyalty, more and
more businesses — from local cruise lines to larger organizations like Bravo TV,
Starbucks and The History Channel — are offering coupons, discounts, free
goods and marketing materials.”320
As users continue to enter personal details as well as update and check-in to
their locations, geofencing applications like Foursquare can “collect a historical
view of consumer habits and preferences and, over time, possibly recommend a

242
 THE A.I. CRUISE LINE

much larger variety of targeted marketing materials in real time — as a consumer

walks into a store to look for a specific item or service.”320
In their paper On the Potential Use of Mobile Positioning Technologies in Indoor
Environments321, Giaglis et al. claim there are six different types of service uses
for mobile positioning technology (see Table 10). Geofencing applications (aka
Location Based Services (LBS)) like Jiepang and Foursquare are useful services for
hotel and cruise line marketers as well. Macau cruise lines, specifically, should
be exploiting this medium because of its high concentration of mobile
subscribers.322
APPLICATION
SERVICES EXAMPLES ACCURACY NEEDS
ENVIRONMENT
Emergency calls Medium to High Indoor/outdoor
EMERGENCY
SERVICES Automotive Medium Outdoor
Assistance
Traffic
High Outdoor
Management
NAVIGATION
Indoor Routing Medium Outdoor
SERVICES
Group
Low to Medium Indoor
Management

Travel Services Medium to High Outdoor

INFORMATION Mobile Yellow

Medium Outdoor
SERVICES Pages
Infotainment
Medium to High Outdoor
Services
MARKETING Banners, Alerts, Medium to High Outdoor
SERVICES Advertisements
People Tracking High Indoor/Outdoor

Vehicle Tracking Low Outdoor

TRACKING
SERVICES Personnel Tracking Medium Outdoor

Product Tracking High Indoor

BILLING Location-sensitive Low to Medium Indoor/Outdoor

SERVICES billing

Table 10: Taxonomy of mobile location services

Source: Durlacher Research

243
 ANDREW W. PEARSON

In his article LBS Opportunities for Casino Marketers in Macau322, Chris Weiners
offers the following ideas for casino operators to get their LBS promotions
rolling:
1. Pick your LBS service and claim your location.
2. Offer tips to customers via LBS.
3. Reward loyalty creatively. Start by offering your most loyal customers
rewards, special access, and other promotions. Those that become your
“Mayor” — or any other significant title — should be rewarded for their
loyalty. This is a great way to identify potential social influencers and
utilize them to further promote your venue.
4. Reward new customers: First time check-ins should receive special
promotions or incentives as it is important to give people a reason to
continuously check in to your establishment.
5. Understand who your loyal customers are online, and work with them.
Develop a plan to utilize these ‘influencers’ and tap into their social
networks. “Casinos do it offline all of the time; develop a similar
approach for high-valued customers online through social connections.
Encourage your followers to promote their checked-in status to their
friends via social networks and micro blogs like Sina and Twitter.”322
6. Promote your services both on- and off-line.
In May of 2013, Lighthouse Signal Systems launched its indoor positioning
system as an open service for Android app developers.323 Developers can use the
technology to create Android apps that will help users find their way through the
vast indoor terrain of Las Vegas’ hotels and casinos.323
Although global positioning systems have made outdoor navigation as simple as
following directions on a mobile device, indoor navigation isn’t so simple, it is
actually one of the last major hurdles that smartphones have yet to truly
conquer.323 However, Cambridge, Mass.-based Lighthouse Signal Systems has
launched a service that covers 20 million square feet of entertainment and retail
space at leading casinos and hotels on the Las Vegas Strip.323
Lighthouse is “making its service freely available to Android app developers,
resort operators, cruise lines, and others seeking to enhance the visitor
experience in Las Vegas. Indoor navigation is the Holy Grail for the mobile
industry, and Lighthouse says it is the first to provide GPS-like indoor positioning
on a wide scale in a major U.S. metro.”323
“We are excited to support app developer partners as they create new mobile
experiences with indoor positioning in Las Vegas, where large resort interiors
have traditionally presented a vexing challenge for visitors,” said Lighthouse co-
founder Parviz Parvizi.323
The standard line is that casinos create circular floors that differ little from
whichever direction you enter or exit them so that patrons will get lost on them

244
 THE A.I. CRUISE LINE

and, therefore spend more money at the casino’s tables and slots, but times are
changing. Casino operators now recognize the importance of getting their
gamblers in front of their preferred gaming table or slot machine as quickly as
possible.
A line stretching out the door at the entrance of a casino in Singapore (because
every guest’s passport must be checked to ensure a Singapore local isn’t
attempting to slip in without paying the local’s entrance fee) means minutes of
lost gaming time per person, which can add up to thousands of dollars of lost
revenue per day.
Giving a gambler direction to his favorite slot machine bank or preferred
baccarat table could mean, at minimum, decreasing a player’s frustration at not
being able to find what he or she is looking for or, at best, increasing gaming floor
revenue by increasing the gaming handle.
“Providing location-based services does not really reduce how much time people
spend at the resorts but instead has the potential to enhance the overall
experience,” said Parviz Parvizi.323 “From a resort owner perspective, the time
that a visitor spends wandering around being lost is a wasted opportunity that
could be better and more profitably spent on gaming or entertainment.” 323
The truly massive nature of today’s integrated resorts also means helping people
navigate through them as quickly as possible will cut down on property wear and
tear. Moving twenty thousand people quickly and efficiently through the halls of
an integrated resort will help reduce wear and tear on things like carpets,
elevators, escalators, toilets, etc., thereby reducing operational costs for the
property quite substantially.
Lighthouse’s platform “includes indoor geofencing: a hosting platform for
location-based offers and user analytics.”323 The apps include user opt-in
agreements and developers cannot use the service to track mobile phone users
without user consent.323
The technology uses “a combination of WiFi fingerprinting and sensor data. As
long as there are WiFi networks in the area, Lighthouse can provide positioning
info.”323 Google, Cisco, Ekahau, Euclid, Shopkick, PointInside, Aisle411,
Sensionlab, Indoor.rs, Yfind, and CSR are all developing similar systems.323
Mobile marketing in general and OTT, MMS and SMS marketing in particular can
help cruise lines create a one-to-one, two-way interactive experience with its
customers. These channels are not just about sending out a simple message, but
rather they are about starting a customer relationship that can be analyzed so
that the cruise line has a 360-degree understanding of its patron. It is an
understanding that includes his or her wants, desires and needs.
Besides geo-fencing applications, social media channels like Facebook,
Instagram, Twitter, WeChat, as well as many others can reveal a patron’s

245
 ANDREW W. PEARSON

location. Instagram tracks a user’s photos even if he or she doesn’t geo-tag them.
As Cadie Thompson warns in her article Social media apps are tracking your
location in shocking detail324, “While the picture sharing app does give users the
option to name the location of where they are uploading an image, it also
geotags an uploaded pic regardless if the user has selected the ‘Add to Photo
Map’ function.”324
Foursquare's check-in app Swarm also broadcasts users’ location even if they
have not selected a specific location for check-in.324 Many live-streaming apps
like Periscope, YouTube, and several Chinese ones will also show the location of
the user and this is information that can be utilized by an IR’s marketing
department if it can exploit the information quickly enough.
Check-ins and geo-posts from sites like Foursquare, WeChat, Instagram, Facebook,
WhatsApp, YouTube, as well as a whole host of other social networks can help
cruise line operators connect with a nearby audience. Underlying these check-ins is
a treasure-trove of collected data. As Aaron Gell explains in his New Yorker article
The Not-so-Surprising Survival of Foursquare 325, “Foursquare’s stockpile of
location-data breadcrumbs has allowed the company to steadily augment its
map of the world, and to test the fuzzy signals it receives from users’ phones (the
service gleans from G.P.S., Wi-Fi, and Bluetooth, and from other markers)
against the eleven billion definitive check-ins provided by its users over the past
seven years.”
“According to Mike Boland, a chief analyst at the market-research firm
BIA/Kelsey, Foursquare can now pinpoint a phone’s location with an accuracy
that matches, and may in some cases surpass, that of much larger rivals,” notes
Gell.325 “Facebook has a much larger sample of data points,” Bolands says, but
“Foursquare has more accurate and reliable data.”325 Foursquare claims its map
now “includes more than a hundred million locations, many of them in tightly
crowded areas, like office buildings and malls, that other services still struggle to
identify.”325 “The accuracy of Foursquare’s Places database has led more than a
hundred thousand other apps and developers — including Snapchat, Twitter,
Pinterest, Uber, and Microsoft — to use its application programming interface
(API) to power their own features,” notes Gell.325
Foursquare in particular is taking this foot-traffic data very seriously. On March 20,
2017, the social network announced a new service, Foursquare Analytics, a foot-
traffic dashboard for brands and cruise lines.326 The platform was made available
to retailers with any number of stores, no matter how small.326 With the service,
retailers could “use the dashboard to see foot-traffic data across metrics like
gender, age and new versus returning customers — on a national or citywide
scale.”326 “Retailers can also compare their foot traffic against a set of
competitors and their category as a whole,” adds Tepper.326
“The data is collected via Foursquare’s existing database of locations (which

246
 THE A.I. CRUISE LINE

powers more than 100,000 apps, including Snapchat), as well as anonymized in-
store-visit data collected from users of Swarm and Foursquare who have opted
in to always-on location sharing. Foursquare then normalizes this data to make
sure it accurately represents the U.S. population as a whole,” explains Tepper. 326
If you want to see where the future of location analytics might be headed,
Foursquare’s ‘Hypertrending’ application, which was rolled out at SXSW 2019,
would be a good place to start. As Dennis Crowley, executive chairman of
Foursquare, explains in his company blob Introducing Hypertrending, Where 10
Years of Foursquare Has Led Us327:
“Hypertrending is a top-down view of all the places and phones
that Foursquare knows about in Austin. The ‘Map’ view gives
you a real-time look at how people are spread throughout the
city — each dot represents a different place, the size of each
dot corresponds to the number of people at each place, and
each color represents a different type of place. If you see it on
the map, you’re seeing it live. The ‘Top 100’ view charts places
and events as they trend up or down in busy-ness (based on the
number of phones inside those places) while the up/down
arrows represent whether that place or event has become
more or less busy in the past 30 minutes.”
Hypertrending is powered by Foursquare’s “Pilgrim” technology, which allows
Foursquare to understand how phones move in and out of more than 100 million
places around the world.327 The data in Hypertrending comes from Foursquare’s
“first-party panel” — "a mix of data from our own apps and other apps that use
our technology.”327 As all the data flowing through Hypertrending is anonymized
and aggregated, Crowley claims that, “Hypertrending lets you see the movement
of the panel population as a whole, without showing you anything about any of
the individuals in the panel.”327
“Foursquare’s approach to analytics focuses on visits (not location trails), and
aggregated and anonymized data (not individuals’ location data),” states
Crowley.327 “Hypertrending only sees phones that are stopped at a specific place
(e.g. the convention center, the bar at the Driskill Hotel, Lamberts BBQ), so the
map won’t show people walking, driving, or otherwise in-between places,”
explains Crowley.327 Foursquare also filtered out stops at what it calls ‘sensitive’
areas, like homes or apartments, religious centers, divorce lawyers’ offices,
etc.327 This means users won’t see these types of places on Hypertrending
either.327
Foursquare only released Hypertrending during SXSW 2019 and hoped the demo
would pique the interest of developers and entrepreneurs, as well as inspire
them to build things with the tool.327 As Crowley explains, “Building city guides
and data viz comes naturally to us — but we want to see what the urban

247
 ANDREW W. PEARSON

planners, the game developers, the folks innovating with AR, etc. would do with
a Hypertrending-esque data set.”327

Going Viral
In their paper Two Hearts in three-quarter time: How to waltz the social
media/viral marketing dance50, Kaplan and Haenlein put forth the six steps of
what they call “waltzing the social media/viral marketing dance”50:
• One. . .way: Viral marketing goes social media.
• Two. . .concepts: Word-of-mouth and viral marketing.
• Three. . .conditions: How to create an epidemic.
• Four. . .groups: Social media viral marketing campaigns.
• Five. . .pieces of advice: Spreading the virus.
• Six. . .degrees of separation: From epidemics to immunity.
Kaplan and Haenlein reference one of humanity’s worse pandemics to explain
how potent social media virality can be50:
“The bubonic plague, also referred to as the Black Death, is
widely considered to be the deadliest pandemic in human
history. Between 1348 and 1350 it killed more than 35 million
people across Europe, corresponding to approximately 50,000
lives lost per day. Yet, as compared to more recent epidemics,
these figures seem modest; according to the U.S. Center for
Disease Control and Prevention (2010), approximately 60
million Americans contracted the H1N1 virus between April
2009 and April 2010 — more than 150,000 per day! Although
only 265,000 were actually hospitalized and 12,000 perished,
many of us won’t soon forget the panic surrounding this ‘swine
flu.’ Now, consider an epidemic of another sort. On July 14,
2010, Procter & Gamble uploaded a 30-second video spot via
the social media application YouTube, to promote its Old Spice
brand. This video, entitled The Man Your Man Could Smell Like,
was viewed 23 million times in 36 hours — representing 15
million ‘infections’ per day. If H1N1 had spread with the same
rapidity, 60 million infections would have been reached after
less than a week, and the 35 million casualties of the Black
Death would have taken no more time than a long weekend.”
Kaplan and Haenlein admit that “watching an online video is certainly not
comparable to getting infected by a potentially deadly disease. Nevertheless,
these numbers illustrate the incredible speed with which so-called ‘viral
marketing campaigns’ can spread at a time when social media start to rule the
world”46 Kaplan and Haenlein believe that, “Viral marketing allows firms to

248
 THE A.I. CRUISE LINE

promote their products and services with very low budgets and still reach the
same levels of awareness that are usually only achievable with high-frequency
TV advertising.”50
Through clever viral marketing low budget horror flicks like The Blair Witch
Project and Paranormal Activity became blockbusters.50 Kaplan and Haenlein
note that, “Brands such as Evian (Roller-Skating Babies), Burger King (Subservient
Chicken), and Old Spice have all benefited from viral marketing epidemics, while
JetBlue, Heinz Ketchup, and others have suffered severely at the same hands.”50
According to Wikipedia, word-of-mouth marketing (WOMM, WOM marketing)
“differs from naturally occurring word of mouth, in that it is actively influenced
or encouraged by organizations (e.g. 'seeding' a message in a networks
rewarding regular consumers to engage in WOM, employing WOM 'agents').”328
While it is difficult to truly control WOM, research329 has shown that there are
three generic avenues to 'manage' WOM for the purpose of WOMM:
1. Build a strong WOM foundation (e.g. sufficient levels of satisfaction,
trust and commitment).329
2. WOMM management which implies that managers only have a
moderate amount of control (e.g. controversial advertising, teaser
campaigns, customer membership clubs).329
3. Direct WOMM management, which has higher levels of control (e.g.
paid WOM 'agents', "friend get friend" schemes).329
WoM has been shown to substantially influence consumer attitudes and
behaviors, and to be up to seven times more effective than traditional print
advertising in impacting brand switching decisions.330 Kaplan and Haenlein
concede that, “Despite what one might initially think, WoM is not a purely
altruistic behavior: it offers advantages to the sender, as well as the receiver.”50
For senders, WoM is an opportunity to help others and to improve their self-
confidence in doing so.331 332 For receivers, it reduces decision-making time as
well as risk333, as friends tend to be perceived as unbiased sources of
information.332
According to academic research and Jonah Berger's bestselling book Contagious:
Why Things Catch On334, there are six key factors that drive what people talk
about and, ultimately, share. They are organized in an acronym called STEPPS,
which stands for:
• Social Currency – the better something makes people look, the more
likely they will be to share it.335
• Triggers – things that are top of mind (i.e., accessible) are more likely to
be tip of tongue.336
• Emotion – when we care, we share. High arousal emotions increase
sharing.335

249
 ANDREW W. PEARSON

• Public – the easier something is to see the more likely people are to
imitate it.336
• Practical Value – people share useful information to help others.334
• Stories – Trojan Horse stories carry messages and ideas along for the
ride.334
“Similar to traditional WoM, electronic WoM — such as book reviews exchanged
on pages like Amazon.com — has been shown to influence purchase behavior337
and to lead to the acquisition of higher value customers,338” claim Kaplan and
Haenlein.50 This presents interesting managerial implications, as the inherent
anonymity of online feedback mechanisms can make such platforms subject to
strategic manipulations by companies which would like to increase their sales
through favorable comments.339 340
Electronic WoM exchanged via newsgroups and the like are easily collected and
analytics for marketing research purposes.341 Kaplan and Haenlein believe that,
“This approach, which is referred to in the literature as netnography342, can lead
to valuable insights due to its ability to observe consumers in an unobtrusive
way.”50
According to Kaplan and Haenlein, when compared to traditional WoM,
electronic WoM has two main advantages; the first is a higher diffusion speed
for new pieces of information and the second is the fact that electronic WoM is
substantially easier to monitor than traditional WoM.50 “When WoM is
exchanged using traditional face-to-face communication, diffusion is limited by
the size of the social network each individual maintains,” claim Kaplan and
Haelein.50 Given that, on average, people have only three close friends 343, and a
total social network of no more than 150344, “chains of WoM communication and
customer referrals tend to die out quickly.”50 Unlike the limitations of personal
communication, “WoM exchanged electronically can reach a much larger group
of other customers.”50 Because WoM is so easily monitored, calculating an ROI
on marketing measures should be effortless.
Kaplan & Haenlein define viral marketing as “electronic word-of-mouth whereby
some form of marketing message related to a company, brand, or product is
transmitted in an exponentially growing way, often through the use of social
media applications.”50 Viral marketing’s two defining elements; “a growth, or
reproduction, rate greater than one; this implies that each receiver passes the
message to more than one other person.”50 “For example, when initially seeded
to one person, a viral marketing message with a reproduction rate of two would
be transferred to 2, 4, 8, 16, 32, 64 (et cetera) new people in the following
periods.”50 If the reproduction rate exceeds one, the resulting growth pattern
goes exponential50 This pattern can be observed in “business (e.g., compound
interest), physics (e.g., nuclear chain reactions), biology (e.g., bacterial growth),
and epidemiology (e.g., spread of a virus).”50

250
 THE A.I. CRUISE LINE

Kaplan and Haenlein state that, “The second characteristic usually associated
with viral marketing is use of social media applications.”50 Social media can be
defined as ‘‘a group of Internet based applications that build on the ideological
and technological foundations of Web 2.0, and that allow the creation and
exchange of User Generated Content.”46 As Kaplan & Haenlein note, Web 2.0 “is
an umbrella term describing different types of applications such as collaborative
projects (e.g., Wikipedia), blogs/micro-blogs (e.g., Twitter), content communities
(e.g., YouTube), social networking sites (e.g., Facebook), virtual game worlds
(e.g., World of Warcraft), and virtual social worlds (e.g., Second Life).”46 “Social
media applications are particularly suited for viral marketing, as the community
element embedded in them makes it convenient to transmit the marketing
message to a large group of people,” argue Kaplan & Haenlein.46 Some
researchers therefore use the terms ‘viral marketing’ and ‘social media
marketing’ interchangeably.345
Viral marketing is a powerful but quite recent phenomenon. The literature
review includes a variety of different terminologies such as word-of-mouse346,
buzz marketing347, stealth marketing348, and word-of-mouth marketing345. The
term ‘viral marketing’ was borne of an article written by Harvard Business
School’s Jeffrey Rayport.349 Published in the business magazine Fast Company,
‘‘The Virus of Marketing’’ “makes reference to the exponential growth pattern
inherent in viral marketing by comparing diffusion of the marketing message
with the spread of a virus.”50 “Successful viral marketing should lead to a growth
pattern similar to major epidemics such as the Black Death in the 14th century,
Spanish Flu in the 20th century, and Swine Flu in the 21st Century,” say Kaplan
& Haenlein.50 Unlike infectious viruses that cause disease, the more resistant and
durable a viral marketing virus is, the better!50
For a marketing message to go viral, Kaplan & Haenlein believe three basic
criteria must be met: the message has to be memorable, the right people need
to get the right message under the right circumstances, and the environment for
the acceptance of the message has to be right.50
“The first critical element in creating a viral marketing epidemic entails finding
the right people to spread the message,” claim Kaplan & Haenlein.50 Consistent
with classical laws of concentration and the Pareto Principle, “20% of
messengers can be expected to carry 80% of the load; it is, therefore, especially
crucial to select wisely the initial hosts for the epidemic.”50 “Three groups of
messengers are required to ensure the transformation of an ordinary message
into a viral phenomenon: market mavens, social hubs, and salespeople.”50
For Feick and Price, “Market mavens are defined as individuals who have access
to a large amount of marketplace information, and proactively engage in
discussions with other consumers to diffuse and spread this information.”350 As
individuals in tune with the cultural zeitgeist, “market mavens are typically
among the first to receive the message and transmit it to their immediate social

251
 ANDREW W. PEARSON

network.”50
“Once a market maven hands over the message to a social hub, a viral epidemic
has begun,” claim Kaplan & Haenlein.50 Social hubs are defined as people with
an exceptionally large number of social connections.351 “They often know
hundreds of different people and have the ability to serve as connectors or
bridges between different subcultures. The exceptional social network of social
hubs can facilitate immediate transmission of the message to hundreds, if not
thousands, of other consumers.”50
For Kaplan and Haenlein, sometimes a “direct link between a market maven and
a social hub is just not enough.”50 “While market mavens may know the message
earlier than others, they might not be particularly convincing in transmitting the
information.”50 “In this case, salespeople could be needed to receive the
message from the market maven, amplify it by making it more relevant and
persuasive, and then transmit it to the social hub for further distribution.”50
Regarding the message, “Even the most perfect combination of market mavens,
social hubs, and salespeople is of limited value when the news itself is not
adapted to become viral. Only messages that are both memorable and
sufficiently interesting to be passed on to others have the potential to spur a
viral marketing phenomenon.”50
“Making a message more memorable and interesting, or simply more infectious,
is often not a matter of major changes but minor adjustments,” suggest Kaplan
and Haenlein.50 “One option is to rely on true stories about real people (‘My
brother has a friend, John Doe…’), which are often more persuasive than
corporate advertising, recommend Kaplan and Haenlein.50
Kamins, Folkes, & Perner recommend “using rumors, especially positive ones
that reflect well on the person telling them, as they have a particularly high
chance of being transmitted to others.”352 Dobele et al. recommend “obvious
safe bets like practical short lists (e.g., ‘The ten best ways to lose weight’),
humorous or even hilarious messages, and sex.”353 Dobele et al. believe
messages with viral potential must trigger an emotional response in the
receiver.353 “Effective messages often contain an element of surprise, combined
with other emotions that can be either positive (e.g., joy) or negative (e.g.,
disgust, fear),” argue Kaplan and Haenlein.50
So many pieces of the virality puzzle must come together at just the right time,
through just the right channels, with just the right influencers that it’s almost
impossible to predict virality. For this reason, marketers who try to make their
content go viral should accept the fact that the process is highly unpredictable;
everything that can be done to ensure a marketing campaign should be done,
but virality is not easy to replicate so failure shouldn’t be viewed in as harsh a
light as normal advertising campaign failures.

252
 THE A.I. CRUISE LINE

Besides getting the right message to the right people, two other environmental
conditions — uniqueness and plain old good luck — make the difference
between viral marketing success or failure.50 “Messengers will only pass on the
message when they think it’s not already something everyone knows about,” say
Kaplan and Haenlein.50 Companies often fail getting virality because they spread
the initial message too broadly.50 “Instead of concentrating on having as many
seeds as possible, firms should instead focus on having an infectious message
(leading to a high reproduction rate) and seeding it to many disconnected
subcultures,” recommend Kaplan and Haenlein.50
Some plain old good luck is also “needed to glue everything together, as it’s often
just not the right time and/or place to launch a viral marketing campaign,” say
Kaplan and Haenlein.50 “This ambiguity makes viral marketing hard to
understand for companies: actions which worked well in the past, or for one’s
competitor, may simply be ineffective in a specific case.”50
Kaplan and Haenlein offer up the example of a May 2009 Starbucks viral
marketing campaign — "The coffee vendor encouraged its customers to take
pictures of themselves in front of the company’s new billboards, and post the
shots to the micro-blogging application, Twitter.”50
At around the same time, “film producer and political activist Robert Greenwald
saw this as a perfect opportunity for promoting his latest documentary about
unfair labor practices at the coffee chain.” 50 he also asked people to take
pictures of themselves, “but while holding signs criticizing the company’s
practices.”50 “Many responded to Greenwald’s calling, and soon about half the
photos distributed on Twitter were very different from those initially intended
by Starbucks,” explain Kaplan and Haenlein.50
It’s unclear why things like this happen, it can be a vast mix of sociological,
political, behavioral and/or a hundred other reasons. Sometimes, as Kaplan and
Haenlein, plainly state, it’s just not your day!50
Viral marketing campaigns emerge from an interaction between a firm and its
customer base.50 The initiator could be either the company or a group of its
consumers.50 Like any other marketing action, viral marketing campaigns can
obviously result in positive or negative reactions.50 Kaplan and Haenlein argue
that, “Combining these two dimensions results in four different types of viral
marketing campaigns: nightmares, strokes-of-luck, homemade issues, and
triumphs.”50
Nightmares include the case of JetBlue Valentine’s Day 2007 promotion.50 When
a JetBlue flight from New York to Cancun was delayed on the tarmac due to a
brutal ice storm, it took JetBlue nearly nine hours to defrost the plane, which
resulted in a near-complete breakdown of JetBlue’s operations. 50 Thousands of
flights had to be cancelled and hundreds delayed. Because of internal IT system
problems, flight crews couldn’t be rescheduled.50

253
 ANDREW W. PEARSON

Unsurprisingly, passengers were outraged, venting their anger on social media

sites; blogs, micro-blogs (e.g., Twitter), vlogs and social networking sites (e.g.,
Facebook) filled up with torrid complaints.50 The fury lasted several days and
“JetBlue had to cancel nearly a quarter of all its flights the following weekend.”50
“The impact of this incident was even more substantial because JetBlue had
worked hard in the previous months to build a positive image and an excellent
reputation,” note Kaplan and Haenlein.50 “Business Week even wanted to crown
the firm as one of four companies with the best customer service, and planned
to promote this prominently on its cover page.”50 Yet, due to the “‘worst
operational week in JetBlue’s seven year history’”354, the magazine balked at the
last minute and replaced JetBlue with cruise line Nordstrom.50
“On Monday, February 19, David Neeleman issued a public apology for the
cancellations and sent a letter to the airline’s clients. Instead of trying to find
excuses or to cover up the situation, he admitted that he was ‘humiliated and
mortified’ by the systems failure, and expressed his deep regrets,” recounts
Kaplan and Haenlein.50 Falling on his sword and admitting guilt worked for
Neeleman and JetBlue has thrived since.
Even though Kaplan and Haenlein believe that “Viral marketing is as much an art
as it is a science,”50 there are several ways of increasing the odds of going viral.
There are some basic rules companies should follow when spreading a virus,
including the following50:
1. Viral marketing is only as good as the remaining marketing mix.
2. Viral marketing needs to be backed up by traditional forms of
communication.
3. Excessive planning and intervention kill any viral marketing campaign.
4. Highly provocative and edgy messages are a tricky business.
5. Successful viral marketing requires a little bit of luck and gut feeling.
Despite all of viral marketing’s advantages, one has to be realistic.50 Even the
most endemic viral buzz won’t be able to sell a worthless product. “To reveal its
true potential, viral marketing needs to be accompanied by changes in the rest
of the marketing mix,” advise Kaplan and Haenlein. Consider Burger King’s
Subservient Chicken campaign for some inspiration.50 “In 2004, Burger King
launched a viral marketing epidemic around an interactive website
(www.subservientchicken.com) where users could give commands to a human
dressed in a chicken costume.”50 The program instantly went viral, partly
“because the advertising agency in charge of it also modified other major
elements of the firm’s marketing mix.”50 “According to CP+B President Jeff Hicks,
the company redesigned most anything it could: from employee uniforms, to
drive-through areas, to ketchup packets. Viral marketing might draw customers
to stores, but they need convincing reasons to come back once the hype is
over!”50

254
 THE A.I. CRUISE LINE

Viral marketing buzz tends to die out quickly, and it usually doesn’t last longer
than a couple of weeks at best.50 Viral marketing is all about excitement, and
nothing is as boring as yesterday’s news.50 Kaplan and Haenlein argue that, “To
maintain momentum, firms therefore need to complement viral marketing with
more traditional forms of communication.”50 A good example of this is
Wilkinson’s Fight for Kisses advertisement.50 “To support the launch of its newly-
developed disposable razor, Quattro Titanium, the company relied on a viral
marketing story about a baby fighting his father for kisses from the baby’s
mother.”50 This campaign consisted of an amusing “animated video and an
interactive computer game, combined with a series of press announcements,
radio spots, and sponsorship of the France — Ireland rugby match that took
place during the same time period.”50 Despite its limited US$90,000 budget, “the
campaign was a huge success and resulted in a five percentage point market
share increase within the target group.”50 “If all forces act in concert, they can
indeed move mountains!” claim Kaplan and Haenlein.50
As with all communication exercises, “viral marketing campaigns need to be
carefully planned prior to their launch.”50 Once unleashed, the virus is set free,
and less control and/or intervention is preferable.50 In particular, companies
should never ask their customers to spread the virus if they are reluctant to do
so.50 A truly compelling viral marketing campaign stands on its own and develops
its own dynamics.50 A good case in point is Evian’s Roller-Skating Babies
campaign, which has been crowned as the Guinness Book of Records most
viewed advertising spot, with more than 45 million online views.50 As Kaplan and
Haenlein relay, “The company engaged in careful planning prior to launching the
video by choosing the right topic (Evian already had an advertising campaign
based on babies 10 years earlier), the right music (a remix of a 30-year-old rap
song), and the right messengers. But once the virus had been unleashed, Evian
limited its role to reacting to — instead of proactively influencing — the viral
phenomenon.”50 The reality is the job of the creator is to make a piece of
advertising that can go viral, not helping it to go viral once it’s released. If the
advert is good enough, the viewers will gladly do the rest.
Marketers should keep in mind that they must walk a fine line. “Good viral
marketing messages need to be both memorable and interesting.”50 However,
“firms must exercise caution and beware of using messages that are too
provocative; there is often a very fine line between being provocative and being
inappropriate.”50
“Computer manufacturer Microsoft learned this the hard way during promotion
of its Perfect Dark Zero game for the Xbox 360 platform,” explain Kaplan and
Haenlein.50 “In the context of a viral marketing campaign, users were invited to
provide the name and email address of a person who Joanne Dark, the assassin
within Perfect Dark Zero, should ‘take care’ of. This person subsequently
received a message with a video, showing a body wearing a toe tag bearing the

255
 ANDREW W. PEARSON

recipient’s name.”50
Quite a provocative idea indeed, but way, way beyond the line of decency. As
expected, most people who received the message reacted with either shock or
disgust.50 “Unless a company and brand want to be remembered for bad taste,
they had better be careful with messages that are too edgy,” recommend Kaplan
and Haenlein.50
Executives should accept “that the transition between careful planning and viral
marketing success is subject to ‘random’ noise and that failure is always a
possibility, even with the best planning and best intentions.”50 “What worked
well yesterday, or is working well for the competition, does not necessarily
guarantee success today,”50 as Heinz Ketchup discovered to its consternation
with one Pepto-Bismol campaign.50
“In 2007, Procter & Gamble organized a highly successful video contest, Be the
Next Pepto Star. Users were encouraged to create funny 60-second videos
portraying the five symptoms aided by Pepto-Bismol (i.e., nausea, heartburn,
indigestion, upset stomach, and diarrhea) and to upload them to YouTube,” say
Kaplan and Haenlein.50
Heinz Ketchup lazily tried to copy the campaign several years later, and the
results were disastrous.50 “People uploaded videos in which they used Heinz
Ketchup as toothpaste or acne cream, and publicly accused the firm of looking
for cheap labor to create ads,”50 not a look any company wants.
One thing seems certain, conclude Kaplan and Haenlein, “viral marketing is still
in the early stages of its life cycle. And, the tremendous potential it offers
companies at very limited cost should make every executive think seriously
about engaging in this new form of communication.”50 However, careful
planning and deep sincerity are two must-have ingredients to ensure a viral
campaign has a chance for success.

Conclusion
Advertising has always been looked down upon, but today it is probably more
important than ever. The Canadian humorist Stephen Leacock once described
advertising as “the science of arresting the human intelligence long enough to
get money from it.”355 The unfortunate reality is advertising is very much a
necessary evil. Will Rogers, one of America’s greatest satirists, put it succinctly
when he said, “One ad is worth more to a paper than forty editorials.” For cruise
lines, which often must spend hundreds of millions of dollars to create massive
ships, advertising is more than a necessary evil, it is an operational imperative.
More than ever before, AI allows marketers to reach consumers at every stage
of the buying process based on their interests and demographics. In his article

256
 THE A.I. CRUISE LINE

How AI Will Change Marketing as We Know It356, Amine Bentahar claims that,
“One particular example of how AI increases the efficiency of marketing is by
making it easier to put customers into distinct groups that will allow for added
segmentation to highly targeted niches.” This means that, “Rather than creating
one ad campaign that you hope will reach your target customers, marketers will
instead be able to create more personalized, natural marketing content that will
be unique for each targeted customer segment.”356
AI will also “allow for more truly data-driven marketing campaigns, where AI will
allow data to be more properly used and integrated into each ad campaign.” 356
The flip side of collecting as much data as a company possibly can is brands are
collecting more data than they can actually use and here, too, AI can help by
giving brands “the ability to seek out and identify patterns that will be beneficial
for marketers in their campaigns.”356
From a content standpoint, AI can help brands keep track of what type of content
consumers are most interested in, which is information that can then be used to
curate a website for each individual user.356 This should help with customer
conversions and should be a part of any brand customer personalization
initiatives.
Overall, AI can help marketers “look at things through a broader, more big-
picture lens.”356 Increased AI use won’t necessarily replace marketing teams, it
will simply allow them to work proactively, as well as help them focus on big-
picture decisions and strategies.356 Creatives will be allowed to be creative once
again.
AI will unquestionably be changing the marketing world as we know it, but
change can be good.356 “AI will allow marketers to create more educated,
personalized campaigns to reach consumers, all while viewing their work
through a big-picture focus that will allow them to be more creative. That is how
the biggest gains possible will be captured through AI,” concludes Bentahar.356
Brands should recognize that there is a radical reorganization of platforms and
delivery channels going on right now as well. All the major software, analytics,
and tech vendors are looking at new ways to monetize their businesses and
marketing is something they are focusing on. Besides Amazon, Microsoft, AT&T,
AOL, Verizon, new players like Roku are getting into the direct advertising
business. Even companies like the now publicly traded company Uber are getting
into the ad business.
Is all of this advertising expense worth it? Well, Henry Ford, the inventor of the
first mass produced car, felt it was not only necessary but imperative; “Stopping
advertising to save money is like stopping your watch to save time,” he wrote. In
the cruise line business, I’d argue that it’s more imperative than most other
industries because a cruise line’s product differs little from one cruise line floor
to another, even from one country to another.

257
 ANDREW W. PEARSON

The psychologist Stuart Henderson Britt put it in a more succinct way, “Doing
business without advertising is like winking at a girl in the dark. You know what
you are doing, but nobody else does.”357 Advertising is very much a seduction
and I think we can all agree that advertising to someone who physically can’t see
or respond to the advertising is a worthless endeavor.
The next chapter tackles social media in both a listening and social media
marketing way. Chapter six is about IoT and operations, explaining how to add
Hadoop and real-time processing to a cruise line’s IT systems. Chapter seven and
eight bring everything together and details the exact way cruise lines and IRs can
fully implement the solutions discussed here.

181 Thompson, Derek. (2018) The Atlantic. The Media’s Post-Advertising Future Is Also Its
Past. December 31, 2018. https://www.theatlantic.com/ideas/archive/2018/12/post-
advertising-future-media/578917/ (Accessed 1 January 2019).
182 Ballentine, Claire. (2018) New York Times. 12 August 2018. Google-Facebook
Dominance Hurts Ad Tech Firms, Speeding Consolidation.
https://www.nytimes.com/2018/08/12/technology/google-facebook-dominance-hurts-
ad-tech-firms-speeding-consolidation.html (Accessed 1 January 2019).
183 Mickel, Tripp & Wells, Georgia. (2018) Wall Street journal. Apple Looks to Expand
Advertising Business with New Network for Apps. June 1, 2018.
https://www.wsj.com/articles/apple-looks-to-expand-advertising-business-with-new-
network-for-apps-
1527869990?utm_source=newsletter&utm_medium=email&utm_campaign=newsletter
_axiosmediatrends&stream=top (Accessed 1 January 2019).
184 Fischer, Sara. (2018) Axios. The Next Big TV Tech Platform: Roku. 22 May 2018.
https://www.axios.com/roku-is-trying-to-become-software-platform-e8602b67-5af4-
443e-affc-634b8d553aea.html (Accessed 1 January 2019).
185 Shields, Ronan. Adweek. Adobe’s New Partnership with Roku Will Help Advertisers
Engage with 27 Million OTT Viewers. https://www.adweek.com/programmatic/adobes-
new-partnership-with-roku-will-help-it-engage-with-27-million-ott-viewers/ (Accessed
27 March 2019).
186 Fischer, Sara. (2018). Axios. How Media Companies Lost the Advertising Business.
June 5, 2018. https://www.axios.com/facebook-google-duopoly-advertising-tech-giants-
media-e382e5e2-21eb-4776-93c0-7d942ba80ada.html (Accessed 1 January 2019).
187 Slefo, George P. (2018). Quartz About-Face on Programmatic Ads Have Big
Implications for Digital Publishing. Ad Age. 8 May, 2018.
https://adage.com/article/digital/publishers-reminder-message-quartz-adopts-
programmatic/313433 (Accessed 24 October 2019).
188 Thompson Derek. (2018). The Atlantic. Where Did All the Advertising Jobs Go?
February 7, 2018. https://www.theatlantic.com/business/archive/2018/02/advertising-
jobs-programmatic-tech/552629/ (Accessed 1 January 2019).
189 Abramovich, Giselle. (2018). Experts Weight in On the Future of Advertising. CMO.
September 19, 2019. https://www.cmo.com/features/articles/2018/6/14/welcome-to-
the-future-of-advertising-cannes18.html#gs.4BvSdQHt (Accessed 15 February 2019).

258
 THE A.I. CRUISE LINE

190 Bartels, Robert (1988), The History of Marketing Thought, 3rd ed. Columbus, OH:
Publishing Horizons.
191 Wedel, Michel and Kannan, P.K. (2016) Marketing Analytics for Data-Rich
Environments. Journal of Marketing: November 2016, Vol. 80, No. 6, pp. 97-121.
https://www.rhsmith.umd.edu/files/Documents/Departments/Marketing/wedel-
kannan-jm-2016-final.pdf (Accessed 4 November, 2017).
192 Reilly, W.J. (1929), Marketing Investigations. New York: Ronal Press Company.
193 Starch, Daniel (1923), Principles of Advertising. Chicago: A.W. Shaw Company.
194 Nixon, H.K. (1924), “Attention and Interest in Advertising,” Archives de Psychologie,
72 (1), 5–67.
195 Guadagni, Peter M. and John D.C. Little (1983), “A Logit Model of Brand Choice
Calibrated on Scanner Data,” Marketing Science, 2 (3), 203–38.
196 Shaw, Robert (1987), Database Marketing, Gower Publishing Co.
197 Winer, Russell S. and Scott A. Neslin, eds. (2014), The History of Marketing Science.
Hackensack, NJ: World Scientific Publishing.
198 Ferber, Robert (1949), Statistical Techniques in Market Research. New York: McGraw-
Hill.
199 Bartels, Robert (1988), The History of Marketing Thought, 3rd ed. Columbus, OH:
Publishing Horizons.
200 Raiffa, Howard and Robert Schlaifer (1961), Applied Statistical Decision Theory.
Boston: Clinton Press.
201 Green, Paul E. (1963), “Bayesian Decision Theory in Pricing Strategy,” Journal of
Marketing, 27 (January), 5–14.
202 Bass, Frank (1969), “A New Product Growth for Model Consumer Durables,”
Management Science, 15 (5), 215–27.
203 Massy, William F., David B. Montgomery, and Donald G. Morrison (1970), Stochastic
Models of Buying Behavior. Cambridge, MA: MIT Press.
204 Little, John D.C. and Len M. Lodish (1969), “A Media Planning Calculus,” Operations
Research, 17 (1), 1–35.
205 Lodish, Leonard M. (1971), “CALLPLAN: An Interactive Salesman’s Call Planning
System,” Management Science, 18, P-25–40.
206 Little, John D.C. (1970), “Models and Managers: The Concept of a Decision Calculus,”
Management Science, 16 (8), B-466–485.
207 Nakanishi, Masao and Lee G. Cooper (1974), “Parameter Estimation for a
Multiplicative Competitive Interaction Model: Least Squares Approach,” Journal of
Marketing Research, 11 (August), 303–11.
208 Coombs, Clyde (1950), “Psychological Scaling Without a Unit of Measurement,”
Psychological Review, 57, 148–58.
209 Green, Paul E. and Srinivasan, V. (1978), “Conjoint Analysis in Consumer Research:
Issues and Outlook,” Journal of Consumer Research, 5 (2), 103–23.
210 Louvie ŕ e, Jordan J. and Woodworth, George. (1983), “Design and Analysis of
Simulated Consumer Choice or Allocation Experiments: An Approach Based on Aggregate
Data,” Journal of Marketing Research, 20 (November), 350–67.
211 Luce, R. Duncan and John W. Tukey (1964), “Simultaneous Conjoint Measurement: A
New Scale Type of Fundamental Measurement,” Journal of Mathematical Psychology, 1
(1), 1–27.
212 Kannan, P.K. and Gordon P. Wright (1991), “Modeling and Testing Structured
Markets: A Nested Logit Approach,” Marketing Science, 10 (1), 58–82.
259
 ANDREW W. PEARSON

213 Gupta, Sunil (1988), “Impact of Sales Promotions on When, What, and How Much to
Buy,” Journal of Marketing Research, 25 (November), 342–55.
214 Schmittlein, David C. and Robert A. Peterson (1994), “Customer Base Analysis: An
Industrial Purchase Process Application,” Marketing Science, 13 (1), 41–67.
215 DeKimpe, Marnik G. and Dominique M. Hanssens (1995), “The Persistence of
Marketing Effects on Sales,” Marketing Science, 14 (1), 1–21.
216 Kamakura, Wagner A. and Gary J. Russell (1989), “A Probabilistic Choice Model for
Market Segmentation and Elasticity Structure,” Journal of Marketing Research, 26
(November), 379–90.
217 Wedel, Michel and Wayne S. DeSarbo (1995), “A Mixture Likelihood Approach for
Generalized Linear Models,” Journal of Classification, 12 (1), 21–55.
218 Rossi, Peter E., Robert E. McCulloch, and Greg M. Allenby (1996), “The Value of
Purchase History Data in Target Marketing,” Marketing Science, 15 (4), 321–40.
219 Andrews, Rick L., Andrew Ainslie, and Imran S. Currim (2002), “An Empirical
Comparison of Logit Choice Models with Dis- crete Versus Continuous Representations of
Heterogeneity,” Journal of Marketing Research, 39 (November), 479–87.
220 Rossi, Peter E. and Greg M. Allenby (2003), “Bayesian Statistics and Marketing,”
Marketing Science, 22 (3), 304–28.
221 Wedel, Michel and Pieters, Rik (2000), “Eye Fixations on Advertisements and Memory
for Brands: A Model and Findings,” Marketing Science, 19 (4), 297–312. White, Percival
(1931), Market
222 Ansari, Asim and Mela, Carl F. (2003), “E-Customization,” Journal of Marketing
Research, 40 (May), 131–45.
223 Montgomery, Alan L., Shibo Li, Kannan Srinivasan, and John C. Liechty (2004),
“Modeling Online Browsing and Path Analysis Using Clickstream Data,” Marketing
Science, 23 (4), 579–95.
224 Moe, Wendy W. and Trusov, Michael (2011), “The Value of Social Dynamics in Online
Product Ratings Forums,” Journal of Marketing Research, 48 (June), 444–56.
225 Rutz, Oliver J., Michael Trusov, and Randolph E. Bucklin (2011), “Modeling Indirect
Effects of Paid Search Advertising: Which Keywords Lead to More Future Visits?”
Marketing Science, 30 (4), 646–65.
226 Sharma, C. H. (2008). Mobile Advertising: Supercharge Your Brand in the Exploding
Wireless Market. John Wiley & Sons, Inc.
227 Kim, K. L. (2012). The typological classification of the participants' subjectivity to plan
the policy and strategy for the smart mobile market. Korean Management Review, 367-
393.
228 Deighton, J. &. (2009). Interactivity’s Unanticpated Consequences for Marketers and
Marketing. Journal of Interactive Marketing, 23, 4 - 10.
229 Gossage, Howard L., Rotzoll, Kim B., Graham, Jarlath, Mussey, June B. (1986). “Is there
any hope for advertising?”, Univ of Illinois Press.
230 https://www.forbes.com/sites/sap/2013/01/16/22-best-marketing-quotes-to-drive-
your-marketing-strategy/ (Accessed 10 December 2019).
231 Sharma, R. S. (2009). The Economics of Delivering. Journal of Media Business Studies,
1-24.
232 Nielsen Company. (2012). Global Trust in Advertising and Brand Messaging. Nielsen
Company.
233 Dushinski, K. (2012). The Mobile Marketing Handbook. Information Today, Inc.

260
 THE A.I. CRUISE LINE

234 Chew, Christie. (2018). The Art and Science Behind Every “Add to Cart.” Adobe Blog.
https://theblog.adobe.com/the-art-and-science-behind-every-add-to-cart/ (Accessed 19
January 2019).
235 Hauser, John R., Urban, Glen, Liberali, Guilherme, Braun Michael. (2009). “Website
Morphing.” Marketing Science 28.2 (2009): 202-223. © 2009 Informs.
https://www.researchgate.net/publication/41822749_Website_Morphing (Accessed 16
February 2019).
236 Hayes, J., C. W. Allinson. 1998. Cognitive style and the theory and practice of
individual and collective learning in organizations. Human Relations 31(7) 847–871.
237 Witkin, H. A., C. Moore, D. Goodenough, P. Cox. 1977. Field dependent and field-
independent cognitive styles and their educational implications. Rev. Educational Res.
47(1) 1–64.
238 Riding, R. J., S. Rayner. 1998. Cognitive Styles and Learning Strategies Understanding
Style Differences in Learning and Behavior. David Fulton Publishers, London.
239 Ansari, A., C. F. Mela. 2003. E-customization. J. Marketing Res. 40(2) 131–145
240 Montgomery, A. L., S. Li, K. Srinivasan, J. Liechty. 2004. Modeling online browsing and
path analysis using clickstream data. Marketing Sci. 23(4) 579–585.
241 Riding, Richard J., Rayner, Stephen (1998), Cognitive Styles and Learning Strategies:
Understanding Style Differences in Learning and Behavior, (London, UK: David Fulton
Publishers).
242 Allinson, Christopher W. and Hayes, John. (1996), “The Cognitive Style Index: A
Measure of Intuition-Analysis for Organizational Research,” Journal of Management
Studies, 33, 1, (January), 119-135.
243 Kirton, Michael J. (1987), Adaption-Innovation Inventory (KAI) Manual (Hatfield, UK:
Occupational Research Centre).
244 Riding, Richard J., and Indra Cheema (1991), “Cognitive Style: An Overview and
Integration”, Educational Psychology, 11, 3&4, 193-215.
245 Kopfstein Donald (1973), “Risk-Taking Behavior and Cognitive Style.” Child
Development, 44, 1, 190-192.
246 Siegelman, Ellen (1969), “Reflective and Impulsive Observing Behavior” Child
Development, 40, 4, 1213-222.
247 Lee, Kavan. Buffer. (2014) Psychological Studies That Will Boost Your Social Media
Marketing. 17 November 2014. https://blog.bufferapp.com/psychological-studies-
marketing (Accessed 14 February 2019).
248 https://en.wikipedia.org/wiki/Endowment_effect (Accessed 14 February 2019).
249 https://en.wikipedia.org/wiki/Reciprocity_(social_psychology)
250 https://www.influenceatwork.com/principles-of-persuasion/
251 Psychologist World. Foot-in-the-door technique as a Persuasive Technique.
https://www.psychologistworld.com/behavior/compliance/strategies/foot-in-door-
technique (Accessed 14 February 2019).
252 Plous, Scott (1993). The psychology of judgment and decision making. McGraw-
Hill. ISBN 978-0-07-050477-6.
253 Kahneman, Daniel, and Amos Tversky, "Choices, Values and Frames," American
Psychologist, April 1984, 39, 341–350.
254 McLeod, Saul. Solomon Asch – Conformity Experiment. Simply Psychology. December
28, 2018. https://www.simplypsychology.org/asch-conformity.html (11 March 2019).
255 Watson, Dorothy (1992). "Correcting for Acquiescent Response Bias in the Absence
of a Balanced Scale: An Application to Class Consciousness". Sociological Methods &
Research. 21 (1): 52–88. 261
 ANDREW W. PEARSON

256 Huber, Joel; Payne, John W.; Puto, Christopher (1982). “Adding Asymmetrically
Dominated Alternatives: Violations of Regularity and the Similarity Hypothesis”. Journal
of Consumer Research. 9 (1): 90-98.
257 https://psychology.iresearchnet.com/social-psychology/emotions/buffering-effect/
(Accessed 11 March 2019).
258 https://en.wikipedia.org/wiki/Propinquity (Accessed 11 March 2019).
259 https://en.wikipedia.org/wiki/Availability_heuristic (Accessed 11 March 2019)
260 https://www.investopedia.com/terms/s/scarcity-principle.asp (Accessed 11 March
2019).
261 Savitz, Eric. Forbes. Listening To Social Media Cues Doesn't Mean Ceding Control.
August 4, 2002. https://www.forbes.com/sites/ciocentral/2012/08/04/listening-to-
social-media-cues-doesnt-mean-ceding-control/#702ffbc2cd52 (Accessed 6 November
2019).
262 https://triplemmarketin.com/index.php/2019/01/21/your-brand-is-a-story-
unfolding-across-all-customer-touch-points/ (Accessed 6 November 2019).
263 Twain, Mark (Samuel L. Clemens).—George Bainton, The Art of Authorship, pp. 87–
88 (1890).
264 https://thedecisionlab.com/bias/framing-effect/ (Accessed 11 March 2019).
265 http://changingminds.org/explanations/theories/acquiescence_effect.htm
(Accessed 11 March 2019).
266 Thu-Huong Ha. (2013). TED blog. The psychology of saving energy. February 27, 2013
https://blog.ted.com/the-psychology-of-saving-energy-alex-laskey-at-ted2013/
(Accessed 11 March 2019).
267 Sunstein, Cass and Thaler, Richard. (2016). The New Yorker. The Two Friends Who
Changed How We Think About How We Think. December 7, 2016.
https://www.newyorker.com/books/page-turner/the-two-friends-who-changed-how-
we-think-about-how-we-think (Accessed 2 May 2019).
268 The Tonic Communications. (2018). Frye festival: how millennial FOMO enabled high
end fraud. https://thetoniccomms.co.uk/fyre-festival-how-millennial-fomo-enabled-
high-end-fraud/ (Accessed 12 March 2019).
269 Lee, Aileen. (2011). Social Proof Is The New Marketing. Techcrunch.
https://techcrunch.com/2011/11/27/social-proof-why-people-like-to-follow-the-crowd/
(Accessed 16 August 2018).
270 https://en.wikipedia.org/wiki/Social_proof (Accessed 16 August 2018).
271 Eric Hoffer Quotes. (n.d.). BrainyQuote.com. Retrieved December 10, 2019, from
BrainyQuote.com Web site: https://www.brainyquote.com/quotes/eric_hoffer_122526
(Accessed 11 December 2019).
272 Wilde, Oscar. De Profundis. February 1905.
273 Cialdini, Robert B. (2006). Influence: The Psychology of Persuasion. Harper Business;
Revised edition (December 26, 2006).
274 Lua, Alfred. The Psychology of Marketing: 18 Ways Social Proof Can Boost Your
Results. https://www.upwork.com/hiring/for-clients/the-psychology-of-marketing-18-
ways-to-use-social-proof-to-boost-your-results/ (Accessed 6 March 2019).
275 Cialdini, Robert B. Harnessing the Science of Persuasion. Harvard Business Review.
October 2001. http://www.coachfinder.club/downloads/Influence%20by%20Cialdini.pdf
(Accessed 15 August 2018).
276 Hallin, Ed. (2014). Fast Company. How to use the psychology of social proof to your
advantage. 05 May 2014. https://www.fastcompany.com/3030044/how-to-use-the-
psychology-of-social-proof-to-your-advantage262 (Accessed 14 August 2018).
 THE A.I. CRUISE LINE

277 Matsakis, Louise. (2018). The Logan Paul Video Should Be a Reckoning For YouTube.
January 3, 2018 https://www.wired.com/story/logan-paul-video-youtube-reckoning/
(Accessed 11 December 2019).
278 Chowdhury, Amit Paul. Analytics India Magazine. Is Machine Learning Key to
Psychometrics? May 26, 2017. https://www.analyticsindiamag.com/machine-learning-
key-psychometrics/ (2017). Accessed 7 March 2019).
279 Grassegger, H., Krogerus, M. December 3, 2016. I Just Showed That the Bomb Was
There. Das Magazin https://www.dasmagazin.ch/2016/12/03/ich-habe-nur-gezeigt-dass-
es-die-bombe-gibt/ (22 November 2017).
280 UPS Customer Solutions & Innovations. (2018). Artificial Intelligence in Logistics, a
collaborative report by DHL and IBM on implications and use cases for the logistics
industry. https://www.logistics.dhl/content/dam/dhl/global/core/documents/pdf/glo-
ai-in-logistics-white-paper.pdf (Accessed 29 April 2019).
281 Facial recognition: Who’s Tracking You In Public. (December 30, 2015) Consumer
Reports. Online: http://www.consumerreports.org/privacy/facial-recognition-who-is-
tracking-you-in-public1/
282 Gates, Kelly A. January 23, 2011. Our Biometric Future: Facial Recognition Technology
and the Culture of Surveillance. NYU Press.
283 Khatchadourian, Raffi. (2015). The New Yorker. We Know How You Feel. January 19,
2015. https://www.newyorker.com/magazine/2015/01/19/know-feel. (Accessed 24
February 2019).
284 Ford, Martin. (2018). Architects of Intelligence. The truth about AI from the people
building it. Packt Publishing; 1 edition. November 23, 2018.
285 https://en.wikipedia.org/wiki/Web_search_engine (Accessed 13 November 2017).
286 https://en.wikipedia.org/wiki/Web_search_engine
287 Lardinois, F. (2013, September 26). Google improves knowledge graph with
comparisons and filters, brings cards & cross-platform notifications to mobile. Retrieved
from TechCrunch: http://techcrunch.com/2013/09/26/google-improves-knowledge-
graph-with-comparisons-and-filters-brings-cards-to-mobile-search-adds-cross-platform-
notifications/
288 Lardinois, F. (2013, September 26). Google improves knowledge graph with
comparisons and filters, brings cards & cross-platform notificiations to mobile. Retrieved
from Techcrunch: http://techcrunch.com/2013/09/26/google-improves-knowledge-
graph-with-comparisons-and-filters-brings-cards-to-mobile-search-adds-cross-platform-
notifications/
289 Moz (2015) ‘Search Engine Ranking Factors 2015 Expert Survey and Correlation Data’,
available at: https://moz.com/search-ranking-factors (Accessed 15th August 2017). Moz’s
rankings survey consists of the compiled results of 150 SEP expert opinions as to what
influences Google’s secret algorithm. It is largely a technical document and relies heavily
on (defined) technical terms, such as ‘Domain-Level-Keyword-Agnostic-Features’, but it is
worth a look nevertheless.
290 https://en.wikipedia.org/wiki/Google_Search (Accessed 13 November 2017).
291 Alpert, Brian. (Spring 2017). Search engine optimization in 2017: A new world where
old rules still matter. Journal of Digital & Social Media Marketing, Volume 5, Number 1,
Spring 2017, pp. 39 – 60 (22). Henry Stuart Publication.
292 Vindicia. (2014). Digital Age/Digital Goods. Retrieved from vindicia:
http://info.vindicia.com/White-Paper---Digital-Age-Digital-Goods-
9essentials_for_acquiring_subscription_and_recurring_revenue_customers.html
(Accessed 15 August 2018). 263
 ANDREW W. PEARSON

293 Pakkala, H., Presser, K., & Christensen, T. (2012). Using Google Analytics to measure
visitor statistics: The case of food composition websites. International Journal of
Information Management, 32(6), 504-512.
294 Plaza, B. (2009). Monitoring web traffic source effectiveness with Google Analytics:
An experiment with time series. In Aslib Proceedings (Vol. 61, No. 5, pp. 474-482).
Emerald Group Publishing Limited.
295 Baye, M. R., De los Santos, B., & Wildenbeest, M. R. (2016). Search engine
optimization: what drives organic traffic to retail sites? Journal of Economics &
Management Strategy, 25(1), 6-31.
296 Kaushik, A. (2010). LDS (Mormon) Church knows SEO with Avinash Kaushik, Google,
at SES Chicago 2010. Available from http://www.youtube.com/watch?v=DjWInquqhMA.
297 Berman, R., & Katona, Z. (2013). The role of search engine optimization in search
marketing. Marketing Science, 32(4), 644-651.
298 Jerath, K., Ma, L., & Park, Y. H. (2014). Consumer click behaviour at a search engine:
The role of keyword popularity. Journal of Marketing Research, 51(4), 480-486.
299 Yao, S., & Mela, C. F. (2011). A dynamic model of sponsored search advertising.
Marketing Science, 30(3), 447-468.
300 Agarwal, V., Bell, G. W., Nam, J. W. & Bartel, D. P. (2015). Predicting effective
microRNA target sites in mammalian mRNAs. elife, 4, e05005.
301 Jeziorski, P., & Segal, I. (2015). What makes them click: Empirical analysis of consumer
demand for search advertising? American Economic Journal: Microeconomics, 7(3), 24-
53.
302 Liu, J., & Toubia, O. (2018). A semantic approach for estimating consumer content
preferences from online search queries. Marketing Science.
303 Strzelecki, A., Furmankiewicz, M., & Ziuziański, P. (2016). The use of management
dashboard in monitoring the efficiency of the internet advertising campaigns illustrated
on the example of google analytics. Studia Ekonomiczne, 296, 136-150.
304 Patil, K. P., & Puri, P. (2015). Small and Medium Enterprises (SMEs) Key Challenges
and innovative approach for using social media marketing (SMM). Journal for Studies in
Management and Planning, 1(5), 450-460.
305 Kumar, A., & Salo, J. (2018). Effects of link placements in email newsletters on their
click-through rate. Journal of Marketing Communications, 24(5), 535-548.
306 Chittenden, L., & Rettie, R. (2003). An evaluation of e-mail marketing and factors
affecting response. Journal of Targeting, Measurement and Analysis for Marketing, 11(3),
203-217.
307 Bickart, B., & Schindler, R. M. (2001). Internet forums as influential sources of
consumer information. Journal of interactive marketing, 15(3), 31-40.
308 Kumar, N., & Benbasat, I. (2006). Research note: the influence of recommendations
and consumer reviews on evaluations of websites. Information Systems Research, 17(4),
425-439.
309 Zhang, J. Q., Craciun, G., & Shin, D. (2010). When does electronic word-of-mouth
matter? A study of consumer product reviews. Journal of Business Research, 63(12),
1336-1341.
310 Chan, Y. Y., & Ngai, E. W. (2011). Conceptualising electronic word of mouth activity:
An input-process-output perspective. Marketing Intelligence & Planning, 29(5), 488-516.
311 Copley, P. (2015, November). For the love of AIDA–developing the Hierarchy of
Effects model in SME social media marketing strategy. In Institute Small Business and
Entrepreneurship (ISBE) Conference.
264
 THE A.I. CRUISE LINE

312 See-To, E. W., & Ho, K. K. (2014). Value co-creation and purchase intention in social
network sites: The role of electronic Word-of-Mouth and trust–A theoretical
analysis. Computers in Human Behaviour, 31, 182-189.
313 Yu, Jim. Searchengingland.com. Supercharging Your SEO with AI Insights, Automation,
and Personalization. June 12, 2018. https://searchengineland.com/supercharging-your-
seo-with-ai-insights-automation-and-personalization-299900 (Accessed February 19,
2019).
314 Simonite, Tom. (2019). Wired. Google Search Now Reads at a Higher Level. 25
October, 2019. https://www.wired.com/story/google-search-advancing-grade-reading/
(26 October 2019).
315 Enterprise Content Team. The Magic of AI in a content-driven world. Using AI to
create content faster. https://www.adobe.com/insights/the-magic-of-AI-in-a-content-
driven-world.html (Accessed 15 February 2019).
316 32 Inspiring Quotes From Marketing & Advertising Experts.
https://www.ventureharbour.com/great-quotes-advertising-marketing-experts/
(Accessed 24 October 2019).
317 Scalco, Dan. Inc.com. 23 Powerful Quotes to Get Your Creativity Flowing. 23 January,
2018. https://www.inc.com/dan-scalco/23-powerful-quotes-to-get-your-creativity-
flowing.html (Accessed 24 October 2019).
318 Abramovich, Giselle. (2017). CMO. Future of Advertising: Automated, Personalized,
and Measureable. 26 September 2017.
https://www.cmo.com/features/articles/2017/9/25/the-future-of-advertising-
automated-personalized-and-measurable-.html
319 TechTarget. (2011). Geofencing definition. Retrieved from TechTarget:
http://whatis.techtarget.com/definition/geofencing
320 Berman, S. J., Battino Bill, Feldman, Karen. 2007. Executive Brief: Navigating the
media divide: Innovating and enabling new business models. IBM Institute for Business
Value.
321 Giaglis, G. M. (2002). On the Potential Use of Mobile Positioning Technologies in
Indoor Environments. 15th Bled Electronic Commerce Conference eReality: Constructing
the eEconomy. Bled, Solvenia.
322 Weiners, C. (2012, March 30). LBS Opportunities for Cruise line Marketers in Macau.
Retrieved from clickz.com: http://www.clickz.com/clickz/column/2281870/lbs-
opportunities-for-cruise line-marketers-in-macau
323 Takahashi, D. (2013, May 22). Lighthouse’s new Android location service could give
you indoor navigation for Las Vegas’ cruise lines. Retrieved from Venturebeat.com:
http://venturebeat.com/2013/05/22/lighthouse-signal-systemss-android-app-will-let-
you-find-your-way-inside-the-biggest-las-vegas-cruise lines/
324 Thompson, Cadie. May 28, 2015. Social media apps are tracking your location in
shocking detail. Business Insider. http://www.businessinsider.com/three-ways-social-
media-is-tracking-you-2015-5
325 Gell, Aaron. (2017) The Not-so-Surprising Survival of Foursquare. The New Yorker.
March 1, 2017. https://www.newyorker.com/business/currency/the-not-so-surprising-
survival-of-foursquare (Accessed 23 November 2017).
326 Tepper, Fitz. Foursquare is launching an analytics platform to help cruise lines
understand foot traffic. Techcrunch. March 20, 2017.
https://techcrunch.com/2017/03/20/foursquare-is-launching-an-analytics-platform-to-
help-cruise lines-understand-foot-traffic/ (Accessed 23 November 2017).
265
 ANDREW W. PEARSON

327 Crowley, Dennis. Foursquare.com. Introducing Hypertrending. March 8, 2019.

https://enterprise.foursquare.com/intersections/article/introducing-hypertrending/
(Accessed 21 August 2019).
328 https://en.wikipedia.org/wiki/Word-of-mouth_marketing#cite_note-
What_Makes_Online_Content_Viral-11 (accessed 23 September 2019).
329 Lang, Bodo; Hyde, Ken (2013). Word of mouth: what we know and what we have yet
to learn. Journal of consumer satisfaction, dissatisfaction and complaining behavior. 26:
1–18.
330 Katz, E., & Lazarsfeld, P. F. (1955). Personal influence: The part played by people in
the flow of mass communications. Glencoe, IL: The Free Press.
331 Phelps, J. E., Lewis, R., Mobilio, L., Perry, D., & Raman, N. (2004). Viral marketing or
electronic word-of-mouth advertising: Examining consumer responses and motivations
to pass along e-mail. Journal of Advertising Research, 44(4), 333—348.
332 Smith, T., Coyle, J. R., Lightfoot, E., & Scott, A. (2007). Reconsidering models of
influence: The relationship between consumer social networks and word-of-mouth
effectiveness. Journal of Advertising Research, 47(4), 387—397.
333 Chui, H-C., Hsieh, Y-C., Kao, Y-H., & Lee, M. (2007). The determinants of email
receivers’ dissemiating behaviors on the Internet. Journal of Advertising Research, 47(4),
524—534.
334 Berger, J. (2012). Contagious: Why things catch on. New York: Simon & Schuster.
335 Berger, Jonah; Milkman, Katherine L (2012). "What Makes Online Content
Viral?". Journal of Marketing Research. 49 (2): 192. doi:10.1509/jmr.10.0353.
336 Berger, Jonah; Schwartz, Eric M (2011). "What Drives Immediate and Ongoing Word
of Mouth?". Journal of Marketing Research. 48 (5): 869. doi:10.1509/jmkr.48.5.869.
337 Chevalier, J. A., & Mayzlin, D. (2006). The effect of word of mouth on sales: Online
book reviews. Journal of Marketing Research, 43(3), 345—354.
338 Trusov, M., Bucklin, R. E., & Pauwels, K. (2009). Effects of word-ofmouth versus
traditional marketing: Findings from an Internet social networking site. Journal of
Marketing, 73(5), 90—102.
339 Dellarocas, C. (2003). The digitization of word of mouth: Promise and challenges of
online feedback mechanisms. Management Science, 49(10), 1407—1424.
340 Dellarocas, C. (2006). Strategic manipulation of Internet opinion forums: Implications
for consumers and firms. Management Science, 52(10), 1577—1593.
341 Godes, D., & Mayzlin, D. (2004). Using online conversations to study word-of-mouth
communication. Marketing Science, 23(4), 545—560.
342 Kozinets, R. V. (2002). The field behind the screen: Using netnography for marketing
research in online communities. Journal of Marketing Research, 39(1), 61—72.
343 Marsden, P. V. (1987). Core discussion networks of Americans. American Sociological
Review, 52(1), 122—131.
344 Hill, R. A., & Dunbar, R. I. M. (2003). Social network size in humans. Human Nature,
14(1), 53—72.
345 Kozinets, R. V., de Valck, K., Wojnicki, A. C., & Wilner, S. J. S. (2010). Networked
narratives: Understanding word-of-mouth marketing in online communities. Journal of
Marketing, 74(2), 71—89.
346 Goldenberg, J., Han, S., Lehmann, D. R., & Hong, J. W. (2009). The role of hubs in the
adoption process. Journal of Marketing, 73(2), 1—13.
347 Thomas, G. M. (2004). Building the buzz in the hive mind. Journal of Consumer
Behaviour, 4(1), 64—72.
266
 THE A.I. CRUISE LINE

348 Kaikati, A. M., & Kaikati, J. G. (2004). Stealth marketing: How to reach consumers
surreptitiously. California Management Review, 46(4), 6—22.
349 Rayport, J. (1996, December 31). The virus of marketing. Fast Company. Retrieved
from http://www.fastcompany.com/magazine/06/virus.html (Accessed 6 November
2019).
350 Feick, L. F., & Price, L. L. (1987). The market maven: A diffuser of marketplace
information. Journal of Marketing, 51(1), 83—97.
351 Goldenberg, J., Han, S., Lehmann, D. R., & Hong, J. W. (2009). The role of hubs in the
adoption process. Journal of Marketing, 73(2), 1—13.
352 Kamins, M. A., Folkes, V. S., & Perner, L. (1997). Consumer responses to rumors: Good
news, bad news. Journal of Consumer Psychology, 6(2), 165—187.
353 Dobele, A., Lindgreen, A., Beverland, M., Vanhamme, J., & van Wijk, R. (2007). Why
pass on viral messages? Because they connect emotionally. Business Horizons, 50(4),
291—304.
354 McGregor, J. (2007). Customer service champs. Business Week, 4024, 52—64.
355 Peter's Quotations: Ideas for Our Time. Laurence J. Peter. Bantam Books, New York
NY, USA. 1977/1979.
356 Bentahar, Amine. (2018). Forbes. How AI Will Change Marketing as We Know It. May
23, 2018. https://www.forbes.com/sites/forbesagencycouncil/2018/05/23/how-ai-will-
change-marketing-as-we-know-it/#52bb93278b78 (Accessed 6 April 2019).
357 Britt, Steuart Henderson. (1978) Marketing Management and Administrative Action.
McGraw-Hill; 4th edition.

267
268
 THE A.I. CRUISE LINE

CHAPTER FIVE: SOCIAL A.I.

Overview
The most important thing to recognize about social media is the fact that most
social media content is user generated. Social networks provide all of the tools
their members require to become content producers and social network
members submit photos, videos, and other forms of multimedia as well as
provide customer reviews, content for blogs and vlogs and links to other social
networking websites that they find noteworthy.358 The content comes from the
users themselves, not from the publishers, and this is an important distinction.358
The publisher supplies all of the necessary tools for the content’s distribution,
but it must remain at arm’s length from the actual content to ensure the
integrity’s content.
Business.com's Top Tools to Measure Your Social Media Success 359 states that
there are five Ws that must be kept in mind when devising a social media
strategy. These are:
1. Who within the company will be using this tool? Will one person or
several people be using the tools, and will they be inside or outside the
organization? Will the primary user be tech savvy or will he or she
require an intuitive interface?
2. What key performance indicators (KPI) are to be measured with this
tool? It is imperative to know how you are going to measure and
benchmark your social media efforts as this will dictate what social
media monitoring tools are the best to use. If sales revenue is a key KPI,
businesses should invest in a tool that integrates with a CRM system to
track impact.
3. Where on the web will the business be engaging customers, and where
does it plan to monitor its social media conversations? If a business is
only interested in tracking specific channels such as Facebook or
Twitter, tools such as Facebook (obviously) and socialmention.com can
help with the former, while Twazzup and TweetEffect can track the
latter. All-encompassing tools that monitor new sites and forums are
useful to monitor mentions from across the entire web.
4. When should the company be alerted of conversations and mentions
within the social media sphere? Options here include general reporting
dashboards or instant notifications via e-mail alerts or RSS feeds.
5. Why is the company engaging in social media? This is, perhaps, the most
important question of all, and a cruise line operator must decide
whether it is turning to social media to manage its online brand

269
 ANDREW W. PEARSON

reputation, to engage its customers and/or potential customers, to

provide real-time customer service, or simply to drive traffic to its
website to influence SEO.
A company is only as strong as its weakest customer relationship and I believe
that social media can help cruise lines reach their customers in highly efficient
and, what can be extraordinarily affordable, ways. Cruise lines should look to
social media to help them in the following ways:
• Adding interactivity to a Website
• Brand and Anti-Brand management
• Brand loyalty enhancement
• Building fanbases
• Crisis management
• Developing a virtual social world presence
• Discovering a customer’s psychological profile
• Discovering important brand trends
• Engaging customers and potential customers
• Harvesting customer feedback
• Influencer marketing
• Marketing to consumers
• Reputation management
• Social Shopping
Studies have shown that 80% of social media users prefer to connect with brands
through Facebook and 43% of people prefer Pinterest over associating directly
with retailers and/or brands.360 This fact alone should underscore the
importance of social media in a business context.
One of the key demographic findings of the 2018 Adobe Digital Insights (ADI)
State of Digital Advertising report was that Millennials and Gen Zers differ from
Generation X, Baby Boomers, and older generations in that, “social channels are
where these generations see the most relevant content in their lives.” 62
According to Taylor Schreiner, “social advertising is clearly a key part of a
paid/owned/earned media strategy, especially if your audience is under 40.” 62
This is a fact that businesses should keep in mind going forward as Millennials
and Gen Zers are now reaching an age when they will have disposable income,
as well as the desire to spend it.
Caesars is one gaming company that has been able to use social media to
measure marketing data quite successfully. In his article At Caesars, Digital
Marketing Is No Crap Shoot361, Al Urbanski explains that:
“While social media networks like Facebook provide metrics
that measure activity within its platform, integrating that data
to enable visibility across a brand's entire marketing

270
 THE A.I. CRUISE LINE

organization is difficult. Caesars, however, unites information

from customers coming through social channels across
business units, program teams, time zones, and languages. A
content-building component allows Caesars' marketers to
listen in and respond in real time.”
No matter where the customer interaction originates, engagement is a key factor
in moving those interactions from the top of the sales funnel to an eventual
purchase.361 “It doesn't matter where customers come in or leave or reenter,”
says Chris Kahle, Caesars Web Analytics Manager, “if they come to your social
page and click your button, or if they go into your content or email and click on
that, it's all the same app and you've got them.”361 Caesars IDs a cookie and if
the prospects come back around on paid search three days later, Caesars tracks
them.361 “We can track them on every website, even if they came in on a Las
Vegas site and then jump markets to Atlantic City,” adds Kahle. 361
Caesars also tracks activity in real time, while responding to customer cues. 361
Unsurprisingly, different types of customers are more responsive to different
interactions from Caesars. Aside from dividing customers into categories such as
“Frequent Independent Traveler” — or FITs and Total Rewards members, the
Caesars team uses tracking data to further segment customers by property or
market as well as to determine how each of their various segments respond to
content.361
Using this data, Caesars evaluates campaigns in regard to KPIs, such as number
of nights booked, and adjusts them on the fly to ramp up conversion rates. 361
“What's really dramatic about this is that you can determine what is engaging
individuals and target them with it,” says Adobe's Langie.361 “The high-roller
segment, for example. They might respond to a very different Web design than
the casual visitor and Caesars tailors the page view to who is visiting. Think of
the website as a canvas. You can paint a still life of a fruit for one person and
something different for another. The canvas is dynamic.” 361
“The speed and the manner with which the chosen website designs and digital
marketing tactics are implemented across the Caesars network may well be the
most transforming development of the company's new data culture,” adds
Kahle.361 And this was no easy task as the Caesars landscape extends over 60
websites for its various properties and services as well as 40 Facebook pages.
“Prior to implementing a data-centric approach to the decision-making process,
it could take as long as two weeks to furnish the field with actionable data. They
now get it done in a matter of hours,” adds Kahle. 361 In 2013, Caesars’
implemented Adobe’s Digital Marketing Suite, which “includes real-time tracking
and segmentation of digital site visitors, analysis of social media’s role in
purchasing, and content testing by segment or individual visitor.” 361
“The people at the individual properties who are managing the content of the

271
 ANDREW W. PEARSON

websites are not all technically sophisticated, but Adobe’s system provides them
with built-in capabilities,” says Kahle.361 “Say one of our properties wants to track
social. Before, they'd have to spend a lot of time manually adding tracking codes.
With Adobe, tracking codes are integrated,” Kahle adds.361
“Right now we can assign a percentage value to social media if a booking doesn't
result right away,” Kahle says.361 “But with social we're going to be
experimenting with a longer funnel, maybe a two-week time frame.”361 “Values
are ascribed to social media for being the site of initial contact with a new
customer, for instance, or for numbers of positive reviews by current
customers.”361
Currently, Caesars can’t measure the total value of a reservation booked online
and also can't determine how much an online booker spends at the tables during
his or her stay.361 This is important information when it comes to truly
understanding a patron. Caesars would also like to know if, for example,
“customers left the Caesars' cruise line in Las Vegas and went to dinner at
Gordon Ramsay's restaurant at the Paris Las Vegas, so they could offer them a
free dinner at the restaurant to close the deal on a future booking.” 361
“Eventually we're going to set a time frame that will never expire [on the sales
funnel],” Kahle says.361 “But for now we've built a sales allocation model that
goes beyond the last click, and that's OK. Most organizations using multiple
marketing channels are still stuck on that last click.” 361

Social Media Analytics

As Melville and Lawrence explain in their article Social Media Analytics:
Channeling the Power of the Blogosphere for Marketing Insight 362, social media
analytics is “the practice of gathering data from blogs and social media websites
and analyzing that data to make business decisions. The most common use of
social media is to mine customer sentiment.” Social media analytics evolved out
of the disciplines of social network analysis, machine learning, data mining,
information retrieval (IR), and Natural Language Processing (NLP).
According to Melville and Lawrence, the automotive analysis of blogs and other
social media sites raise the following intriguing marketing questions362:
1. Given the enormous size of the blogosphere, how can we identify the
subset of blogs and forums that are discussing not only a specific
product, but higher-level concepts that are in some way relevant to this
product?
2. Having identified this subset of relevant blogs, how do we identify the
most authoritative or influential bloggers in this space?
3. How do we detect and characterize specific sentiment expressed about
an entity (e.g., product) mentioned in a blog or a forum?

272
 THE A.I. CRUISE LINE

4. How do we tease apart novel emerging topics of discussion from the

constant chatter in the blogosphere?
As Margaret Rouse explains in her article Social Media Analytics363, step one of
a social media analytics initiative is “to determine which business goals the data
that is gathered and analyzed will benefit. Typical objectives include increasing
revenues, reducing customer service costs, getting feedback on products and
services and improving public opinion of a particular product or business
division.” Once these business goals have been identified, “key performance
indicators (KPIs) for objectively evaluating the data should be defined. For
example, customer engagement might be measured by the numbers of followers
for a Twitter account and number of retweets and mentions of a company’s
name,” states Rouse.363
Through social networks like Twitter and Weibo, organizations can pick up
customer satisfaction in real time.108 “Social media is enabling companies such
as Coca-Cola, Starbucks, and Ford to go beyond standard customer satisfaction
data gathering to innovate by setting up and participating in communities to gain
feedback from customers.”108 A good example is MyStarbucksIdea.com, it is a
website where “Starbucks customers can relate their experiences and offer ideas
about how to improve the Starbucks experience, from drinks to foods to
ambiance.”108
When looking at what objectives companies were seeking when implementing
customer analytics technologies with social media data (see Figure 21), TDWI
Research found that gaining a “deeper customer understanding” topped the list
at 56%108 “Social media listening can provide an unprecedented window into
customer sentiment and the reception of an organization’s marketing, brands,
and services.”108
Besides the broad objective of gaining deeper customer understanding, nearly
one-third (31%) of companies seek to identify attribution, or paths to buying
decisions, which can be done on a limited scale with services like Google
Analytics as well as other Web site analysis applications.108 Google webmaster
tools also allow cruise line to understand the organic search traffic that is linking
customers to them.
30% or respondents sought to discover customer sentiment, which is important
because it helps companies discover positive and negative comments in social
media channels, on customer comment and review sites.108 “Sentiment analysis
often focuses on monitoring and measuring the ‘buzz’ value, usually through
volume and frequency of comments around a topic.” 108
Simply deciding which social media sites’ data to analyze can be one of the
biggest challenges facing businesses going down the analytics path.
“Organizations have to research where their customers are most likely to express
themselves about brands and products. They need to spot influencers who have

273
 ANDREW W. PEARSON

networks of contacts and take it upon themselves to play an advocacy role.” 108
“About 20% of respondents are interested in differentiating influencers from
followers in social media.108 “Link analytic tools and methods specialize in
identifying relationships between users in social communities and enabling
organizations to measure users’ influence.” 108 “With some tools, data scientists
and analysts can test variables to help identify social communities as ‘segments’.
Then, as they implement segmentation models for other data sources, they can
integrate these insights with social media network analysis to sharpen models
and test new variables,” explains Stodder.108

Which of the following objectives does your organization seek to

achieve by implementing customer analytics technologies and
methods with social media data? (Please select all that apply.)

Gain deeper customer understanding 56%

Identify customer paths to buying decision 31%

Monitor and measure sentiment drivers 30%

Determine value of social media engagement
29%
to marketing campaigns
Discover new audience segments 27%

Gain insights for new product development 24%

Analyze social networks, links, and graphs 22%

Differentiate influencers from followers in
20%
social media
Increase engagement beyond passive social
19%
media monitoring
Analyze competition's "share of voice" 18%
Monitor and analyze social activity in real
14%
time
Improve "long-tail" analysis of buying by
11%
small groups of customers
We do not analyze social media data 32%

0% 10% 20% 30% 40% 50% 60%

Figure 21: Customer Analytics and Social Media Objectives

Based on 1,546 respondents from 418 respondents; a bit more than three
responses per respondent, on average.
Source: TDWI Research

Analytics are critical in helping organizations “make the right decisions about
when, where, and how to participate in social media. It isn’t enough to just listen;
organizations must insert themselves and become part of the conversation.”108

274
 THE A.I. CRUISE LINE

When doing so, however, companies should keep in mind advice from The
Cluetrain Manifesto364 — “Conversations among human beings sound human.
They are conducted in a human voice,” as well as this: “When delivering
information, opinion, perspectives, dissenting arguments or humorous asides,
the human voice is typically open, natural, uncontrived.”364
One interesting strategy is for a cruise line operator to start viral campaigns via
Twitter, using hashtags for a topic; the campaign could be a part of a larger
marketing strategy. The cruise line can then “monitor social media to see what
people say and analyze how the campaign is playing among influencers and
across networks.”228
Klear, a social media and social data platform that focuses on influencer
marketing, offers a product that can help cruise lines understand the effects of
their influencer marketing. Klear’s campaign reports contain the following
summaries365:
• How many influencers participated in the campaign?
• Number of updates the influencers posted during the campaign.
• Engagements metrics.
• Number of people who saw the content.
The report also includes a drill-down analysis for each and every influencer. For
each influencer the report will show365:
• Who is the influencer?
• The influencer’s expertise.
• Fanbase across different social networks.
• Top posts during the campaign.
• Engagements for these updates.
• A direct link to the influencer’s profile on Klear.
This is a paid service, but most of the information is publicly available and this is
something an IR could build up in-house, should they want a customized
solution.
Influencer marketing taps directly into what Deighton and Kornfeld call the five
paradigms of digital interactive marketing, i.e., social exchanges — building
identities within virtual communities — and cultural exchanges — firms offering
culture products that will compete in buzz markets.228 This peer-to-peer
interactivity should motivate the desire to exchange and share information,
which should help market any IR event.228
In their article Social Media Monitoring: An Innovative Intelligent Approach 366,
Emmanouil Perakakis, George Mastorakis and Ioannis Kopanakis argue that
“Social media marketing today has become too complex. There are too many
social media channels, devices and technologies in the marketing stacks, creating

275
 ANDREW W. PEARSON

a huge volume of data that is no longer humanly possible for marketers to

manage. This makes the art of campaigning extremely time-consuming and
inefficient.” The writers add that, “Artificial intelligence can play a crucial role in
social media marketing by introducing new features for tasks automation.”366
“AI is poised to elevate the power of marketers’ data capabilities and enable
powerful contextual marketing programs,” say Perakakis et al.366 “By
understanding the customer, brands can determine relevant marketing
messages, find the right social media influencers, refine their content marketing
strategy and gain insightful information about their users,” add Perakakis et al.366
“In addition, AI will play a crucial role in social media monitoring tools, by
enabling new features for smart suggestions and intelligent decisions based on
the analyzed data (e.g., mentions),” notes Perakakis et al.366
“As intelligent social media monitoring improves, it will become easier for brands
to find the right influencers,” argue Perakakis et al.366 Finding and hiring these
influencers will become less resource-intensive, less time-consuming and more
accurate.366 There are already social media monitoring companies that make use
of influencers’ followers, posts and interactions to determine if they are suitable
for a brand.366
Timing, as they say, is everything, especially in social media marketing. The
channel can also be important. “Sometimes social media marketers have great
content available but the time or the medium, at which they post could be
wrong. This means that the content does not receive the engagement that it
deserves,” explains Perakakis et al.366
If a social media post is uploaded at the wrong time, it could receive customer
engagement for just a few seconds.366 “Choosing the right time and platform for
posting content is a process that can be simplified by using an intelligent social
media monitoring tool,” recommend Perakakis et al.366 Tools embedded with
intelligent features can “analyze data about the reach of posts with their time
and platform to come up with an effective content posting strategy.”366
Social media is a great channel for cruise lines to connect with their customers
and potential customers.366 “Whether digital marketers look to grab the
attention of prospects or to answer queries of their customers, social media is
the way to go. However, social media, unlike other forms of media, thrives on
real-time interactions,” note Perakakis et al.366 This means marketers can’t
afford a long delay in responding to their customers.366 Immediacy is
imperative.366
“Social media monitoring tools can help marketers identify opportunities for
interacting with customers by providing vital insights. Intelligent features in a
social monitoring tool allow for determining when and how to respond back to
customers,” claim Perakakis et al.366 In futures years, AI will undoubtedly be even
more involved in the customer service process than it currently is.366 Cruise lines

276
 THE A.I. CRUISE LINE

need to jump on this trend ASAP.

Today, it’s easy for social media marketers to get overwhelmed by the amount
of social media posts referencing their brand.366 Moreover, they can also get
confused between which posts are referencing their brand and which aren’t; not
all industry keywords will be relevant to a brand’s marketing message but it’s
almost impossible to keep up with it all, especially for international brands.366
AI automates the process of scanning through social media posts, ensuring key
opportunities for customer engagement aren’t missed.366 “Social media
marketers can leave the sorting of the posts to an intelligent social media
monitoring tool. This enables them to focus on responding back to the
customers,” say Perakakis et al.366 The reality is, “no matter how many social
media marketers are assigned to the task of sorting social media posts, it is nearly
impossible to carry out this task without the assistance of automation.”366 “AI
can play a key role in helping sort these posts and it is expected that intelligent
features will improve such processes even further. Learning algorithms enables
for making informed calculations, perform analytics and even make automated
recommendations,” contend Perakakis et al.366
Social media marketers can also mine an intelligent monitoring tool to generate
customer leads.366 “AI-powered insights allow social media marketers to
improve conversion rates and generate more sales,” argue Perakakis et al.366
“For efficient social media monitoring, brands need to go through and learn from
all user-generated content, including images,” contend Perakakis et al.366
“Intelligent tools have to be able to recognize brand-related images and draw
meaning from them,” say Perakakis et al.366 “Through this, brands can learn in-
depth insights about the customers’ feedback on their products. AI enables
brands to recognize possible sales opportunities and cross-promotional
opportunities through visual content analytics,” say Perakakis et al.366 With
information gleaned from images, social media marketers can capture the
available opportunities and release targeted marketing content for customers
and even potential customers.366
Customers who regularly post images or blogs about a product, can be sent
targeted promotions or even just messages of appreciations showing them that
the company cares.366 This should increase customer loyalty as well as encourage
them to keep posting content about the brand.366 It becomes a positive
reinforcement loop that will help with word of mouth marketing for the
company. “Moreover, analyzing the visual content that customers post can also
help social media marketers developing detailed buyer personas,” say Perakakis
et al.366 These buyer personas provide priceless information and data about the
groups of people that buy from the brand, information that the marketing
department should covet.366
In a general context, AI plays a key role in monitoring social media platforms and

277
 ANDREW W. PEARSON

other customer review forums.367 “Since social sharing continues to improve

across different platforms, as a result, brands face a serious challenge when it
comes to monitoring what customers say on these platforms,” claim Perakakis
et al.366 With the overwhelming amount of feedback and reviews that businesses
can get on these platforms, it is important to have an advanced monitoring tool
that captures both quantity and nuance.366 This ensures that no harm will come
to the brand’s reputation, which is extremely important in this highly sensitive
culture we’re currently living in.366
With AI, businesses can effectively monitor mentions on social media coming
from a multitude of channels and networks.366 AI can help identify a customer
who is complaining about a particular company product or service.368 AI can be
used to identify positive comments from select customers.366 This social proof
can then be shared on the company’s social channels, which will help it build its
online reputation.366 A Python NLP script can easily analyze and classify positive
and negative comments. Alerts can be set up for negative comments, which can
quickly be acted upon to ensure the company’s reputation isn’t damaged by
incensed or annoyed customers.366
Personalized customer experiences also help with online reputation
management, argue Perakakis et al.366 To deliver one, a business needs to clearly
understand it customer demands.366 By collecting customer data on social media
and correlating that with a customer’s purchase history, companies can
understand the exact customer demands.366
AI can be used to recognize customer preferences, which will help deliver
personalized experiences. For marketers, AI helps meet the demand for highly
individualized experiences.369 Thus, businesses can better reach and target their
consumers, which should results in a better buying experience.366
AI can also help businesses track customer posts to identify any possible pain
points.366 AI can also ensure customer service is effectively delivered.366 Today,
customers almost demand instant replies from a business they frequent and any
slight delays will lead to customers complaining about inefficiency, often to any
and every one on their social media feed.366 This can risk tarnishing or flat out
ruining a company’s reputation.366 Through the use of AI and chatbots, customer
concerns can be addressed effectively.
Other uses for AI in online reputation management include offering a
personalized customer experience, countering the spread of fake news and
gathering customer testimonials.370
What Perakakis et al. propose is an innovative intelligent system that would
analyze social media data and automatically draw useful conclusions for
company marketers without needing any human intervention.366 Perakakis et al.
named this system (Social Intelligence Advisor (S.I.A.), and it is based on the
following analysis steps366:

278
 THE A.I. CRUISE LINE

1. “Data gathering: This is the practice of harvesting data from web and
social media. Data is often provided anonymized and it is not exactly as
seen from actual human social media users.”366
2. “AI micro-analysis of gathered data (Sentiment analysis, text analysis,
image recognition, etc.): In this step, each post gathered is individually
analyzed by intelligent algorithms to enrich it with semantic
information. E.g., the sentiment of text, the language it was written, the
gender of the author, the country that it originated, etc.”366
3. “Social intelligent analysis: In this step, the data is treated as a set,
rather than individual posts. In this way, patterns are discovered, and
useful insights can be drawn automatically. This step was only
previously possible in this area by human analysts.”366
4. “Expert advice: Furthermore, the conclusions from step 3 are not only
shown to the users, but customized advice sets can be generated,
personalized for different cases. This advice is aimed at helping the
marketers take action to improve their marketing efforts.”366
5. “Re-evaluate: Since the goal of the system is to generally improve a
brand’s online marketing efforts in social media, it is important to
follow-up and repeat the measurement at regular times. A comparison
can be made between the new values and the old ones in order to
evaluate if metrics show improvement, and advise the marketer again
based on the new results.”366
“The introduction of S.I.A. and ‘intelligent’ features in social media monitoring
allows marketers to focus on the gist of marketing itself,” say Perakakis et al.366
Instead of focusing on collecting and converting data into useful information,
S.I.A. enables marketers to focus on drawing meaning from customers data to
improve company decision making based on engagement reach, sentiment and
other social metrics.366
Using an S.I.A., a cruise line can see engagement rates, reach, as well as the
negative sentiments on a particular post.366 It can help uncover the brand’s fans
and ambassadors that you may have previously gone unnoticed.366 It can reveal
the best days for posting based on past performance.366 “S.I.A. can analyze the
data about the reach of posts with their time and platform to come up with an
effective content posting strategy.”366 An S.I.A. can automatically find hashtags
that perform better in the company’s posts and its industry, so they can be
utilized to increase exposure.366 An S.I.A. can remind the business when reposts
are needed.
An S.I.A. can quickly identify content that has the potential to go viral and it can
remind the business when it needs to repost content.366 This allows a brand to
take appropriate actions before it’s too late.366 The business might be able to
detect a viral event, while it is spreading, potentially helping the virality.366
An S.I.A. can help a business find the most popular site where a competitors is

279
 ANDREW W. PEARSON

appearing and provide advice on how to appear there as well. 366 Through an
S.I.A., a cruise line would be able to capture every site that its competitors
appear on.366 As such, the cruise line could identify the positive impact of such
advertising efforts.366
“S.I.A. in influencers’ discovery lets brands target social media influencers with
followers who are interested in the brand,” say Perakakis et al.366 The application
can identify the follower’s interests, as well as discover influencers through their
contextual relationships with both the brand and amongst their followers.366
AI can also “be applied in monitoring the performance of competitor’s online
campaigns.”366 This helps social media marketers pinpoint how and where a
competitor might be getting their customers.366 “Through monitoring the
performance of competitors’ campaigns, social media marketers are allowed to
know the expectations of their customers,” say Perakakis et al.366 “With an AI-
enabled advertisement analytics tool, social media marketers are aware of the
type of adverts run by their competitors that gain more engagements. With this
knowledge, they are then able to come up with better advertisements and target
the same audience,” offer Perakakis et al.366
“An intelligent social media monitoring tool not only focuses on what
competitors are saying, but such a tool is also capable of highlighting what
customers say about products,” say Perakakis et al.366 Should a brand
competitor’s customers complain about certain elements missing from that
company’s product, the brand’s social media marketers can quickly advertise
one of their products that contain these missing elements, thereby capturing
some market share.366 As if that weren’t enough, companies can also directly
target any complaining clients.366 “In addition, through AI-enabled competitors
monitoring, social media marketers will have the ability to get real-time
intelligence.”366 Comments can be responded to immediately, which should
increase customer conversion.366

Sentiment Analysis
In the TDWI Customer Analytics in the Age of Social Media108 Research report
about the same percentage (30%) of respondents sought to monitor and
measure sentiment drivers. “Sentiment analysis enables organizations to
discover positive and negative comments in social media, customer comment
and review sites, and similar sources. Sentiment analysis often focuses on
monitoring and measuring the ‘buzz’ value, usually through volume and
frequency of comments around a topic.”108 However, it is not just the buzz that
is important, many organizations want more analytical depth so that they can
understand what the buzz is all about, where it comes from, and who is
benefiting the most from it.108

280
 THE A.I. CRUISE LINE

For more sophisticated sentiment analysis, text analytics tools that use word
extraction, natural language processing, pattern matching, and other
approaches to examine social media users’ expressions are employed. 108
“Sentiment analysis can give organizations early notice in real time of factors
that may be affecting customer churn; the research shows that 14% are
interested in monitoring and analyzing social activity in real time.” 108
In 2011, Toyota started testing social media monitoring and sentiment-analysis
tools. After a few years of research, they discovered that by filtering for such
words as “Lexus”, “decide”, “buy” and “BMW”, they were able to quickly identify
active shoppers who were choosing between theirs and their competitor’s
brands.371
Today, Toyota uses social media data analysis across many areas — sales,
service, quality, marketing and product development. 371 For example, if a
customer expresses interest in a car, Toyota “can determine engagement by
analyzing the frequency of dealership visits via their Foursquare check-ins,
understand their dealership experiences, and even understand what features
may have sparked their interest in a competitor's product.” 371
Armed with this information, Toyota stratifies its leads based on their readiness
to buy, moving stronger leads to the top of the funnel and weaker ones to the
bottom.371 By analyzing free-form text, Toyota can learn what customers think
of specific vehicles.371 In the quality area, “Toyota can look for information like
whether new-car owners are hearing a slight rattle and pass that on to their
quality engineers.”371 They are also working on using sentiment analysis to
increase the accuracy of their sales predictions; an important goal, if ever there
was one.371
Cruise lines should keep these ideas in mind when developing their own use
cases. A “rattle” for the cruise line wouldn’t be an engine problem, of course
(except in a company bus, maybe), but rather a poor customer experience on
the ship.
Toyota also wants to deepen its understanding of its customers' other interests,
like what a Camry owners' favorite TV show might be, as well as which other
brands they might like.371 This can help with product placement and brand tie-
ins down the line.371
Sentiment analysis is also key to understanding a competitors’ relative strengths
and weaknesses in the social sphere.108 The TDWI research found that “18% of
respondents are examining social media data to analyze a competition’s ‘share
of voice.’”108
As Joe Mullich explains in his article Opposition Research: Sentiment Analysis as
a Competitive Marketing Tool372:
“When a leading bank wanted to find out how it stacked up

281
 ANDREW W. PEARSON

against competitors, it assumed customers would focus on

lending terms and interest rates. To the bank’s surprise, the
most enthusiastic discourse on blogs and specialized financial
forums related to a smartphone app a competing financial
institution had just put out. The bank had dismissed apps as a
generic marketing gimmick, like the old custom of giving away
a toaster for opening an account. After learning how much
customers valued the app, the bank quickly created its own with
the same prized features as its competitor.”
“You get the benefits of corporate espionage without doing corporate
espionage,” notes Joseph Carrabis, founder of NextStage Evolution, the
company that did the analysis for the bank.372
Sentiment analysis can also provide early insight into a competitor’s new product
initiatives.372 “Very often companies will test market before they release a
product,” says Mullich.372 “And no matter what you get people to sign saying that
they won’t share information, they’ll go online and talk about products they’re
excited about,” warns Mullich.372 You can’t change human nature, but
sometimes you can make it work for you.
In addition, sentiment analysis can alert companies about new competitors who
are bubbling up to the surface or even coming out of left-field.372 Ford would
obviously consider Chevy a competitor, but it might not think of public transport
as being troubling competition.372 However, Carrabis argues that a car company
should realize that it needs to analyze online discussion boards to try to
understand why people are making different transportation choices so it can
change its product offerings or marketing campaigns to emphasize their
customers’ growing environmental concerns. 372 “We have to think broader and
wider than we used to,” Carrabis advises.372 The lesson here is, don’t just look at
your closest competitors as your competition.
This is why it is so imperative for a cruise line to understand how and why people
discuss competitors online. “When car shoppers talk online they don’t talk about
‘quality,’” says Susan Etlinger, an analyst with the Altimeter Group. 372 “They’ll
say, ‘I love the leather interior’ or ‘the cup holder fell out.’ It takes meticulous
work to roll together all the indicators of quality.” 372
Etlinger suggests that “social-media listening teams work with the groups in the
organization that handle keyword search terms and search-engine optimization
effort, since they have a solid grasp on how people online actually talk about the
industry and products.”372
Another thing to keep in mind: “At any point in time, the way people feel about
a brand can be distorted online, because things like Twitter are so volatile and
affected by the news of the day,” warns Etlinger.372 “But over time, you can get
directional trends — why do people love or hate you, how do they feel about

282
 THE A.I. CRUISE LINE

your product compared to the competitor’s products.”372

“My belief is that the sweet spot for social media is not conversion, but
nurturing,” said Brian Ellefritz, vice president of global social media at SAP.108
“Whether it’s in your community, through Twitter, or through Facebook pages,
you want to build an increasing conviction that your company is the one to do
business with,” says Ellefritz.108 “It’s about establishing a belief system that
becomes robust with the support of fans and followers. The question is how you
measure that and create value out of that investment,” he adds. 108
When it comes to setting strategies for customer and social media analytics,
Stodder recommends the following108:
• Use social media data to support an active, not passive social media
strategy. “In competitive, fast-moving markets, organizations cannot
just passively listen to and analyze social media data. The analytics
should plug into strategies for engaging users and customers on social
networks and comment sites. Predictive analytics can help
organizations anticipate the results of active strategies. Special events
such as tweet-ups can build on customer data analysis and create
positive exchanges and engagement.”108
• Take a holistic view of the potential contributions of social media data
analytics. “Understanding behavior in the social sphere can have a
positive impact, not just on marketing and sales functions, but also on
services and other processes in the organization. Marketing executives
should use social media insights to improve brand awareness and
reputation throughout the organization.” 108
• Give CMOs and marketing executives the ability to understand the
financial impact of certain decisions.
• Apply analytics to gain a more accurate understanding of marketing
attribution. “Last-touch” attribution may be easy to affix, but it is not
always reliable. Powerful analytics, along with big data, can help
organizations get a better understanding of what truly affects a
customer’s purchase decision.
In its article Sentiment Analysis: Types, Tools, and Use Cases373, Altexsoft states
that the goal of sentiment analysis is “to know a user or audience opinion on a
target object by analyzing a vast amount of text from various sources.” “It’s not
only important to know social opinion about your organization, but also to define
who is talking about you,” says Altexsoft. 373 Measuring mention tone can also
help define whether industry influencers are discussing a brand and, if so, in
what context. The power of sentiment analysis software is it can do all of the
above in real time and across all channels, thereby making it useful for both
sentiment analysis and customer service.373
“You can analyze text on different levels of detail, and the detail level depends

283
 ANDREW W. PEARSON

on your goals,” says Altexsoft.373 “For example, you may define an average
emotional tone of a group of reviews to know what percentage of customers
liked your new clothing collection,” explains Altexsoft. 373 “If you need to know
what visitors like or dislike about a specific garment and why, or whether they
compare it with similar items by other brands, you’ll need to analyze each review
sentence with a focus on specific aspects and use or specific keywords,” add
Altexsoft.373
When a brand wants to analyze sentiment, it first needs to gather all relevant
brand mentions into one document.373 Selection criteria must be carefully
considered – should mentions be time-limited, should only one language be
used, should specific locations be locked in, etc., etc. 373 Data must then be
prepared for analysis, read, cleansed, and any irrelevant content should be
excluded from the analysis.373 Once the data has been prepared, full analysis can
begin and sentiment extracted.373 Of course, since hundreds of thousands or
even millions of mentions may need analysis, the best practice is to automate
this tedious work with software and many of the tools I have mentioned
throughout this book can help.373 I have also included a list of social media
monitoring tools at the end of this section.
Altexsoft mentions various customer experience software, such as InMoment
and Clarabridge that “collect feedback from numerous sources, alert on
mentions in real-time, analyze text, and visualize results.”373 “Text analysis
platforms (e.g. DiscoverText, IBM Watson Natural Language Understanding,
Google Cloud Natural Language, or Microsoft Text Analytics API) have sentiment
analysis in their feature set,” adds Altexsoft.373
“InMoment provides five products that together make a customer experience
optimization platform,” explains Altexsoft.373 “One of them, Voice of a Customer,
allows businesses to collect and analyze customer feedback in a text, video, and
voice forms. The number of data sources is sufficient and includes surveys, social
media, CRM, etc.,” says Altexsoft.373
Clarabridge is a CEM platform that “pulls and analyzes text from chats, survey
platforms, blogs, forums, and review sites,” notes Altexsoft. 373 “Users can also
gain insights from emails, employee and agent notes, call recordings and
Interactive Voice Response (IVR) surveys: The system can convert them into
text.”373 Clarabridge provides social media listening as well. 373 According to
Altexsoft, “The system considers industry and source, understanding the
meaning and context of every comment. Sentiment analysis results display on
an 11-point scale. Users can modify sentiment scores to be more business-
specific if needed.”373
Another useful platform is DiscoverText, “a cloud-based collaborative text
analytics system for researchers, entrepreneurs, and governments.” 373
“Capterra users note the solution is great for importing/retrieving, filtering, and

284
 THE A.I. CRUISE LINE

analyzing data from various sources, including Twitter, SurveyMonkey, emails,

and spreadsheets,” says Altexsoft.373
IBM Watson Natural Language Understanding is a set of advanced text analytics
systems that supports analysis in 13 languages.373 “Analyzing text with this
service, users can extract such metadata as concepts, entities, keywords, as well
as categories and relationships,” says Altexsoft. 373 “It also allows for defining
industry and domain to which a text belongs, semantic roles of sentence parts,
a writer’s emotions and sentiment change along the document,” says
Altexsoft.373 Tools for developers to build chatbots and other NLP solutions are
provided using IBM Watson services.373
“Microsoft Text Analytics API users can extract key phrases, entities (e.g. people,
companies, or locations), sentiment, as well as define in which among 120
supported languages their text is written,” explains Altexsoft. 373 “The Sentiment
Analysis API returns results using a sentiment score from 0 (negative) to 1
(positive),” explains Altexsoft.373 The software can detect sentiment in English,
Spanish, German, and French texts.373 Developers recommend that “the analysis
be done on the whole document and advise using documents consisting of one
or two sentences to achieve a higher accuracy.”373
Google Cloud Natural Language API can “extract sentiment from emails, text
documents, news articles, social media, and blog posts.” 373 It can also extract
“insights from audio files, scanned documents, and documents in other
languages when combined with other cloud services.”373 “The tool assigns a
sentiment score and magnitude for every sentence, making it easy to see what
a customer liked or disliked most, as well as distinguish sentiment sentences
from non-sentiment sentences,” notes Altexsoft.373
Competitive analysis that involves sentiment analysis can also help cruise lines
understand their strengths and weaknesses and maybe find ways to stand out
from the crowd.373 In times of crisis, sentiment analysis can be instrumental in
helping douse the flames.
Altexsoft believes, “There is one thing for sure you and your competitors have in
common — a target audience.”373 Brands can track and research how society
evaluates competitors just as they analyze attitudes towards their business.
“What do customers value most about other industry players? Is there anything
competitors lack or do wrong? Which channels do clients use to engage with
other companies?” — these are all important questions that sentiment analysis
help answer.373 Brands can use this knowledge to improve “communication and
marketing strategies, overall service, and provide services and products
customers would appreciate.”373
Most brands grapple with the question of how to bring a desired product to the
market?373 The only approach, claims Altexsoft, is to ask people what they
want.373 Successful companies build a minimum viable product (MVP), gather

285
 ANDREW W. PEARSON

early feedback, and continuously try to improve a product, even after its
release.373 “Feedback data comes from surveys, social media, and forums, and
interaction with customer support,” argues Altexsoft. 373 Sentiment analysis can
be extremely handy here.373 It helps brands learn about product advantages and
drawbacks.373 Armed with strong sentiment analysis results, “a product
development team will know exactly how to deliver a product that customers
would buy and enjoy.”373
Using sentiment analysis, marketers can study consumer behavior patterns in
real time, which can help to predict future brand trends.373 “Another benefit of
sentiment analysis is that it doesn’t require heavy investment and allows for
gathering reliable and valid data since its user-generated,” says Altexsoft.373
Sentiment analysis lets businesses harness an enormous amount of free data to
help them understand their customers’ attitude towards their brand. 373 This
analysis can take customer care to the next level.

Sentiment Analysis Tools

Table 11 lists the Social Media Tools and websites available to business users to
track engagement and customer feedback.
Name Comments

BoardReader allows users to search multiple message boards

simultaneously, allowing users to share information in a truly
global sense. Boardreader is focused on creating the largest
Board Reader repository of searchable information for our users. Users can find
answers to their questions from others who share similar interests.
Our goal is to allow our users to search the “human to human”
discussions that exist on the Internet.
Buffer makes your life easier with a smarter way to schedule the
great content you find. Fill up your Buffer at one time in the day
and Buffer automagically posts them for you through the day.
Buffer
Simply keep that Buffer topped up to have a consistent social
media presence all day round, all week long. Get deeper analytics
than if you just post to social networks directly.
Analyze what content performs best for any topic or competitor.
Find the key influencers to promote your content:
• Discover the most shared content across all social
networks and run detailed analysis reports.
Buzzsumo • Find influencers in any topic area, review the content
they share and amplify.
• Be the first to see content mentioning your keyword; or
when an author or competitor publishes new content.
• Track your competitor’s content performance and do
detailed comparisons.

286
 THE A.I. CRUISE LINE

Name Comments

Can help you organize, increase, and manage your followers, and
can do so across multiple accounts and profiles. At a glance you can
Commun.it see different aspects of your community management, like the
latest tweets from your stream and which new followers might
appreciate a welcome message.
Crowdfire is a powerful phone app and online website that helps
Crowdfire you grow your Twitter and Instagram account reach. This tool has
a variety of functions designed to understand your social analytics
as well as manage your social publishing.
Cyfe is an all-in-one dashboard software that helps you monitor
Cyfe and analyze data scattered across all your online services like
Google Analytics, Salesforce, AdSense, MailChimp, Facebook,
WordPress and more from one single location in real-time.
Shows a variety of valuable information related to your Facebook
page, such as growth, engagement, service and response time, and
Fanpage Karma of course Karma (a weighted engagement value). FanKarma also
provides insight into Twitter and YouTube; the latter could be
particularly valuable if you're creating a video marketing strategy.
Followerwonk is a cool social media analytics tool thet lets you
explore and grow your social graph. Dig deeper into Twitter
Followerwonk analytics: followers, their locations, when do they tweet. Find and
connect with influencers in your niche. Use visualizations to
compare your social graph to competitors.
Google Alerts are email updates of the latest relevant Google
results (blogs, news, etc.) based on your searches. Enter the topic
Google Alerts you wish to monitor, then click preview to see the type of results
you’ll receive. Some handy uses of Google Alerts include:
monitoring a developing news story and keeping current on a
competitor or industry.
Trends allows you to compare search terms and websites. With
Google Trends you can get insights into the traffic and geographic
Google Trends visitation patterns of websites or keywords. You can compare data
for up to five websites and view related sites and top searches for
each one.
Monitor and post to multiple social networks, including Facebook
and Twitter. Create custom reports from over 30 individual report
modules to share with clients and colleagues. Track brand
Hootsuite sentiment, follower growth, plus incorporate Facebook Insights
and Google analytics. Draft and schedule messages to send at a
time your audience is most likely to be online. HootSuite has the
dashboard for your iPhone, iPad, BlackBerry and Android.
Monitor and post to multiple social networks, including Facebook
HowSocialable and Twitter. Create custom reports from over 30 individual report
modules to share with clients and colleagues. Track brand
sentiment, follower growth, plus incorporate Facebook Insights

287
 ANDREW W. PEARSON

Name Comments

and Google analytics. Draft and schedule messages to send at a

time your audience is most likely to be online. HootSuite has the
dashboard for your iPhone, iPad, BlackBerry and Android.
Key metrics about your Instagram account. Number of likes
received, your most liked photos ever, your average number of
Iconosquare likes and comments per photo, your follower growth charts and
more advanced analytics. Track lead conversations, send private
message as on Twitter, and improve communication with your
followers.
Social media monitoring, analytics and reporting. Influencer
marketing, find and create relationships with the top influencers in
Klear your sector and build your community. Competitive analysis tracks
your social media landscape, see what’s working for them and
develop your strategy.
Klout’s mission is to help every individual understand and leverage
Klout their influence. Klout measures influence in Twitter to find the
people the world listens to. It analyzes content to identify the top
influencers.
Kred is a social-media scoring system that seeks to measure a
person’s online influence. Kred, which was created by the San
Francisco-based social analytics firm PeopleBrowsr, attempts to
Kred also measure a person or company’s engagement, or as they call it,
outreach. PeopleBrowsr hopes that that combination can offer a
more informed metric for non-celebrities like entrepreneurs and
those whom they follow and look to for advice.
This Facebook analysis tool comes up with stats and insights into
your page and begins every report with a list of recommendations.
LikeAlyzer Keep track of where your Facebook page stands compared to other
pages by following the comparison to average page rank, industry-
specific page rank, and rank of similar brands.
Mention prides itself on “going beyond Google Alerts” to track
absolutely anywhere your name or your company might be
Mention mentioned online. When you subscribe to Mention's daily email
you get all these wayward hits right in your inbox, and the Web
dashboard even flags certain mentions as high priority.
Explore your Twitter network. Discover which people interact the
most and what they’re talking about. It’s also a great way to find
relevant people to follow. The visualization runs right in your
Mentionmap browser and displays data from Twitter. Mentionmap loads user’s
tweets and finds the people and hashtags they talked about the
most. In this data visualization, mentions become connections and
discussions between multiple users emerge as clusters.
Must Be Built by the team at Sprout Social, Must Be Present searches your
Present Twitter account to find how quickly you respond to mentions. Their
engagement reports place you in a percentile based on other

288
 THE A.I. CRUISE LINE

Name Comments

accounts so you can see how you stack up to the speed of others.

A super-powered Twitter search tool, NeedTagger runs language

filters and keyword searches to determine which Twitter users
NeedTagger
might need your products or services. The tool shows you real-time
search results and sends a daily email digest of new finds.
Collects your activity on Facebook, LinkedIn, and Twitter (and even
places like Yelp and Foursquare) to provide an email overview of
your accounts. You set how often and when you want to receive
NutshellMail the recap emails. Put it to use: If you have a weekly metrics plan
you can have NutshellMail send a message once a week with an
overview of your accounts. You can then extract the data and
insights straight into your weekly report.
Omgili helps you find interesting and current discussions, news
Omgili stories and blog posts. Direct access to live data from hundreds of
thousands of forums, news and blogs. Very easy to use, no signup
for web interface.
Find out how many people are pinning from your website, seeing
your pins, and clicking your content. Pick a time-frame to see how
Pinterest your numbers trend over time. Get better at creating Pins and
Analytics boards with metrics from your Pinterest profile. Learn how people
use the Pin It button on your site to add Pins. See how people
interact with your Pins from whatever device they use. Get a glance
at your all-time highest-performing Pins.
Pluggio is a web-based social media tool to help marketers easily
grow and manage their social media profiles (Facebook and
Pluggio Twitter). It includes a suite of tools to organize and keep track of
multiple accounts, get more followers, and automate the finding
and publishing of excellent targeted content.
The full set of social media tools. Post content to over 10 social
networks with one single click of a button. Get real time click-
Postific through statistics with your domain name. Measure and analyze
the best results from your social posts. Monitor the social media
conversations that are important for your business.
Quintly is the professional social media monitoring and analytics
solution to track and compare the performance of your social
Quintly media marketing activities. Whether you are using Facebook,
Twitter or both, Quintly monitors and visualizes your social media
marketing success. Benchmark your numbers against your
competitors or best practice examples.
Sentiment was born in 2007 and now boasts a team of bright
Sentiment enthusiastic people dedicated to provide the best social customer
service and engagement platform for business.
SocialMention SocialMention tracks areas such as sentiment, passion, reach, and
strength to not just tell you what's being said about your search

289
 ANDREW W. PEARSON

Name Comments

but how those reactions feel. While you track your brand or
yourself, you can also see how your sentiment changes over time.
Identifies your top 10 followers in three specific areas: Best
Followers, Most Engaged, and Most Valuable. Your most engaged
followers are those who interact with you most often (replies,
Social Rank retweets, and favorites); your most valuable followers are the
influential accounts; and your best followers are a combination of
the two. Social Rank will run the numbers for free and show you
the results today, then follow-up each month with an email report.
Schedule tweets, track keywords, extended Twitter profiles, save
Social Oomph and reuse drafts, view @mentions and retweets, purge your DM
inbox, personal status feed — your own tweet engine, unlimited
accounts.
Keeps track of your hashtag campaign or keyword on Twitter,
This tracking
Instagram, or Facebook with a full dashboard of analytics,
tool
demographics, and influencers.
TipTop Search is a Twitterbased search engine that helps you
discover the best and most current advice, opinions, answers for
Tip Top any search, and also real people to directly engage and share
experiences with. A search on any topic reveals people’s emotions
and experiences about it, as well as other concepts that they are
discussing in connection with the original search.
A powerful search engine for Twitter content. Want to know how
Topsy a certain term is being used on Twitter? You can search links,
tweets, photos, videos, and influencers.
Offers real-time monitoring and analytics for Twitter on any name,
keyword, or hashtag you choose. The Twazzup results page delivers
Twazzup
interesting insights like the top influencers for your keyword and
which top links are associated with your search.
Has a number of useful Twitter features, many of which fall into a
couple categories: managing your followers and supercharging
who you're following. For management, you can unfollow in
Tweepi
batches those who don't follow you back, and you can bulk follow
another account's complete list of followers or who they're
following.
A Twitter management tool for iOS and Android devices and
provides the basics of what you'd expect from a Twitter dashboard
Tweetcaster
plus a few fun extras: enhanced search and lists, hiding unwanted
tweets, and photo effects for your images.
Lets you track, organize, and engage with your followers through a
Tweetdeck customizable dashboard where you can see at a glance the activity
from different lists, followers, hashtags, and more.

290
 THE A.I. CRUISE LINE

Name Comments

Shows you the reach and exposure of the tweets you send,
collecting data on who retweets you and the influence of each.
TweetReach
Identify which of your tweets has spread the furthest (and why)
and then try to repeat the formula with future tweets.
Twitter Counter is the number one site to track your Twitter stats.
Twitter Counter provides statistics of Twitter usage and tracks over
TwitterCounter 14 million users. Twitter Counter also offers a variety of widgets
and buttons that people can add to their blogs, websites or social
network profiles to show recent Twitter visitors and number of
followers.
Provides a snapshot of your Twitter profile and can even track
Facebook and Instagram as well. Two of Twtrland's most helpful
tools are a live count of how many followers are currently online
Twtrland and advanced search functionality that includes keywords,
locations, and companies. Local companies can perform a location
search to see which area accounts are most popular and potentially
worth following.
SumAll is a powerful social media analytics tool that allows our
customers to view all of their data in one simple, easy-to-use
SumAll visualization. Social media, e-commerce, advertising, e-mail, and
traffic data all come together to provide a complete view of your
activity.
Pin Alert feature lets you track what are people pinning from your
website, who is pinning the most and what images from your
website are trending on Pinterest. Thousands of social media
ViralWoot
marketers and agencies use Viralwoot for their clients. You can
manage & grow multiple Pinterest accounts with a single Viralwoot
account.
WhosTalkin is a social media monitoring tool that lets you search
for conversations surrounding the topics that you care about most.
Whether it be your favorite sports team, food, celebrity, or brand
WhosTalkin name; Whostalkin will help you find the conversations that are
important to you. WhosTalkin search and sorting algorithms
combine data taken from over 60 of the most popular social media
sites.

Table 11: Social Media Tools

Source: Dreamgrow.com374

Social Media Monitoring

It is high time to revise Wanamaker’s oft-made quote that he didn’t know which
half of his marketing spend was useful (as it is probably the most over-quoted

291
 ANDREW W. PEARSON

quote in the history of marketing) because now we not only have the ability to
figure out which advertisement works for which customer, but we can also
extrapolate how that advertising will work on customers similar to the ones we
might want to target.
Today, digital advertising should employ a multi-screen strategy that follows its
audience throughout his or her digital day. As previously mentioned, successful
mobile advertising requires three things — reach, purity and analytics. Analytics
“involves matching users’ interests — implicit and explicit, context, preferences,
network and handset conditions — to ads and promotions in real time.”38
A cruise line operator can utilize analytics to enhance its marketing, campaign
management, increase sales, conduct market research, ferret out fraud and for
risk/management, contact center operations, supply chain management, as well
as for a whole host of other things.
In their Measuring Social Media Performance and Business Impact (Part 1) 375,
Hamill and Stevenson put forth their ‘6Is’ of social media monitoring framework
that include:
1. Involvement – the number and quality of customers involved in your
various online networks
2. Interaction — the actions taken by online network members – read,
post, comment, review, recommend, etc.)
3. Intimacy — the brand sentiments expressed, level of brand ‘affection’
or ‘aversion’
4. Influence — advocacy, viral forwards, referrals, recommendations,
retweets, etc.
5. Insights — the level of customer/actionable insight delivered from
monitoring online conversations
6. Impact — business impact of your social media activities benchmarked
against core business goals and objectives.
In 1999, The Cluetrain Manifesto364 warned, “Reviews are the new advertising.”
Today, this is truer than ever before. There are a multitude of platforms that
allow users to rate or comment on a restaurant, a retail establishment, a hotel,
a cruise line property, or even a local handyman or plumber.
Used properly, reviews can be the new advertising currency for a cruise line’s
marketing department. Companies such as Dell, Cisco, Salesforce.com, the
American Red Cross, and Gatorade are creating Social Media command centers
that monitor the social conversations about their companies. These social media
centers enable company employees to monitor conversations from the social
web on channels such as Twitter, Facebook, and YouTube, amongst others, in an
attempt to keep track of the health of a company’s social brand.
In his article Taking Back The Social-Media Command Center376, Scott Gulbransen

292
 THE A.I. CRUISE LINE

argues that, “To do the command-center model right, a setup has to envision a
real-time workflow empowered to take action on all of the relevant content
being analyzed, whether it be insights derived from real-time monitoring,
opportunities to respond, or great discovered content to feature that elevates
you and your fans.” Gulbransen recommends breaking down a command center
into the following critical functions376:
1. Identify trends and insights — track not only the key themes, but also
how they evolve over time.
2. Review the content — monitor a wide variety of terms that are
meaningful to the brand and assign employees to sort through the
responses, deciding which one warrants a response, and what might
interest the community at large.
3. Curate the best stuff — leverage the great content that is being said
about the company as well as champion those great content providers.
4. Listen and Respond — this is a two-way conversation, listen and
respond quickly and accordingly.
Unlike casual conversations, comments, updates, likes and dislikes uploaded to
social networks are collected and, therefore, analyzable, and measurable. This
results in “a data tsunami: the actions and content generated by participants in
social media create ‘Big Data’ sources that are full of potential for tracking and
understanding behavior, trends, and sentiments.” 108 Remember, this can be
highly quantifiable data. Cruise operators should be studying attribution analysis
for their social media campaigns on platforms like Facebook, YouTube, Twitter,
Weibo, etc., etc.
Getting people to actually state their feelings and opinions about a product is
paramount and it can help with attribution analysis, which can reveal such things
as what kinds of campaigns most influence customer behavior. 108 In digital
advertising, attribution is traditionally done at a user-specific level, where a
consistent user identifier can be established across all analyzed events. In
traditional media, attribution is generally done at the macro, user-group level,
as there is no consistent user identifier available.
In its Social Media Analytics: Making Customer Insights Actionable377, IBM
believes that the “mistake many organizations make is to treat social media as
distinct and separate from other customer data and divorced from revenue
generating imperatives.” IBM recommends companies venturing into the social
media space do the following377:
• Integrate company-wide information from different data sources to
drive the business through deeper consumer insight.
• Define the real value of the company’s brand — its equity, reputation
and loyalty — at any moment in time, in any place in the world; and

293
 ANDREW W. PEARSON

• Understand emerging consumer trends, both globally and locally and

apply predictive models to determine actions with the highest
probability of increasing relevance and maximizing marketing campaign
ROI.
IBM recommends businesses ask the following questions when devising a social
media plan377:
• Assess — also referred to as “listening,” at this stage a company should
monitor social media to uncover sentiment about its products, services,
marketing campaigns, employees and partners. The questions that
need to be asked at this stage include:
o What are you company objectives? Are you looking to:
▪ Attract customers?
▪ Increase the value of existing customer relationships?
▪ Retain customers?
o How do customers interact with you today?
o What are they interested in?
o Where and when do they use social media?
o Are there significant influencers who speak to your brand or
products?
• Measure — proactive analytics can uncover hidden patterns that can
reveal “unknown unknowns” in the data. Questions that businesses
need to ask at this stage include:
o Who are you targeting with your social media initiatives and
why?
o What will you be measuring:
▪ Share of voice
▪ Activation
▪ Brand sentiment
▪ Influencers
o Sales over the life of the customer relationship?
• Integrate — social media can give businesses both a broad view of their
operations as well as a detailed and intimate view of their individual
customers. Questions to ask at this stage include:
o What is your vision for social media and its integration into the
company’s operational marketing systems?
o Do you have a profile of your customer advocates? Can you
predict sentiment on products, services, campaigns?
o How do you measure the effects of social media on brand
equity and reputation, pipeline, and sales orders and margins?
o How will you integrate social analytics into other customer
analytics?
Regardless of the sophistication and scope of any social media initiative, the end

294
 THE A.I. CRUISE LINE

goal, IBM argues, should be in alignment with corporate imperatives and goals
as well as produce a measurable ROI.377

Social Media Marketing

In their article An Exploratory Study of Gambling Operators’ Use of Social Media
and the Latent Messages Conveyed378, Gainsbury et al. argue that social media
has become a powerful marketing channel because it “enables gambling
operators to promote products and brands with fewer constraints than in
traditional forms of media.”378 The study attempts to quantify features of social
media presence among several popular Australia gambling operators, a majority
of which did have some social media presence; Facebook was the most popular
social media platform used.378
In the Gainsbury et al. study, “Information posted on Facebook and Twitter was
inspected to examine content promoted or discussed via the respective social
media channels. Many operators posted the same or similar content across
various social media platforms or linked between these. This was particularly the
case for Facebook and Twitter linking to YouTube videos.”378
According to Gainsbury et al., the types of content posted on social media sites
included the following378:
• Information about the venue/operator, including information on
gambling facilities as well as other non-gambling services available,
contact details, and hours of operation.
• Promoting gambling products, offers, and specific events to encourage
users to place bets. “Posts involved text and graphics that illustrate the
types of gambling products available, how to use these, potential
returns, and types of customers who use the products.”378 Land-based
venues often posted information about in-venue events, such as bingo
tournaments and upcoming jackpots.”378
• Posts about competitions and promotions.
• Promoting gambling wins.
• Promoting features to assist with betting, including betting and
payment options.
• Betting tips
• Sports and racing news
• Promoting in-venue events, which include drawing users’ attention to
special or regular weekly events, encouraging interaction with users.
• Promoting food and beverages.
• Encouraging customer engagement by encouraging users to follow, like,
and share their own posted content.
• Links to sports teams.

295
 ANDREW W. PEARSON

• Promoting community engagement.

Many gambling operators also posted content unrelated to gambling, but this
was “designed to be engaging, humorous, and encourage likes and sharing
among user networks.”378 For example, “cartoons and images were often used
to make jokes about the frenetic and mundane chaos as well as the drudgery of
family life, with betting presented as a superior alternative option (e.g., a
Facebook post by IASbet.com featuring a photo of a man surrounded by children
thinking about being at the race with the caption ‘Where would you rather be’,
May 16, 2013).”378
In its Q2 2019 Paid Search and Paid Social Benchmark Report 379, Adstage
provides benchmark data as well as trend analysis for multiple social media
channels, offering a good overview of the current social media marketing
environment.
“Facebook may soon be making changes to the News Feed itself according to
tests it’s been running,” says Adstage.379 Facebook has admitted that there’s no
more room for ads in News Feed, and as users get more comfortable with
Swipeable Stories, it may be where Facebook puts its focus next.379
In an effort to strengthen privacy, Facebook is letting users clear their entire
histories, but marketers fear it could mess with targeting options.379 “The option
would give people the ability to see and delete the data that third-party sites and
apps share about them. For example, Custom Audiences relies in part on data
about visits people make to an advertiser’s website or app,” says Adstage.379
When one hears “influencers”, one automatically thinks of Instagram, a fact that
the social media platform is well aware of and now wants to utilize.379 That’s
why Instagram now make influencer posts shoppable.379 According to Adstage,
“Influencers can now use the app to tag and sell products directly. Previously,
only Instagram pages owned by brands could link out to specific products using
‘shoppable’ posts.”379 Advertisers can now work in concert with influencers to
sell products on their behalf.379 “Creators and brands will receive ‘shared
insights’ from Instagram.”379
Instagram also allows advertisers to boost Branded Content Partnerships as
ads.379 As Instagram explained in their announcement: “We're introducing the
ability for advertisers to promote creators' organic branded content posts as
feed ads. 68% of people say they come to Instagram to interact with creators.
With branded content ads, businesses have an opportunity to tell their brand
stories through creators' voices, reach new audiences and measure impact.”379
“Previously, branded content partnership posts would only reach the followers
of the influencer. Branded content ads let advertisers promote the posts just like
they would any other ad,” contends Adstage.379
In an effort to boost privacy, “Google is bringing users more transparency around

296
 THE A.I. CRUISE LINE

how they’re being tracked across the web with cookies, which could negatively
affect advertisers.”379 Some analysts fear Google might now give its own tracking
preferential treatment.379 With these new changes, “Chrome will let users clear
all tracking cookies, while not affecting single domain cookies – those are the
cookies that store info like logins and settings.”379 “Users will also be able to see
exactly which sites are setting the tracking cookies.”379
Google is also “prohibiting another type of online tracking called fingerprinting,
or device recognition.”379 But similarly, this will only aid Google's bottom line.379
Without fingerprinting, the amount of data third parties can collect on internet
users is limited.379
On the product side, Google’s announced a new mobile search ad format that
queues up giant gallery ads.379 According to Adstage, advertisers can feature up
to eight images in the ads, to “make it easier for you to communicate what your
brand has to offer,” said Prabhakar Raghavan, SVP of Google Ads and
Commerce.379 “Google is also rolling out a mobile search redesign featuring a
black ‘Ad’ label, which replaces the green outlined label and a new look for
favicons on organic listings.”379 “New advanced location targeting is meant to
reach commuters, whether they’re at home, at work, or on the road,” said
Adstage.379 “Marketers will see the targeting option label has changed from
‘People in your targeted locations’ option to ‘People in or regularly in your
targeted locations.’ Now businesses can reach people not just when they’re
physically in the targeted locations at search time, but if they regularly pass
through the targeted location.379
Finally, one of Google’s most used products – maps – is courting advertisers.379
Over the past 14 years ago, maps has been relatively free of ads, but now the
app will regularly highlight sponsored locations and show additional paid listings
when people search for nearby businesses.379 Also, for the last two years,
“Google has also tested ‘promoted pins,’ which show an advertiser’s brand
regardless of whether or not the user searched for that business.”379
“If there’s any platform that believes in the power of account-based marketing,
it’s LinkedIn,” explains Adstage.379 The company has “teamed up with Adobe and
Microsoft to accelerate account-based marketing for their joint customers.”379
“LinkedIn says that by extending account-based marketing capabilities to Adobe
Experience Cloud users, they’re creating a seamless way for them to identify and
target the right audiences on LinkedIn with meaningful content.”379 This is great
news for B2B marketers as “Adobe and Microsoft are also working together to
align key data sources to populate account-based profiles from Adobe
Experience Cloud, including Marketo Engage and Microsoft Dynamics 365 for
Sales.”379 This makes it “much easier for B2B marketers to easily identify,
understand, and engage B2B customer buying teams.”379
“In an effort to demonstrate a broader focus on ad inventory, data, and targeting

297
 ANDREW W. PEARSON

capabilities, Bing Ads rebranded itself to Microsoft Advertising,” explains

Adstage.379 In the announcement post, Microsoft said, “It’s a simple shift
because our clients and partners already know us as Microsoft, and many are
already tapping into our new advertising products that go above and beyond
search, such as the Microsoft Audience Network.” 379
“Microsoft Advertising also debuted Sponsored Products, which allow
manufacturers to promote their products in shopping campaigns with their
cruise line partners.”379 According to Adstage, “this gives manufacturers access
to new reporting and optimization capabilities, and retailers get additional
product marketing support with a fair cost split.”379 Also, “Microsoft's new
Chromium-based Edge browser, an overhauled app now available for Windows,
MacOS, Android and iOS that's based on Google's open-source browser
technology will block intrusive ads.”379 This is another clear sign that browsers
are now competing to make the online experience better – and more private –
for consumers, even if it deters businesses.379 “Product managers say Edge will
also go a step further with a privacy control that consumers could use to block
advertisers and publishers from tracking them across the web. The feature offers
three levels of tracking constraints: unrestricted, balanced and strict,” concludes
Adstage.379
In its Q3 2019 Paid Search and Paid Social Benchmark Report380, Adstage reports
that “Facebook continued on its quest to make users feel more comfortable on
the platform by making it easier for them to understand why an ad was being
shown.” “Now when someone clicks on ‘Why am I seeing this ad?’ they’ll see
detailed information like interests or categories they’ve shown an affinity for
that may have been the reasons they were targeted for the ad,” explains
Adstage.380 Facebook will also be “more upfront about where it received
information about a user’s interests.”380 “That means anyone can find out which
advertiser uploaded a list with their information and used it to run an ad in the
past 7 days,” says Adstage.380
“Facebook is finding more ways to pack advertising opportunities into its pages,”
notes Adstage.380 “In Q3, Facebook started letting more advertisers place ads in
Facebook’s search results.”380 “‘Facebook search’ now pops up for some
advertisers when they’re creating a Newsfeed campaign. That means businesses
have to place ads in Facebook search results and in the Newsfeed.”380 “When a
business places an ad in Facebook search, it shows up for search terms related
to the business’s offerings,”380 which means a Facebook business page should be
highly detailed and updated constantly.380
In Q3, Facebook also rolled out automated lead generation capabilities in
Messenger.380 “After tapping on a click-to-Messenger ad, users go through an
automated series of questions that can be answered with pre-filled or free form
responses,” explains Adstage.380 “To continue the process, businesses can
integrate Messenger with their CRM provider to capture leads info. Pages can

298
 THE A.I. CRUISE LINE

also manually continue the conversation through Pages Inbox, Pages Manager
App, or a third-party live chat provider,” notes Adstage.380
Google recently updated “its Keyword Planner tool so that it now shows the most
relevant keyword ideas based on the seed keyword.”380 “The update is designed
to help campaign managers prioritize which keywords to add to their Google Ads
account,” says Adstage.380 Also, “Google Ads will now only suggest
recommended keywords if they’re estimated to drive additional traffic beyond
existing keywords.”380 In addition, Google Ads’ keyword recommendations can
now include broad match modifiers.380 “Google Ads also expanded close variants
to include words that have the same meaning as the original keyword.”380 “Now,
close variants can match for queries that don’t contain the keywords at all, as
long as they share the same meaning.”380
On the analytics side, “Google is making it easier to measure users’ interactions
across platforms with unified app and website analytics.”380 “’Reports for this
new property use a single set of consistent metrics and dimensions, making it
possible to see integrated reporting across app and web like never before. Now
you can answer questions like: Which marketing channel is responsible for
acquiring the most new users across your different platforms?” says Google.380
According to Adstage, Google officially launched Gallery Ads in beta in the 3rd
quarter of 2019.380 “Gallery ads are a swipeable, image-based ad unit that appear
at the top of search results. Much like Carousels on Facebook, they’re designed
to give businesses more real estate to showcase their brand visually with the
ability to include up to 8 images,” says Adstage.380
“There’s a new ad unit available to advertisers on YouTube — masthead ads,”
notes Adstage.380 “Previously, the only way to buy YouTube masthead ads was
to do a full day takeover. That meant everyone who visits YouTube on a
particular day would see the ad. You can still do that, but the cost is enormous,
and there are no targeting options. Now advertisers can buy the YouTube
Masthead on a cost-per-thousand impression (CPM) basis and use advanced
audience solutions to customize who sees it,” says Adstage.380
YouTube’s new “Video Reach campaigns let advertisers upload multiple videos
for a single campaign.”380 “The capabilities include three different asset types —
six-second bumper ads, skippable in-stream ads, and non-skippable in-stream
ads. The units are available on a CPM-basis and currently run on YouTube’s
desktop and mobile platforms,” says Adstage.380
In Q3, “LinkedIn gave marketers a handful of exciting new advertising goals and
insights capabilities.”380 “The first of these was three new ways for marketers to
achieve their advertising goals. In Campaign Manager, marketers can now
optimize their LinkedIn marketing campaigns for the following objectives:
• Brand awareness

299
 ANDREW W. PEARSON

• Website conversions
• Job applicants”380
With this change, LinkedIn is optimizing its click pricing to align with the selected
objective of helping marketers best utilize their budgets.380 “For example, if your
objective is website clicks, you’ll only be charged for clicks that go to a landing
page or your website,” explains Adstage.380
“In Q3, Twitter introduced a 6-second video bid unit where advertisers will be
charged only if their ad is viewed for 6 seconds, with pixels at 50% in view,” says
Adstage.380 Twitter also revealed that, “according to their data, a tweet can
attract 10x the engagement as a tweet without a video.”380 Even videos of non-
professional quality could be just as engaging as something that’s more
polished.380
Microsoft Advertising is giving marketers the ability to see where their ads are
showing up in search results seems like a given.380 Also, new metrics include380:
• Top impression share.
• Top impression share lost to rank.
• Top impression share lost to budget.
• Absolute top impression share.
• Absolute top impression share lost to rank.
• Absolute top impression share lost to budget.
“A top impression is anything in the top section of search results. An absolute
top impression means the ad was the first thing shown in search results,”
explains Adstage.380
Microsoft is giving users the ability to test it with a feature called
“Experiments”.380 “As the company explained in the announcement,
‘Sometimes, it’s not immediately clear whether a new bidding strategy, setting,
or feature is the best move for you… With experiments rolling out globally, you
can now test out those campaign changes with full confidence.’”380 “This gives
advertisers the ability to test changes without launching them across the whole
campaign for a pretty effective A/B test,” notes Adstage.380
Microsoft is also giving “marketers more real estate in Dynamic Search Ads with
longer titles and descriptions. This makes their ads more comparable to what
Google’s ads provide after their recent headline, description, and character
expansions.”380 The update gives marketers380:
• An additional description field.
• Up to 90 characters each for the two descriptions.
• Longer dynamically generated ad titles.
In another bid to keep pace with Google, Microsoft opened up Responsive search
ads – ads that automatically adjust themselves according to the search query –

300
 THE A.I. CRUISE LINE

to all advertisers,” says Adstage.380 Previously it was only for select accounts.380
Finally, Microsoft built in a new audience targeting option.380 “The new method
combines customer data with technology to hit people with ads when they’re
most ready to buy by finding prospects who are similar to those who already buy
from you. Microsoft says the new option should give marketers a boost in the
performance of shopping campaigns with better conversion rates and lower
CPA,” notes Adstage.380

Conclusion
Social media listening can provide a cruise line with an ongoing real-time window
into customer sentiment, as well as give the business verifiable information
about the company’s marketing campaigns, brands, and services.
The beauty of this system is that it can be a real win-win when it comes to a
company’s marketing plan as customers who are happy with a business’s
products and/or services will often comment and blog about the products
and/or services they like, while those who aren’t happy with it can be reached
out to and, hopefully, converted into satisfied customers. Often, the simple act
of responding to a customer’s comments can turn the tide of negativity and, as
long as the remedies are constructive, turn a hostile customer into a happy one,
and, possibly, one who might even tout the company’s excellent customer
service at a later date.
Mullich offers the following tips on how to get the low-down on rivals372:
1. Understand that day-to-day online chatter can be misleading, but, over
time, An cruise line can find directional trends important to its business
and industry.
2. The deepest insights often come not from general sources, like
Facebook and Twitter, but from blogs and forums that are specific to an
industry.
3. Think broadly about the nature of one’s “competitors” — sentiment
analysis can help a business prepare for unexpected entries that might
be preparing to take a piece of its business. Keyword search teams can
help.
4. The information you can gain online about competitors is limited, and
often must be combined with your own internal data to bring
actionable insights.
There are, of course, limits to what competitive sentiment analysis can provide.
“The challenges you might address, using your company’s own customer,
product, and transactional data, are far more extensive than those you can tackle
via available competitor data,” says Seth Grimes, an analyst who runs the annual
Social Analysis Symposium.372 “For instance, you’re not going to have access to

301
 ANDREW W. PEARSON

your competitors’ contact-center notes and warranty claims, or to your

competitors’ customer profiles and transaction records. But with your own
company’s, you can create some very rich analyses,” he adds.372
For the above reasons, competitive analysis is usually just one piece of the vast
data mosaic.372 For example, one company that noticed a drop in sales of its
flagship product analyzed online chatter and found customers were talking
enthusiastically about a new product a competitor had just released.372 “When
the company analyzed its contact-center data, it found that returns correlated
to discontent about an attribute its own product lacked, but the new competing
product offered.”372 The company was quickly able to identify the problem and
by using a combination of competitive sentiment analysis, discovery from its own
internal data, it was able to tweak its own product to make it much more
competitive.372
As Grimes says, “Sentiment analysis can help you understand how the market
perceives you and your competitors’ products and services, but keep in mind
that sentiment is only an indicator, useful in measuring and projecting market
impact, not a substitute for strong human judgment.”372
There is a dark side to all this tracking as the case of IFA and Shopsense showed.
Health Insurer IFA Insurance teamed up with Shopsense, a grocery chain in
Midwest America, and bought their loyalty card customer data.381 The insurance
company discovered some intriguing patterns in the loyalty card data, such as
the correlation between condom sales and HIV-related claims, for example.381 It
also discovered such things as households that buy cashews and bananas
quarterly are the least likely to develop symptoms of Alzheimer’s.381 Although
this information did prove to be highly profitable for IFA, I believe it is a clear
violation of customer trust and privacy.
As Katherine Lemon explains in her article, How Can These Companies Leverage
the Customer Data Responsibly382, “Customer analytics are effective precisely
because firms do not violate customer trust. People believe that retail and other
organizations will use their data wisely to enhance their experiences, not to harm
them. Angry customers will certainly speak with their wallets if that trust is
violated.”382
Another concern for consumers is what Lemon calls “battered customer
syndrome.”382 She explains that, “Market analytics allow companies to identify
their best and worst customers and, consequently, to pay special attention to
those deemed to be the most valuable.”382
“Looked at another way, analytics enable firms to understand how poorly they
can treat individuals or groups of customers before those people stop doing
business with them. Unless you are in the top echelon of customers — those
with the highest lifetime value, say — you may pay higher prices, get fewer
special offers, or receive less service than other consumers,” Lemon adds.382

302
 THE A.I. CRUISE LINE

“Despite the fact that alienating 75% to 90% of customers may not be the best
idea in the long run, many retailers have adopted this ‘top tier’ approach to
managing customer relationships. And many customers seem to be willing to live
with it — perhaps with the unrealistic hope that they maybe reach the upper
echelon and reap the ensuing benefits.”382
“Little research has been done on the negative consequences of using marketing
approaches that discriminate against customer segments. Inevitably, however,
customers will become savvier about analytics. They may become less tolerant
and take their business (and information) elsewhere,” warns Lemon.382

358 Outing, S. (2007, September). Enabling the Social Company. Enthusiast Group.
359 Business.com. (2010, November 8). Top Tools to measure your social media success.
Retrieved from Business.com: http://www.business.com/info/social-media-monitoring-
tools
360 Honigman, B. (2012, November 29). 100 fascinating social media statistics and figures
from 2012. Retrieved from Thehuffingtonpost.com: http://huffingtonpost/brian-
honigman/100-fascinating-social-me_b_2185281.html
361 Urbanski, Al. At Caesars Digital Marketing Is No Crap Shoot. DMNews. February 01,
2013 www.dmnews.com/marketing-strategy/at-caesars-digital-marketing-is-no-crap-
shoot/article/277685/ (Accessed 20 November 2017).
362 Melville, P. &. (2009). Social Media Analytics : Channeling the Power of the
Blogosphere for Marketing Insight. Retrieved from citeseerx.ist.psu.edu:
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.157.3485&rep=rep1&type=
pdf
363 Rouse, M. (n.d.). Social Media Analytics. Retrieved from techtarget.com:
http://searchbusinessanalytics.techtarget.com/definition/social-media-analytics
364 https://cluetrain.com (Accessed 13 December 2019).
365 https://klear.com/ (Accessed 13 December 2019).
366 Perakakis, E., Mastorakis, G. and Kopanakis, I. (2019). Social Media Monitoring: An
Innovative Intelligent Approach. e-Business Intelligence Laboratory, Department of
Management Science and Technology, Hellenic Mediterranean University, Estavromenos,
71004 Heraklion, Greece. 20 May 2019. https://www.mdpi.com/2411-9660/3/2/24/pdf
(Accessed 1 December 2019).
367 Sterne, J. Artificial Intelligence for Marketing: Practical Applications; John Wiley &
Sons: Hoboken, NJ, USA, 2017. 9. Marinch
368 Marinchak, C.M.; Forrest, E.; Hoanca, B. Artificial Intelligence: Redefining Marketing
Management and the Customer Experience. Int. J. E-Entrep. Innov. (IJEEI) 2018, 8, 14–24.
369 Dimitrieska, S.; Stankovska, A.; Efremova, T. Artificial Intelligence and Marketing.
Entrepreneurship 2018, 6, 298–304.
370 Pitt, C.; Eriksson, T.; Dabirian, A.; Vella, J. Elementary, My Dear Watson: The Use of
Artificial Intelligence in Marketing Research: An Abstract. In Boundary Blurred: A Seamless
Customer Experience in Virtual and Real Spaces. AMSAC 2018. Developments in
Marketing Science: Proceedings of the Academy of Marketing Science; Krey, N., Rossi, P.,
Eds.; Springer: Berlin, Germany, 2018; p. 325.

303
 ANDREW W. PEARSON

371 Hicks, Z. (2013, August 28). Toyota Goes All-in With Social Media Monitoring.
Retrieved from CIO.com: http://www.cio.com/article/2383143/social-media/toyota-
goes-all-in-with-social-media-monitoring.html
372 Mullich, J. (2012, December 10). Opposition Research: Sentiment Analysis as a
Competitive Marketing Tool. Retrieved from Wellesley Information Services: http://data-
informed.com/opposition-research-sentiment-analysis-as-a-competitive-marketing-
tool/
373 Altexsoft. (2018). Sentiment Analysis: Types, Tools, and Use Cases. September 21,
2018. https://www.altexsoft.com/blog/business/sentiment-analysis-types-tools-and-
use-cases/ (Accessed 15 April 2019).
374 Dreamgrow.com. https://www.dreamgrow.com/69-free-social-media-monitoring-
tools/ (Accessed 22 November 2017).
375 Hamill, J. and Stevenson, A. 2010. Step 3: Key Performance Indicators (Post 1).
Available at: www.energise2-0.com/2010/06/27/step-3-key-performance- indicators-
post-1/ [accessed: 12 February 2011].
376 Gulbransen, Scott. January 22, 2014. Taking Back The Social-Media Command Center,
Scott Gulbransen. Forbes.
http://www.forbes.com/sites/onmarketing/2014/01/22/taking-back-the-social-media-
command-center/#3c283a5d6513
377 IBM. (2013, February). Social Media Analytics: Making Customer Insights Actionable.
Retrieved from IBM.com: http://www-01.ibm.com/common/ssi/cgi-
bin/ssialias?infotype=SA&subtype=WH&htmlfid=YTW03168USEN
378 Gainsbury, Sally M., Delfabbro, Paul, King, Daniel L., and Hing, Nerilee. March 2016.
An Exploratory Study of Gambling Operators’ Use of Social Media and the Latent
Messages Conveyed.
379 Adstage. Q2 2019 Paid Search and Paid Social Benchmark Report.
https://www.adstage.io/resources/adstage-benchmark-reports/ (Accessed 15 October
2019).
380 Astage Q3 2019 Paid Search and Paid Social Benchmark Report.
https://www.adstage.io/resources/adstage-benchmark-reports/ (Accessed 15 October
2019).
381 Davenport, T. (2006). Competing on Analytics. Harvard Business Review.
382 Lemon, K. (2007, May). How Can These Companies Leverage the Customer Data
Responsibly. Retrieved from http://blog.hansacequity.com:
http://blog.hansacequity.com/Portals/11224/docs/article%20on%20Analytics.pdf

304
 CHAPTER SIX: THE SMART CRUISE
LINE
“Smart homes and other connected products won't just be
aimed at home life. They'll also have a major impact on
business. And just like any company that blissfully ignored the
Internet at the turn of the century, the ones that dismiss the
Internet of Things risk getting left behind.”
~Jared Newman
Fast Company

Overview
Not only is AI old, but it is also a difficult technology to implement. One of the
first things companies that want to implement AI need to do is to get its data
house in order. In its Conquer the AI Dilemma by Unifying Data Science and
Engineering383, Databricks says that only “1 in 3 AI projects are successful and it
takes more than 6 months to go from concept to production, with a significant
portion of them never making it to production — creating an AI dilemma for
organizations.”
Databricks believes that data-related challenges are hindering 96% of
organizations from achieving AI.383 Nearly all of the respondents (96%) cited
multiple data-related challenges when moving projects to production. 383
“According to the survey, 90% of the respondents believe that unified analytics
— the approach of unifying data processing with ML frameworks and facilitating
data science and engineering collaboration across the ML lifecycle, will conquer
the AI dilemma.”383
Databricks argues that, Unified Analytics make AI much more achievable for
organizations.383 “Unified Analytics makes it easier for data engineers to build
data pipelines across siloed systems and prepare labeled datasets for model
building while enabling data scientists to explore and visualize data and build
models collaboratively.”383 A unified analytics platform can “unify data science
and engineering across the ML lifecycle from data preparation to
experimentation and deployment of ML applications — enabling companies to
accelerate innovation with AI,” concludes Databricks.383
The rest of this chapter will focus on how to build the backbone of an IT system
that will incorporate a structure that can help a cruise line utilize AI throughout

305
 ANDREW W. PEARSON

its operation. General sections on data governance, IoT, deep learning

frameworks, amongst others, will lay out the most common questions that a
marketer should ask about an EDW, cloud systems, storage, CPUs, and an AI, ML
and deep learning world. The rest of this chapter will detail specific business
areas that can be improved with these technologies.
The term “Smart” has become synonymous with a vision of technology
integrating with multiple information and communication platforms, including
Internet of Things (IoT) solutions to enhance quality, performance and
interactivity of services, as well as to reduce costs and resource consumption and
to improve contact between different parties, including between a citizen and
his or her government.
In my estimation, The A.I. Cruise Line has answers to the following questions:
• How can a cruise line save on it resource use?
• How can a cruise line improve the management of patron movements?
• How can a cruise line best utilize its transportation fleet?
• How can a cruise line reduce its labor needs?
• How can a cruise line cut down on waste?
• How can a cruise line ensure its security?
These are all important questions that need to be asked by today’s cruise line
executives. Usually, cruise lines are part of massive corporations, with
international operations, and the IoT could be helpful for them to become more
predictive. Sensors have become so small and so cheap they can be put almost
anywhere. IoT sensors can be used for smart parking, smart lighting, as well as
part of a mini smart grid. They can also be used for silo stock calculation —
measuring the emptiness level and weight of goods, as well as waste
management, tracking movements on the cruise line floor, and perimeter access
control.
For a cruise line’s logistics department, IoT aids quality of shipment conditions,
item location, storage incompatibility detection, and fleet tracking. IoT sensors
can even be installed to ensure a buildings structural health, as well as part of a
swimming pool remote management system.
As Figure 22 shows, analytics can be used to boost service quality, reduce
operating costs, as well as increase ROI. As the IBM Thought Leadership White
Paper Descriptive, predictive, prescriptive: Transforming asset and facilities
management with analytics384 states, “As facilities and assets become more IT-
like — instrumented, intelligent and interconnected — the convergence of
physical and digital infrastructures makes their management increasingly
complex. And in a physical world outfitted with millions of networked sensors,
vast amounts of facilities- and asset-generated data make extracting meaning
increasingly difficult.“384

306
 THE A.I. CRUISE LINE

Figure 22: Understanding analytics, definitions, applications

Source: IBM384

When it comes to capital planning, real estate lease management, operations,

facilities maintenance and energy consumption, IRs must tame the three Vs of
big data — volume, variety, and velocity — and analytics are the best way to do
that.
Utilizing the right analytics in the right place can yield impressive ROI results. As
IBM states384:
“From basic to advanced capabilities, analytics can yield
dramatic results. One study found that an organization that
uses basic automation to expand its reporting capabilities can
improve its return on investment (ROI) by 188 percent. But
adding additional capabilities such as data management,
metadata to ensure uniform data interpretation, and the
ability to gather and analyze data from outside the
organization, can boost ROI to as high as 1,209 percent.”
IBM’s specific examples include:
• “Effective facilities and asset management uses data analytics to
proactively manage facilities and maintain equipment, optimize
utilization, prevent breakdowns, lower occupancy and operational
costs, and extend asset life.”384
• “Utilizing analytics to monitor energy-intensive equipment across the
facilities portfolio, identify operating anomalies in real time, and

307
 ANDREW W. PEARSON

generate corrective work orders can dramatically reduce energy

consumption.”384
• “To help mitigate risks of failure in facilities and assets, analytics can
detect even minor anomalies and failure patterns. Identifying issues
early helps organizations deploy limited maintenance resources more
cost-effectively, maximize equipment uptime and improve customer
service levels.”384

Data Governance
Today, Talend believes that, “Data governance is not only about control and data
protection; it is also about enablement and crowdsourcing insights. Data
governance is a requirement in today’s fast-moving and highly competitive
enterprise environment.”385 Ultimately, “Now that organizations have the
opportunity to capture massive amounts of diverse internal and external data,
they need the discipline to maximize that data’s value, manage its risks, and
reduce the cost of its management,” claims Talend.385
Data governance is not optional in today’s highly complex and fast-moving IT
environment.385 An effective data governance strategy provides so many crucial
benefits to an organization, including385:
• A common understanding of data: “Data governance offers a consistent
view of, and common terminology for, data, while individual business
units retain appropriate flexibility.” 385
• Improved data quality of data.
• A data map.
• A 360-degree view of each customer and other business entities.
• Consistent compliance with government regulations.
• Improved data management because a human dimension is brought
into a highly automated, data-driven world.
• Easy accessibility.
To find the right data governance approach for your organization, Talend
recommends brands look for “open source, scalable tools that are easy to
integrate with the organization’s existing environment.”385 Additionally, a cloud-
based platform lets brands “quickly plug into robust capabilities that are cost-
efficient and easy to use.”385 “Cloud-based solutions also avoid the overhead
required for on-premises servers,” notes Talend.385 When comparing and
selecting data governance tools, brands needs to focus on choosing ones that
will help them realize the business benefits laid out in their data governance
strategy.385 Any chosen tool should help in the following ways 385:
• Capture and understand data through discovery, profiling,
benchmarking and capabilities.385 For example, the right tools can

308
 THE A.I. CRUISE LINE

automatically detect a piece of personal data, like a national ID or social

security number, in a new data set and then trigger an alert.
• Improve the quality of a brand’s data with validation, data cleansing,
and data enrichment.
• Manage a brand’s data with metadata-driven ETL and data integration
applications so data pipelines can be tracked and traced with end-to-
end, forward-looking and backward-looking data lineage.
• Control a brand’s data with tools that actively review and monitor it. 385
• Empower the people who know the data best, so they can contribute
to the data stewardship tasks with self-service tools.385
Modern data governance is about both minimizing data risks and maximizing
data usage.385 There is a need for a more agile, bottom-up approach, which
“starts with the raw data, links it to its business context so that it becomes
meaningful, takes control of its data quality and security, and thoroughly
organizes it for massive consumption.”385 In addition, due to headline-grabbing
data scandals and data leaks, government are enacting a proliferation of new
regulations and laws that put higher stakes on data protection. 385
Talend believes that, “New data platforms empower this new discipline, which
leverages smart technologies like pattern recognition, data cataloging, data
lineage, and machine learning to organize data at scale and turn data governance
into a team sport by enabling organization-wide collaboration on data
ownership, curation, remediation, and reuse.” 385
With data storage prices plummeting, data is becoming less commoditized.385
large data repositories such as data lakes are creating vast reservoirs of known
and unknown datasets.385 Although it might take seconds to ingest data into a
modern EDW or data lake, it could take weeks for this data to be made available
to a business user.385 At the same time, business users might not even be aware
that the data they need is even available for use. 385 Humans are inventive
creatures and they often employ data work-arounds, which can create additional
governance headaches.385 When business users add their own rules atop newly
created data sources, multiple versions of “the truth” result. This can lead to data
governance nightmares.385
The challenge “is to overcome these obstacles by bringing clarity, transparency,
and accessibility to your data assets.”385 Wherever this data resides, proper data
screening must be established so businesses have a holistic view of the data
sources and data streams coming into and out of their organization.385
In the past, data experts might have manually processed the data using
traditional data profiling tools.385 However, this approach no longer works.385
“The digital era’s data sprawl requires a more automatic and systematic
approach,” says Talend.385

309
 ANDREW W. PEARSON

Modern data cataloging tools can help schedule the data discovery processes
that crawls an EDW or a data lake and intelligently inspect the underlying data,
so that it can be understood, documented, and actioned, if necessary.385 Today’s
data catalogs “can automatically draw the links between datasets and connect
them to a business glossary.”385 Talend argues that, “this allows an organization
to automate the data inventory and leverage smart semantics for auto-profiling,
relationships discovery and classification thanks to an integrated semantic
flow.”385 The benefits are twofold; data owners and providers get an overview
of their data and can take actions; data consumers get visibility into the data
before consuming it.385
Data profiling is the process of discovering in-depth and granular details about a
dataset. It helps in accurately assessing a company’s multiple data sources based
on the six dimensions of data quality — accuracy, completeness, consistency,
timeliness, uniqueness, and validity.385 It will help a brand identify if and how its
data is inaccurate, inconsistent, and, possibly, incomplete.385
Often, the people who know the data best are not the data experts. 385 Sales
admins, sales representatives, customer service reps, and field marketing
managers know the data quality issues probably better than their company’s
central IT team.385 Not only do they know the data best, but they are also the
ones who most keenly feel the pain of data quality issues because it directly
impacts upon their day-to-day job.385
Of course, these people can’t become data quality experts so they must be
provided with smart tools that can hide the technical complexity of data
profiling. Many vendors provide data preparation tools that have “powerful yet
simple built-in profiling capabilities to explore data sets and assess their quality
with the help of indicators, trends, and patterns.” 385 “While automatic data
profiling through both a data catalog and self-service profiling addresses the case
for bottom-up data governance, a top-down approach might require a deeper
look into the data,” says Talend.385
With products like Talend Data Quality, users “would start by connecting to data
sources to analyze their structure (catalogs, schemas, and tables), and store the
description of their metadata in its metadata repository.” 385 Users would then
“define available data quality analyses including database, content analysis,
column analysis, table analysis, redundancy analysis, and correlation analysis,”
says Talend.385 “These analyses will carry out data profiling processes that will
define the content, structure, and quality of highly complex data structures,”
adds Talend.385
A “trust index” can be created out of all this data discovery and it can be
calculated, reported, and tracked on a regular and automated basis. 385 Trigger
alerts can be set when index moves beyond a certain comfortable threshold. 385
According to Talend, “Data quality is the process of conditioning data to meet

310
 THE A.I. CRUISE LINE

the specific needs of business users.” 385 However, data quality is not a
standalone operation or problem. 385 “To make it successful and deliver trusted
data, you need to operate data quality operations upfront and natively from the
data sources, along with the data lifecycle to ensure that any data operator or
user or app could consume trusted data at the end,” argues Talend. 385
“Successful data governance frameworks require setting accountabilities and
then delegating that authority appropriately,” argues Talend 385 For example,
Talend says, “a data protection officer in a central organization might want to
delegate tasks to data stewards or business users in the operations: a sales
engineer might be best positioned to ensure that contact data for his or her
accounts are accurate and kept up-to-date. A campaign manager is the one that
should ensure that a consent mechanism has been put in place and captured
within its marketing database.”385 To support this kind of delegation,
organizations need to provide workflow based, self-served apps to different
departments, recommends Talend.385 This provides additional autonomy
without putting the data at risk.385
“The cloud drastically extends the boundaries of data. Lines of business use their
own applications, and products, people, and assets create their own data
pipelines through the web and the Internet of Things. Data can also be
exchanged seamlessly between business partners and data providers,” says
Talend.385
“Data preparation is not just a separate discipline to make lines of business more
autonomous with data; it’s a core element for data quality and integration,”
contends Talend.385 Not only does it unlock people’s data productivity, but it also
captures the actions taken on that data, which can help make the data more
trustable.385 In addition to improving personal productivity, the true value of
these collaborative and self-service applications is to drive collaboration
between business and IT, however that’s not always an easy thing to do.385
Once the incoming data assets are identified, documented and trusted, it is time
to organize them for massive consumption by an extended network of data users
within an organization.385 “This starts by establishing a single point of trust; that
is to say, collecting all the data sets together in a single control point that will be
the cornerstone of your data governance framework,” explains Talend.385
Datasets then need to be identified; roles and responsibilities have to be
assigned directly into a single point of control.385
“It is one of the advantages of data cataloging: regrouping all the trusted data in
one place and giving access to members so that everybody can immediately use
it, protect it, curate it and allow a wide range of people and apps to take
advantage of it,” notes Talend.385 “The benefit of centralizing trusted data into a
shareable environment is that it will save time and resources of your
organization once operationalized,” they add.385

311
 ANDREW W. PEARSON

“Within a data catalog, a business glossary is used to define collections of terms

and to link them to categories and sub-categories. Building a business glossary
can be as simple as dragging in an existing well-documented data model,
importing the terms and definitions from other sources (e.g., CSV, Microsoft
Excel),” says Talend.385 Once published, the glossary can be accessed company-
wide anyone who has proper authorizations.385
Talend believes that, “As you are about to deliver access to your catalog to
others, your dataset will become a living artifact, as you will enable authorized
people to edit, validate, or enrich the data directly into data sets.”385 “Doing it
automatically through a data catalog will allow you to save lots of time and
resources,” they contend.385
Talend claims that data lineage functionality gives users the ability to track and
trace their data flows from source to final destination.385 Data lineage can
dramatically accelerate the speed to resolution of problematic data by helping
users spot the specific problem at the right place and ensure that the data is
always accurate.385 Moreover, if new datasets come into an EDW and/or a data
lake, data lineage rapidly helps identify these new sources. 385 Errors can quickly
be uncovered and accountability understood. A data chain is both forward and
backward-looking, upstream or downstream impact is easily seen and acted
upon.
Once the data categories have been defined, a more accurate picture of the data
environment sources can be created.385 “It will also enable you to define better
data owners: who is responsible for this particular data domain? Who is
responsible for viewing, accessing, editing and curating the data sets?” explains
Talend.385
At this step, using a RACI Model — a model derived from the four key
responsibilities most typically used: Responsible, Accountable, Consulted, and
Informed — will help users save time defining and assigning roles and
responsibilities between stakeholders.87
“The next step is to define data owners who are ultimately accountable for one
or more data categories and subcategories,” says Talend.385 “These data owners
will be responsible for day-to-day operations regarding the data or appoint data
stewards those operational data-centric tasks. They will identify critical datasets
and critical data elements (CDEs) as well as establish standards for data
collection, data use, and data masking.”385 “A data catalog may also catalog
owners and stewards for data categories and sub-categories and assign their
related roles and workflows,” adds Talend.385
For example, a data cataloger “may catalog the data owners for ‘customer’ as
well as ‘customer identity’, ‘customer billing’, ‘customer contact’ and ‘customer
ship-to information’.”385

312
 THE A.I. CRUISE LINE

The RACI Model is a good example of a responsibility assignment matrix that is

both easy to understand and use.87 “It’s particularly useful if your data
governance will involve different departments and divisions in your
organization,” says Talend.385
According to Wikipedia, data curation “is the organization and integration of
data collected from various sources. It includes annotation, publication, and
presentation of data to make sure it’s valid over time.” 386 This will be enabled
once you put in place an explicit RACI Model that clearly describes who can
define, edit, validate, and enrich data in the systems.385
“A data governance project is not just intended to let trusted data be accessible
to all,” claims Talend.385 “It’s also about promoting data custodians’
accountability to the rest of the organization so that they can enrich and curate
trusted data and produce valuable, accurate insights out of the data pipelines,”
they add.385 “In many cases, data owners realize that they should not manage
everything in their data domain, and thus need act as orchestrators rather than
doers,” notes Talend.385
“The data governance team may also delegate responsibilities for data
protection,” adds Talend.385 Data masking is a prime example for delegation. 385
“In a data lake, for example, IT specialists might not be the ones responsible for
data masking and might even not have the authorization privileges to process
the data before it has been masked,” claims Talend.385 Data protection tasks can
be delegated to people who might not be technical experts with deep expertise
in the data masking discipline.385
“This is why it is important to empower a large audience to mask the data on
their own so that once they identify specific scenarios where sensitive data may
be exposed, they can proactively act on it automatically with a user-friendly
tool,” says Talend.385 For example, Talend offers the case of a campaign manager
who prepares an event with a business partner that doesn’t have explicit consent
to see the personal data of a customer because of a lack of third party privacy
consent.385 Thankfully, the campaign manager can utilize data prep tools that
can mask the data directly on the data so that the data can be easily shared
without violating data privacy rules.385
Once the data is accessible in a single point of access and reconciled properly, “it
is time to extract all its value by delivering at scale to a wide audience of
authorized humans and machines,” says Talend.385 Technologies like
automation, data integration and machine learning can help enormously.385
“Advanced analytics and machine learning help democratize data governance
and data management because they make things much simpler,” argues
Talend.385 “They improve developers’ productivity and empower non-data
experts to work with data as well by suggesting next best actions, guiding users
through their data journey,” say Talend.385

313
 ANDREW W. PEARSON

“Machine learning also allows the capture of knowledge from business users and
data professionals,” says Talend.385 One typical use case is data error resolution
and matching.385 Self-service tools can be used to deduplicate records on a data
sample and then machine learning can be applied to a whole data set in a fully
automated process, which turns low value and time-consuming tasks into an
automated process that can be scaled up to handle millions of records. 385
Data masking allows a company to selectively share production quality data
across their organization for development, analysis and more, without ever
disclosing any Personally Identifiable Information (PII) to people not authorized
to see it.385
Failing to establish strict data privacy controls can leave a company exposed to
financial risk, negative reputation, and stiff data privacy regulatory penalties. 385
To deal with this growing threat, businesses need to find ways to automatically
spot sensitive datasets.385 Data cataloging technologies can help with this.385
“A data catalog is the typical starting point for automating the personal data
identification process,” says Talend.385 Once data elements have been defined
with a PII, data sets that relate to them can automatically be spotted and
masked, if necessary.385 If personal data is not necessary for testing or analytics,
why risk exposing it?385
In the past, disciplines like data masking were sparingly used, but with the
explosion of data privacy scandals and the proliferation of regulations, a much
more aggressive approach to data masking is needed.385 Only then can
businesses share production-quality data across their organizations for analysis
and business intelligence, without exposing personally identifiable
information.385
“Many data governance approaches fail because they cannot be applied in a
systematic way,” claims Talend.385 Modern data governance controls “need to
be embedded into the data chain, so that it can be operationalized and cannot
be bypassed.”385 It needs to become part of the process.
Data governance can help data engineers orchestrate and automate all of a
company’s data pipelines, whether they are physical EDWs or cloud-based ones,
or even data that surfaces through a company app. 385 “It will act as an
orchestrator to operationalize and automate any jobs or flows so that you keep
on structuring and cleaning your data along the data lifecycle, all the while
putting stewards at work for validation, users for curations or business users for
data preparation,” says Talend.385
A data catalog makes “data more meaningful for data consumers, because of its
ability to profile, sample and categorize the data, document the data
relationships, and crowdsource comments, tags, likes and annotations.” 385 “All
this metadata is then easy to consume through full text or faceted search, or

314
 THE A.I. CRUISE LINE

through visualization of data flows,” explains Talend.385 Data catalogs make it

“possible to locate, use, and access trusted data faster by searching and verifying
data’s validity before sharing with peers.”385

Capacity and Demand Management

In their Capacity management for hospitality and tourism: A review of current
approaches54, Pullman and Rodgers acknowledge that many leisure enterprises,
such as casinos and massive integrated resorts, require substantial capital
investment and expenditure. Pullman and Rodgers contend that, “While capacity
management has generally fallen within the domain of operations management,
and demand management within the domain of marketing, intersecting
methodologies such as yield and revenue management, the partitioning of
visitors (by length of stay), and/or the adjustment of visitor participation levels
rely on expertise from both of these functional areas to work effectively.” 54
Although capacity management has been referred to as ‘‘demand
management’’387 by Crandall and Markland and ‘‘managing capacity and
demand’’ by Fitzsimmons and Fitzsimmons, capacity and demand management
are, in fact, distinct concepts. Taylor believes that demand management
attempts to influence when and how many visitors attend or use a service388,
whereas Klassen and Rohleder contend that capacity management ensures that
sufficient capacity exists to meet this demand.389
Pullman and Rodgers argue that, capacity strategy is a key operational function
for all leisure-related enterprises.390 391 51 “The extent to which capacity satisfies
demand has an impact on visitor experience, employee satisfaction, profitability,
and long-term sustainability of both the resources and the enterprise itself,” they
contend.54
In the leisure industry, excess capacity not only underutilizes the workforce and
other physical resources, but can also lead to increased waste and demand
stimulating and profit reducing price reductions.391 “Inadequate capacity, on the
other hand, can impair the visitor experience through degradation of facilities,
overuse of natural resources, crowding, and increased waiting time, warn
Pullman and Rodgers.54 It can also open the door for competitors to enter the
market as well as overtax the workforce, leading to employee burnout and
increased turnover.54 “Fortunately, firms can borrow a number of ideas from
other industries to facilitate better matching of supply to demand in their
particular enterprise,” advise Pullman and Rodgers.54
Capacity “can be separated into two distinct perspectives, the strategic or long-
term perspective and the tactical or short-term perspective.”54 Strategic capacity
decisions are made during the planning stages for a project as managers consider
macro-level responses to existing and potential future demand. 392 393 391 For

315
 ANDREW W. PEARSON

Pullman and Rodgers, “Such decisions may include the number of hotel rooms
an area can or should support; the land space, energy, and water required for a
project; the available labour and skills; the anticipated overall size of the
enterprise in terms of parking, seating, and production requirements (in food
services); and/or the carrying capacity issues for natural resources.” 54
Later in the planning process, the capacity focus shifts to the set of short-term
actions taken to fulfill the planned strategies, often called the “tactics.” 393
“Examples of these decisions include the determination of the number of
employees needed to meet peak demand during a summer lunch period and the
best mix of table configurations to accommodate dinner demand given different
party sizes,” say Pullman and Rodgers.54 As per Figure 23, “Long versus short
term decisions are distinctly different, and the methodologies for addressing
these problems often involve alternate approaches.” 54

Figure 23: Capacity decisions

Source: Pullman and Rogers, Capacity management for hospitality and tourism:
A review of current approaches54

According to Porter, “In the strategic capacity planning phase, enterprises make
decisions to ensure achievement of their desired objectives based on an
assessment of their current situation or position, capabilities, shortcomings, and
overall competitive position; the alternatives and risks involved; and the
timing.”392

316
 THE A.I. CRUISE LINE

For Pullman and Rodgers, “Strategic capacity decisions include capacity size and
expansion, carrying capacity or optimal use of the resource, and capacity
flexibility. Capacity size refers to the maximum physical size of a facility or the
optimal size of a workforce.” 54 Carrying capacity is the desired or optimal level
of utilization of a given resource; beyond that capacity, both Lawson and Baud-
Bovy394 as well as Mathieson and Wall395 argue that negative impacts exceed
levels specified by evaluative standards, such as those for physical deterioration
or minimum quality of recreational experience. “Capacity flexibility represents
the ability to respond to fluctuating demand using additional labour or
adjustments to physical space.”54 For Schroeder, “These strategies address the
amount and timing of capacity changes and types of facilities needed for the long
run.”393 “For physical facilities, capacity strategy is part of the total operations
strategy; it is not merely a series of incremental capital-budgeting decisions,”
argue Pullman and Rodgers.54 “Both physical and labour capacity strategies
include proactive planning for future growth, reactive response to existing
demand, or a mix of these two strategies,” claim Davis and Heineke. 391
“Short-term capacity management comprises the set of actions taken to fulfill
the firm’s strategy — specifically, the methods of implementation or the
requirements for the strategic plan to take effect,” say Pullman and Rodgers54;
for example, diners or visitors served per hour, rooms cleaned per shift, or skiers
per day.54 “Managers and planners are concerned with meeting demand in real
time by means of such things as labour scheduling, flexible seating or
partitioning, or opening up more seating capacity at lower price points as an
event or departure date approaches,” explains Pullman and Rodgers. 54
Other decisions can address such things as “the design of the service process,
including the service interaction time (the average amount of time an employee
spends with a visitor); the level of visitor participation or self-service; the
capacity cushion (extra space or labour added as a contingency); and queue
reservations or partitioning.”54 Disney’s Fast-Pass program or Universal’s VIP
pass for special fast lines are examples of reservations systems put in place to
ease traffic and customer headaches.54 “The appropriate focus for these capacity
decisions depends on the overall size and type of leisure enterprise,” argue
Pullman and Rodgers.54
It is useful to view leisure-related enterprises in terms of the service process
matrix, which was originally developed by Schmenner 396 for the field of
operations management. As per Table 12, this framework classifies enterprises
“according to their degree of labour intensity versus degree of
interaction/customization.”54 “Labour intensity is defined as the ratio of labour
cost incurred to the value of the facilities and equipment.”54 In general, “highly
labour-intensive businesses require relatively little capital expenditure for
facilities and equipment, while businesses with low labour intensity have low
levels of labour cost relative to costs for facilities and equipment.” 54 “The degree

317
 ANDREW W. PEARSON

of interaction and customization reflects the level of service customization for

the visitor and the level of visitor interaction with the service process. When both
of these levels are high, visitors actively participate in the service process and
the business works to satisfy the visitors’ particular preferences,” say Pullman
and Rodgers.54

Degree of labour Degree of interaction and customization

intensity Low High
Service Factory Service Shop
Airlines
Luxury hotel
Standard hotels
Low Luxury restaurants
Resorts
Luxury cruise ships
Cruise ships
Spas
Recreation
Theme parks
Mass Service Professional Services
High Retailing Tour guides
Cafeterias Instructors
Fast food Concierges
Table 12: Application of the service process matrix 396 to the field of hospitality
and tourism
Source: Pullman and Rogers, Capacity management for hospitality and tourism:
A review of current approaches54

“For enterprises with low ratios of labour to capital intensity — that is, Service
Factory and Service Shop processes — physical capacity decisions are
paramount. Here, capacity cannot be augmented easily, so it is essential to utilize
each capacity unit effectively (each airline seat, hotel room, restaurant or gaming
table seat, amusement park ride seat, etc.),” contend Pullman and Rodgers.54
“Strategic physical capacity decisions encompass the long-term planning of
construction or renovation of buildings, planes, or cruise ships, or the purchasing
of new equipment, rides, vehicles, etc. Short-term capacity management, in
contrast, includes approaches that enable firms to react to short-term
fluctuations in demand,” argue Pullman and Rodgers.54 “The latter is typically
achieved through scheduling, yield or revenue management, and the addition or
withdrawal of support services. Mass Service and Professional Service processes,
on the other hand, have high degrees of labour intensity,” explain Pullman and
Rodgers.54 “Strategic capacity management for these enterprises focuses more
on the management of labour capacity, including the hiring and selection of
employees with the appropriate skill mix,” note Pullman and Rodgers. 54 “Short-
term capacity management focuses on the implementation of scheduling and

318
 THE A.I. CRUISE LINE

other incremental capacity methods that allow the enterprise to react to short-
term fluctuations in demand with greater flexibility,” conclude Pullman and
Rodgers.54
“Capacity measures serve as the foundation of all analytical approaches for all
capacity decisions,” explain Pullman and Rodgers.54 “For example, a capacity
measure could provide the planning objective (e.g., an 80% employee utilization
goal). Or the measures could show the relationship of capacity to the overall
performance of a system (e.g., at 80% and 90% maximum capacity, a park
experienced 89% and 75% customer satisfaction ratings, respectively),” add
Pullman and Rodgers.54 But as pointed out by Wall, capacities are difficult to
measure because of definitional problems.397 “The appropriate design and
outcome measures depend on whose perspective is emphasized,” note Pullman
and Rodgers.54 Table 13 provides some examples of operational indicators that
are particularly relevant to managers and visitors.

Visitors’ perspective Managers’ perspective

Availability Revenue or profits
Right time
Right price Number of visitors
Right type
Crowding and space
Too many people/too little space Utilization
Too few people/too much space
Intimacy/privacy Percentage of cost of depletion or wear
Interaction/sociability Excess capacity or waste
Perceived service time Labour costs
Variable and fixed physical resources
Perceived wait time Actual service time
Actual wait time
Table 13: Examples of capacity measures relevant to visitors’ versus managers’
perspective
Source: Pullman and Rogers, Capacity management for hospitality and tourism:
A review of current approaches54

“In most for-profit environments, management is concerned with maximizing

revenue through the number of paying visitors using the enterprise or revenues
generated per visitor and the firm’s capacity utilization of resources,” explain
Pullman and Rodgers.54 “Visitor objectives (including less crowding and shorter
perceived waiting times) and their other specific measures of relevance differ
from those of the enterprise management,” add Pullman and Rodgers. 54
For Wall, “In hospitality and tourism, both biophysical (the quality of the

319
 ANDREW W. PEARSON

environment) and behavioural components are recognized.”397 An increase in

physical capacity utilization can lead to queuing and a general reduction in the
perceived quality of service warn Davis and Heineke.391 This can have profoundly
negative effects on customer satisfaction, obviously.
In the mainstream literature, the critical levels that are needed to support
certain facilities are often referred to as capacity thresholds. 398 Muller provides
the example of the high occupancy of an airplane or a quick-service restaurant,
which can lead to untidiness because the service crew simply doesn’t have time
to maintain cleanliness; from the customer’s perspective this can be a big
negative.399 A nearly empty cruise liner, on the other hand, does not provide the
expected social ambience and affects the employees’ interactions with
customers argues Kandampully.400
“Clearly, managers must monitor visitors’ perceptions and preferences and try
to best meet both management’s and their customers’ expectations,” advise
Pullman and Rodgers.54 For example, Kilby and Fox explain in Casino Operations
Management, “an Australian casino improved the level of patrons’ comfort by
decreasing the playing spots offered on each gaming table, which subsequently
increased the blackjack hold.401 However, as Muller illustrates, a reduction in the
number of customers can be compensated for by higher revenues only if the
demand is price inelastic, which is not always the case.399
“Different capacity indicators linking the ability of an enterprise to control the
environment with the overall customer experience can be used. For example, in
a restaurant setting, capacity-related indicators include average customer
waiting time and average duration of each service,” explain Pullman and
Rodgers.54 These indicators have been shown to have “a causal relationship with
the number of incidents/complaints, the general ratings of the service, and the
food quality.”402
“For a recreational site, capacity indicators can include the following: the
satisfaction of the visiting public, the intention to return, the percentage of
people feeling overcrowded, and the number of complaints,” say Simón et al. 403
Several researchers have shown that the links between capacity decisions and
other performance measures such as profit, quality (overall visitor satisfaction),
and market share are complex and nonlinear, including work by Easton and
Pullman404 and Sill.405
As previously mentioned, “strategic capacity planning of physical resources is of
primary concern for Service Factory and Service Shop processes (hotels, resorts,
cruise ships, recreational sites and theme parks).” 54 In practice, this planning is
part of the detailed design that includes capacity estimation, an integral part of
a feasibility study406 and the main indicator of the price of a capital asset. 407
Strategic physical capacity decisions include size, configuration, layout, and built-
in flexibility. These decisions made during the planning stage have long-term

320
 THE A.I. CRUISE LINE

implications for the overall resource productivity of a property. 408

The capacity-related criteria for the ‘‘goodness’’ of a design includes the grossing
factor53, which is calculated as the proportion of income generating areas
relative to the total floor space. According to Pullman and Rodgers, “The grossing
factor is reduced by the need to provide technical areas (business centres, sport
and recreational areas, etc.) and functional features (architectural elements such
as atriums, balconies, and grand staircases), as well as by the need to meet
legislative and environmental requirements such as land zoning and usage
restrictions, building codes, fire and maritime safety requirements, and disability
access regulations.”54
“Codes for food-service premises, for example, stipulate the allocation of
adequate space for each food production area and regulate the physical layout
of the facility to ensure the ease of cleaning/sanitation (minimum gaps between
equipment units and walls, minimum number of hand-washing sinks and
maximum walking distance to reach them, etc.), explain Pullman and Rodgers. 54
These requirements for land-based409 410 and cruise-ship411 kitchens lead to an
increase in the size of the production areas, which are non-revenue-
generating.54 “By means of the grossing factor, various areas of a facility can be
assessed for the contribution margin of the different service products they
support,” explain Pullman and Rodgers.54 For example, Kimes and McGuire show
that in the Raffles City Convention Centre in Singapore, the contribution margin
is 30–35% for food services, 85–95% for room rentals, and 50–95% for audio–
visual equipment.412
“In addition to size, the actual physical capacity depends on the configuration
and layout of a facility, since the combination of these affects the flow of
products and customers,” notes Pullman and Rodgers. 54 For example, in
restaurant planning, table types and locations (banquettes, booths, or anchored
or unanchored tables) affect the customer experience and, hence, the average
spending per minute.413 414 Table 14 provides other examples of other design
solutions for different industries and sectors.
The capacity-related criteria for the ‘‘goodness’’ of a design include the grossing
factor53, “which is calculated as the proportion of income generating areas
relative to the total floor space.”54 “The grossing factor is reduced by the need
to provide technical areas (business centres, sport and recreational areas, etc.)
and functional features (architectural elements such as atriums, balconies, and
grand staircases), as well as by the need to meet legislative and environmental
requirements such as land zoning and usage restrictions, building codes, fire and
maritime safety requirements, and disability access regulations,” note Pullman
and Rodgers.54

321
 ANDREW W. PEARSON

INDUSTRY LAYOUT SOLUTION OUTCOMES

The hub is a central facility offering
restaurants, shopping arcades, entertainment,
Hub and spoke system
conference rooms, and other amenities, while
the spokes are chosen destination areas.415
Airlines,
The space is designed for people to move from
theme parks,
Point to point one destination area to the next without
ski resorts
moving through a central location.416
Customers sign up for virtual queue, go to
Virtual queue other destination areas and return at
appointed time.417
The decrease in the net bedroom size coupled
Increased bathroom with an addition of a walk-in shower and
Hotels
size dressing area resulted in a 13% improvement
of the design efficiency ratio.53
A steady, controlled process flow reduced
complexity and confusion in a kitchen 418
Better distribution of serving capacity
Process flow principle between cold and hot sections in restaurant
buffets resulted in faster table turns, shorter
lines, better food temperature control, and
Restaurants decreased construction costs.405
Customers in booths spent more on average
Table type and
than those in banquettes; those in poor table
location
locations spent more.413
Creating a busy feel (small dining spaces)
Reduce table spacing
increased customer satisfaction.
Positioning of the upper category of cabins at
Horizontal class the lower desk to minimize rolling, which is
sandwich amplified at the upper deck as well as at the
Cruise ships bow and the aft.419
Vertical class barrier with the provision of a
‘‘Cocooned isolation’’ view, a private balcony, and no need to change
floors to reach the nearest pool.419

Table 14: Examples of layout applications for physical resources54

Source: Pullman and Rogers, Capacity management for hospitality and tourism:
A review of current approaches54

Food production is the only manufacturing function in the hospitality sector 420,
“but it faces more unpredictable demand patterns than most goods
production.”54 This production is manageable, however; Creed recommends
short shelf-life systems (cook-hot-hold and cook-chill), which offer a relatively

322
 THE A.I. CRUISE LINE

small time buffer between food preparation and consumption and are suitable
for settings with predictable demand, such as banqueting or in-flight catering.421
“Extended shelf-life systems (cooked-in-a-bag and cook-freeze), on the other
hand, which allows for a longer time buffer, are suitable for unpredictable
demand patterns such as those of railway food services or restaurants that are
open to the public,” note Pullman and Rodgers. 54 “The practical implication of
this for tourism-oriented food services is that the capacity of their physical
resources can be better utilized through the production of packaged meals for
new markets (local cafes and restaurants as well as the retail sector) during a
slow season or periods of tourism downturn,” argue Pullman and Rodgers. 54

Sector Units Method

Cost-volume-profit422
Beds, rooms, percentage of
Hotels Regressions analysis423
occupancy
Economic order quantity424
Number of seats x hours of
Available seats per hour
service/service-cycle time399
Restaurants Degree of table TABLEMIX computer program425
“combinability”
Production equipment units Empirical formulae426
Average blackjack Minimum break-even bets401
Gaming
hands/hour

Theme parks and Rides Break-even analysis415

resorts Ski lifts Simulation modeling427
Benchmarking against the
Convention and consumption patterns (number
Square meters of exhibition
exhibition of customers per event, space,
area, tables, flatware, chairs
centers time requirements, etc.) and
future trends in demand428
Cruise ships Berths Dynamic game theory429

Table 15: Examples of strategic physical capacity estimation techniques

Source: Pullman and Rogers, Capacity management for hospitality and tourism:
A review of current approaches54

“The relationships between demand, capacity, managerial objectives, and the

visitor experience are complex and difficult to evaluate without appropriate
analytical techniques,” assert Pullman and Rodgers.54 Table 15 lists several
different methods used in strategic capacity estimation for industries such as

323
 ANDREW W. PEARSON

hotels, restaurants, gaming, theme parks, convention and exhibition centers and
cruise ships.54 The range of strategic estimation approaches run “from the
planning of overall tourism or hotel carrying capacity in a city (for long range
tourism or special events such as the Olympics) to such micro-level decisions as
the sizing of hotel water heaters,” note Pullman and Rodgers.54
Generally, the major capacity indicators – the number of berths on a cruise ship,
rooms in a hotel, seats in a restaurant, etc.) are determined during the initial
stages of major infrastructure projects. These indicators are then used later in
the estimation of other capacity elements,” contend Pullman and Rodgers.54
“Closed-form/cost-oriented methods are typically used in the accommodation
sector, which is capital intensive, since the capital costs, such as construction
costs, furnishings, property taxes, and insurance, vary with capacity” add
Pullman and Rodgers.54 An example of a closed-form solution is a typical break-
even formula such as the cost-volume-profit method430 put forth by Liu and Var:
𝑃 = 𝑄(𝑆 − 𝑉) − 𝐹
where P = total profit; Q = volume of sales (e.g., visitor-days); S = sale price per
unit volume; V = variable cost per unit volume; F = total fixed cost.
“Here, if a target profit is set, the volume of sales (Q) can be derived from this
equation, which together with the expected occupancy level would indicate the
capacity needed to meet the given profit objective. With this approach, the
capacity depends on the targeted sales price,” claim Pullman and Rodgers. 54
“Another closed-form approach, proposed by Carey423 uses regression models to
determine overall hotel carrying capacity for specific locations. Here the
dependent variable — optimal room occupancy — is a function of the
independent variables, price and overall market capacity,” note Pullman and
Rodgers.54
“A feasibility assessment of a new project also includes the determination of
variable or semi-variable costs affected by furniture/equipment needs and
layouts,” say Pullman and Rodgers.54 They add that, “These can be classified as
‘capacity elements’ and are usually estimated using empirical formulae with
results expressed in a variety of units (square meters for different functional
areas; number of parking places/rides/boats, etc.; customers per lift cycle for a
ski resort; standing spots for cruise line games; and others).”54 “These
approaches are usually simpler than the methods used in determining the major
capacity indicators and reflect the factors shaping the capacity on the micro-
level, which are usually not directly linked to profits,” explain Pullman and
Rodgers.54 For example, Muller provides a very simple empirical formula to
estimate capacity for a restaurant based on estimates of hours of service,
number of seats, cover count, and planned service-cycle time399:
seats x hours of service
maximum seating capacity =
service − cycle time

324
 THE A.I. CRUISE LINE

Pullman and Rodgers argue that, “The closer a capacity element is to the
customer interface in a physical sense, the more customer-related variables are
used in capacity estimation — the number of customers served per hour for a
dining-room or the percentage of people travelling by car for parking, for
example.”54 “To estimate the number of rides in a theme park, Wanhill415
provides a closed-form solution integrating the catchment’s population and
penetration rate, demand fluctuations between weekdays and weekend,
average ‘consumption’ of entertainment units per customer, and average ride
throughput,” claim Pullman and Rodgers.54
“Typically closed-form solutions must assume that variables such as pricing or
demand are constant or that certain conditions must be ignored to create
problem tractability. As an alternative, probabilistic models allow for more
complexity and realism. These approaches range from modifications of closed-
form approaches to large simulation models,” claim Pullman and Rodgers. 54 In
his Analysis of Las Vegas strip casino hotel capacity: an inventory model for
optimization424, Gu gives an example of a modified closed-form solution, that
“treated the supply of room nights of Las Vegas Strip cruise line hotels as
inventory units and integrated a probabilistic model based on the economic
order quantity (EOQ) equation to reflect the perishable nature of hotel services.”
“This approach allowed for the prediction of the cycles of over- and
undercapacity,” state Pullman and Rodgers.54
“In previous research, simulation modelling has been applied to capacity
problems such as restaurants, cruise ships and ski resorts,” note Pullman and
Rodgers.54 For example, Wall397 uses simulation modelling to account for
multiple inputs and the cyclic nature of capacity in tourism. For restaurant
capacity, the TABLEMIX model developed by Thompson
“provides the optimal percentage of combinable tables in a
restaurant by integrating the number and duration of peak
dining periods, the probabilities of different numbers of party
arrivals for the dining period, the probabilities of different size
parties, the maximum number of waiting parties, the
distribution of dining periods (normal or lognormal), and the
choice of table arrangement rule (e.g., assign an available
table to the largest party or to the party waiting the
longest).”425
“Using probabilistic models of market share and computer simulation to more
accurately integrate the relationship between capacity and demand,” Pullman
and Thompson developed a method for determining the profit-maximizing
capacity strategy for different hospitality environments.427 “This method
requires customer utility models (reflecting real customer’s preferences for
multiple capacity-related aspects of a service such as queue duration) with the
potential costs and revenues from the implementation of different capacity

325
 ANDREW W. PEARSON

management strategies,” explain Pullman and Rodgers.54 The writers continue,

saying that, “Using discrete-event simulation models of a resort’s subsequent
flow performance under different capacity scenarios, the profit maximizing
strategy simultaneously addresses both management and customer preferences
as shown below:

where t = time period (set T periods); Nt = number of potential visitors in period

t; Pjt = price of service j during time period t; V = variable cost per visitor for
service j; F = fixed cost of service j; pjt = potential market share or probability
that a customer group will select service j from the set of J services in period t.” 54
The above market share is derived from actual customer preference models
based on choice experiments by Verma et al.431 In this case, the probability that
an aggregate group chooses service j out of a set of J competitive service
establishments is as follows54:

where Uj = aggregate customers’ overall systematic utility for service design j.

Pullman and Rodgers see the advantages of these mixed-model approaches are
as follows:
“first, complex enterprises can be modelled with a great deal
of realism, including variable costs, dynamic customer arrivals
and flow patterns, multiple combinations of strategies, and
heterogeneous customer mixes. Alternate forecasts of visitor
behaviour and customer flows can be tested. Second, utility
models are typically based on actual customer preferences for
different attributes of an enterprise (for example, crowding
level, average queue time, price, retail and amenities level, mix
of activities, or skill level). Third, by integrating the customer
and enterprise perspectives, diverse criteria can be optimized,
including profit, revenue, and customer satisfaction. This
approach allows for the testing of various capacity
management strategies, enabling management to evaluate
their estimated costs prior to the capital investment.”
For enterprises, strategic labour capacity decisions cover a long-term planning
horizon, usually of 6-18 months.393 “As with physical capacity decisions, firms’
functional capacity strategies should reflect their competitive priorities,” argue
Pullman and Rodgers.54

326
 THE A.I. CRUISE LINE

From an operations management perspective, there are three typical strategies

for labour capacity planning: level, chase, or mixed. 432 Pullman and Rodgers
provide the following example: “hotels with a high service/competitive priority
might want to retain the majority of their staff throughout the year (a level
strategy), regardless of demand, to keep both the service level and the morale
high. A cost-focused hotel, on the other hand, might only staff as needed (a chase
strategy) to minimize labour costs.”54
Sasser argues that, a level strategy is more appropriate when demand is
quantifiable before its time of use, or the firm has strict requirements about
service availability, and/or the employees are scheduled according to pre-
existing criteria such as strict union contracts.432
“Typically, Service Factory processes with low interaction, customization, and
labour intensity use this strategy. Demand is scheduled to fit into an existing
plan, such as airline flight or tourist bus schedules. The visitor or guest reserves
a space in advance and the labour capacity is fixed for the flight or bus trip
regardless of the number of people on that trip,” explain Pullman and Rodgers. 54
“This approach is best suited to an environment in which labour requirements
are predetermined based on physical capacity, such as under safety regulations
(e.g., the number of flight attendants on a plane is fixed regardless of the number
of passengers),” state Pullman and Rodgers.54
A chase strategy tries to meet demand as it occurs.432 Pullman and Rodgers
contend that, “For this type of strategy, the labour requirements change to
match the demand forecast throughout the planning horizon. This approach is
best suited to an environment in which visitors cannot be prescheduled and
there is a high degree of labour interaction and customization (Professional
Services), or to other environments with unpredictable visitor arrivals such as
hotel front desks, shops, parks, and non-reservation restaurants.”54
Mixed strategies use some combination of level and chase approaches. 432
“Because of the seasonality of many tourism enterprises, it makes sense to use
level strategies for certain staff areas (e.g., cruise ship operators and
maintenance or other departments that must function regardless of the number
of visitors) and chase strategies for others (maid service, ski instructors, front
desk personnel, wait-staff, bell-staff, etc.),” recommend Pullman and Rodgers.54
“Often a leisure enterprise will change its labour capacity strategy as the
company grows,” claim Pullman and Rodgers.54 “For example, Port Aventura, a
Spanish theme park, used a chase strategy for their first season to match
employees to demand. During that season, the workforce reached 2800 on the
busiest days but went down to 200 employees during the winter season.” 54
Later, Huete and Segarra report, “the company signed a collective agreement
with their unions that offered permanent or permanent-seasonal contracts to
50% of the long-term workforce, thus requiring a mixed strategy.”433

327
 ANDREW W. PEARSON

“Labour capacity estimation for various strategies depends on the industry and
often on the planned physical capacity,”54 even with integrated resorts that are
now some of the biggest buildings in the world. “Here, industry standards play a
significant role, dictating how many employees of each type are required for a
plane, restaurant or hotel of a certain size,” argue Pullman and Rodgers.54
Sill provides examples of this approach for hospitality: An enterprise first
estimates a key demand input (e.g., number of expected guests, menu-mix
history, or potential bar sales), multiplies that by a given standard (staffing
standard, recipe, table-setting used) and gets a capacity output (staffing for the
dining-room, kitchen, dishwashing, food production, or bar). 434 “To cite another
example, in the hotel industry, union contracts might dictate how many rooms a
housekeeper may clean in a shift (perhaps 12–14 standard rooms), and the
required number of housekeepers can then be directly calculated from the
expected room occupancy,” explain Pullman and Rodgers.54
“For capacity estimation that involves a partial or full chase strategy, a typical
approach to labour staffing for long-term planning employs aggregate planning
techniques that calculate the economic trade-off between full- and part-time
employees, overtime, and hiring and lay-off costs to determine the labour
requirements,” argue Pullman and Rodgers.54
In their paper Introduction to Operations and Supply Chain Management 435,
Bozarth and Handfield believe, “These problems are solved optimally using linear
programming in which the objective is to minimize cost or maximize profit
subject to constraints such as the meeting of demand, limits on materials and
equipment time, and labour requirements for standard and overtime wages.” In
a typical inventory-less model formulation, Vollman et al.436 argue that the linear
objective function for cost minimization is the following:

subject to the following constraints436:

• that regular-time hours must not exceed the allowable maximum to be
worked per employee per month;
• workforce level change between months Ht Ft; and Xt Ot Dt
where CH, CF, CR, Co, and CU = the cost of hiring an employee, the cost of firing
an employee, the cost per labour-hour of regular time, the cost per labour-hour
of overtime, and the cost per labour-hour of idle regular time, respectively; Ht,
Ft, Xt, Ot, and Ut = the number of employees hired, employees fired, regular-
time hours, overtime hours, and idle regular-time hours in month t, respectively;
Dt = the number of hours of service to be sold in month t; and M = the number
of months in the planning horizon.436

328
 THE A.I. CRUISE LINE

Vollman et al. also believe that, “Similar linear programming models can be set
up to model more complex environments, such as different product or customer
classes, different employee classes (part-time and full-time), and many other
service situations.”436
Pullman and Rodgers argue that, “During the actual operation of an enterprise,
the ability to meet demand depends on both the flexibility of the physical
resources and the willingness of the price/market segment to pay for an
available capacity unit.”54

Physical capacity
Physical flexibility
Rent capacity
Share capacity
Hire sub-contractors
Change resource allocations
Change hours of operation
Provide off-site access
Use automation
Price/segment flexibility
Partition visitors (status and length of transaction)
Yield management
Revenue management
Human capacity
Labour flexibility
Schedule employees
Allow overtime
Allow idle time
Cross-train employees
Change work speed and process
Hire permanent employees
Lay-off employees
Use temporary employees
Use part-time employees
Visitor flexibility
Allow waiting
Allow balking
Turn away visitors
Provide rewards or incentives
Provide diversions or complementary services
Camouflage the queue
Pay for VIP queues
Change level of visitor participation
Schedule visitors/take reservations
Inform/educate about alternative options

Table 16: Short-term capacity management approaches

Source: Pullman and Rogers, Capacity management for hospitality and tourism:
A review of current approaches54

329
 ANDREW W. PEARSON

Table 16 provides a comprehensive list of the different physical resource

approaches, most of which are self-explanatory.389 402 “The majority of
price/segment flexibility approaches use capacity partitioning or yield/revenue
management techniques,” say Pullman and Rodgers.54 “Traditionally, yield
management has been used by airlines (price and class), lodging enterprises
(price and stay duration), restaurants (price, style, and duration of service/meal),
and more recently, by gaming venues (gaming spots per table and minimum
bets) and the golf-course industry (no-show fees, rental fees, playing times),”
state Pullman and Rodgers.54 The most basic methodology is the linear modelling
approach, which attempts “to maximize revenues from different customer
classes subject to the capacity constraints of the enterprise as illustrated below:

where i = rate class or segment, xi = number of items sold in rate class i, di =

demand for rate class i,ri = revenue from selling items in rate class i, and c =
capacity of the service.”54
The work of Lindenmeier and Tscheulin 437 and Sanchez and Satir438 have shown
that yield management methodologies have evolved to include probabilistic
demand variables, economic approaches, network approaches, customer
satisfaction implications, etc. “In addition, yield management has been applied
to almost all leisure and hospitality sectors, claim Pullman and Rodgers. 54
Table 17 provides examples of the various short-term capacity estimation
techniques used in different sectors.54
Sector Tactics References
Packages: budget, moderate, and deluxe Kandampully439
Categories: couples, singles, and families
Hotel/resort
Subcategories for children: kids, petit, and mini
baby/nursery
Off-peak discounts Muller399
Increase production rate by application of more
efficient equipment and simpler

Restaurant Reduce randomness via well-designed Kimes et al.440

reservation system
Increase sophistication through application of
queuing theories, modelling and simulations,
linear programming, and forecasting

330
 THE A.I. CRUISE LINE

Sector Tactics References

In low-demand periods: assign multiple duties to
the host, increase menu variety, offer regular or
reduced prices, increase promotion, have servers
bring condiments
Capacity upgrades: improved lifts Pullman and
Thompson427
Capacity expansion: additional terrain
Queuing information
Ski resort
Inter-daily demand smoothing: surcharge during
weekends
Class variation: mix of beginners and
intermediate
Reduce the uncertainty of arrival by means of Kimes442
forecasting, overbooking, requiring guaranteed
reservations, reconfirming reservations by
phone, instituting a no-show fee and/or a fixed
fee for reservations
Reduce the uncertainty of duration by selling
Golf course rounds of predictable length, flexible course
design with a global positioning system to direct
flow, requiring own golf carts or reducing the cart
rental fee during busy times, using course
marshals and caddies, posting golfers’ playing
times
Reduce the tee-time interval
Coping: close part of year, maintain the premises, Getz and
‘reduce staff’ seek other income, borrow money Nilsson441
Combating: stay open all year, develop other
Small tourist segments, cater to residents, augment the
Inn/B&B product’s appeal, add value, develop export
products
Capitulating: shrink, sell, or terminate the
business

Table 17: Capacity estimation in different hospitality and tourism sectors 54

Source: Pullman and Rogers, Capacity management for hospitality and tourism:
A review of current approaches54

“The capacity requirements of different customer groups cannot only affect the
revenue but the overall utilization of a facility,” say Pullman and Rodgers. 54 They

331
 ANDREW W. PEARSON

provide the example of a ski resort that attempted to increase utilization; “a ski
resort with a large percentage of terrain for expert or advanced skiers increased
its promotions for the ‘family skier’ segment (skiers with beginner-level and
intermediate skills).”54 Although the change reduced the service quality in terms
of queuing, a resulting increase in spending on lessons and amenities was
observed. 427
In a golf course setting, experienced golfers usually play faster than novices.54
Kimes found that offering separate playing times for beginners relieved the
experienced players’ frustration and improved overall flow. 442 Simón et al.
discovered that, for outdoor recreation enterprises, the attraction of
environmentally aware tourists in order to minimize ecological damage and
increase tourism carrying capacity had a similar effect on the operations. 403
“Tactical human capacity approaches involve labour and/or customer flexibility.
From the labour perspective (Table 17), most of the flexibility comes from hiring
an appropriate number of employees, scheduling employees to meet demand as
it occurs, and adequately training employees to do their respective jobs or cross-
training them to do multiple jobs,” say Pullman and Rodgers. 54 “The majority of
these labour-related tactical capacity management approaches derive from the
operations area and are based on queuing theory,” note Pullman and Rodgers. 54
According to Sundarapandian, queueing theory is the mathematical study of
waiting lines, or queues.443
In Labour scheduling, part 1: forecasting demand 444, Thompson explains that,
matching visitor demand to labour capacity involves a four-step process:
1. forecasting demand;
2. translating the demand forecasts into requirements for employees;
3. developing a workforce schedule that, ideally, only has employees
working when necessary to deliver the service; and
4. controlling the delivery of the service in real time.
“From the customer perspective, flexibility tactics involve the management of
waiting lines or ‘capacity buffers,’” says Pullman and Rodgers. 54 The process,
Pullman and Rodgers contend, “can be designed to allow visitors to merely wait;
to let them leave rather than keeping them ‘captive’ in roped areas; to provide
diversions to improve the line experience; to hide the queue; to offer higher
prices in exchange for a shorter queue; or to allow virtual queuing (e.g.,
electronic queuing, such as Disney’s Fast Pass system).”54
Using queuing theory and the appropriate model, analysts can determine the
appropriate number of workers based on line configuration, desired length of
lines, customer wait duration, and average service times, says Kleinrock. 445
“Alternatively, visitors can increase their own level of participation in the
process, allowing the enterprise to increase the available labour capacity,” offer
Pullman and Rodgers.54 They provide the example of a self-guided audio-tour

332
 THE A.I. CRUISE LINE

system for museums, parks, convention centers, arenas, and other tourism sites
to reduce the number of guides required.54
A wide range of labour capacity techniques are available that address these
problems. Thompson444 446 447 448 offers a comprehensive tutorial on the
development of optimal workforce schedules for multiple employee
classifications (part-time, breaks, and reliefs). Thompson’s main goal is to
minimize employee cost, while meeting customer service-time goals.54
Klassen and Rohleder ran multiple simulated scenarios that combined a number
of labour and visitor flexibility options simultaneously to determine the best
combinations of approaches.389 They found that enterprises with limited
flexibility in terms of maximum schedulable staff should make use of customer
participation, employee cross-training, and price segmentation and should
inform or educate customers about other alternatives. 389
Finally, the real-time adjustment of labour schedules448 “refers to methods in
which managers observe actual visitor demand and correct any capacity
problems as they occur. Here the goal is to improve service and cost
performance in real time.”54 For example, Pullman and Rodgers recommend, “if
a restaurant’s customer flow is slower than expected, employees can go on
break or leave early; if it is busier than expected, employees can defer breaks
until a later point in their schedule.”54
Real-time schedule adjustments include changing employees’ station
assignments, cancelling or changing their breaks, starting their shifts early/late
or asking them to leave early/ late, calling in more workers, or cancelling shifts.449
Thompson indicates that schedules developed with preset employee breaks
outperformed those with real-time break adjustment in terms of overall cost.448
However, Hur et al. found that the active adjustment of work schedules is
beneficial as long as the direction of demand change is accurately identified. 449
Table 18 (across two pages) contains a detailed analysis of the advantages and
disadvantages of various physical and labour capacity estimation approaches. 54
As per Pullman and Rodgers, “Main criteria include the simplicity of use, the
modelling skills and amount of data needed, and the adaptability of the
approach to different environments and over- and undercapacity cycles.”54
As previously shown, “service capacity utilization has a significant influence on
visitors’ perceptions of quality: both over- and underutilization can be
undesirable.”54 “Models that include both quality and productivity measures
may enable enterprises to determine optimal physical as well as social carrying
capacity for a variety of environments,” conclude Pullman and Rodgers.54

333
 Type Description Time frame Purpose Limitations Advantages
Physical Method uses historic Performed during To estimate hotel Requires historic data for Allows for cycles of
capacity: data with regression the planning phase carrying capacity or the specific location. over- and under-
statistical and/or probabilistic of the project. room occupancy. Difficult to extrapolate to capacity.
methods models to predict new locations.
requirements.
Physical capacity: Method weighs cost, Performed during To estimate the number Requires predictions of Relatively simple
closed-form/cost volume and profit to the planning phase of capacity units (berths prices and costs. Does approach with closed-
focus determine breakeven of project. on ships, seats in plane not account for form solution.
ANDREW W. PEARSON

capacity. or restaurant, hotel competitive

rooms, etc.) based on environment and
desired profit and costs. economic fluctuations.

334
Physical capacity: Method weighs Performed during To estimate appropriate Requires market Allows for most
mixed statistical customer preferences, the planning phase capacity based on research on customer realistic view of
and customer competitive market, of project. customer preferences preferences and existing multiple performance
preference and simulation of for different attributes of or proposed competitors measures under any
models enterprise scenarios. the proposed enterprise along with modelling of number of future
in a competitive market. enterprise function. scenarios.
Physical capacity: Method models the Performed during To simulate the Requires sophisticated Allows for high degree
simulation proposed facility the planning phase customer flow and modelling skills, of realism and
models under various of the project. queuing under different projected demand data, multiple scenario
conditions. levels of capacity software, and complete evaluations. Can test
flexibility, layout, and understanding of design changes of
capacity choices. operating environment existing enterprises
and rules. without disrupting
flow.
 Source: Pullman and Rogers, Capacity management for hospitality and tourism:
Type Description Time frame Purpose Limitations Advantages
Physical Estimate uses Performed as To estimate the right unit Requires historical Determines optimal
capacity: yield historical and current needed, of capacity to sell to the data, programming price, customer class
management demand to determine depending on the right customer at the right ability or specialized size, and capacity

Table 18: Capacity estimation approaches and their application

optimal capacity use. demand changes price and for the right software.
for the enterprise. duration.
Labour capacity: Method links labour Performed after To estimate the number of May not allow for Relatively simple to
deterministic requirement to physical capacity is employees based on flexibility or implement.
physical capacity determined. heuristic or industry adaptation to actual
THE A.I. CRUISE LINE

units or regulations. regulations. needs.

335
Labour capacity: Method finds Performed as To estimate the number of Requires Determines optimal cost
linear minimal labour needed (daily, employees needed, given programming ability and profit schedules.
programming requirements based weekly, monthly) projected demand, or specialized

A review of current approaches54

on fluctuating depending on the employee service rate, software. Can be
demand and supply industry. contract schedule difficult to account
conditions. requirements, and for different
constraints on service employee skills or
level. productivity levels.
Labour or Methods allow Performed during To free up capacity by Requires an Allows customers to
physical customer to choose enterprise allowing customer to self- additional channel for choose their experience
capacity: service method. operations. serve or choose virtual selfservice or a virtual and reduces capacity
customer lines. queue system costs.
flexibility tactic
 ANDREW W. PEARSON

Cost of Outages
According to Ponemon Institute’s Cost of Data Center Outages January 2016
Data Center Performance Benchmark Series450 study, “the average cost of a data
center outage has steadily increased from $505,502 in 2010 to $740,357 today
(or a 38 percent net change).” The institute’s benchmark analysis focused “on
representative samples of organizations in different industry sectors that
experienced at least one complete or partial unplanned data center outage
during the past 12 months.”450 The Institute claims that, “Utilizing activity-based
costing methods, this year’s analysis is derived from 63 data centers located in
the United States. Following are the functional leaders within each organization
who participated in the benchmarking process450:
• Facility manager
• Chief information officer
• Data center management
• Chief information security officer
• IT operations management
• IT compliance & audit
• Operations & engineering
Utilizing activity-based costing, Ponemon captured information about both
direct and indirect costs, including but not limited to the following areas 450:
• Damage to mission-critical data.
• Impact of downtime on organizational productivity.
• Damages to equipment and other assets.
• Cost to detect and remediate systems and core business processes.
• Legal and regulatory impact, including litigation defense cost.
• Lost confidence and trust among key stakeholders.
• Diminishment of marketplace brand and reputation.
Key findings of Ponemon’s benchmark research study included450:
• The average cost of a data center outage rose from $690,204 in 2013 to
$740,357, a 7% increase. The cost of downtime increased 38% since the
first study in 2010.
• Downtime costs for the most data center-dependent businesses are
rising faster than average.
• Maximum downtime costs increased 32% since 2013 and 81% since
2010. Maximum downtime costs for 2016 are $2,409,991.
• UPS system failure continues to be the number one cause of unplanned
data center outages, accounting for one-quarter of all such events.
• Cybercrime represents the fastest growing cause of data center
outages, rising from 2% of outages in 2010 to 18% 2013 to 22% in the
latest study.

336
 THE A.I. CRUISE LINE

Utilizing activity-based costing, Ponemon’s study “addresses eight core process-

related activities that drive a range of expenditures associated with a company’s
response to a data center outage.”450 The activities and cost centers used in our
analysis are defined as follows450:
1. Detection cost: “Activities associated with the initial discovery and
subsequent investigation of the partial or complete outage incident.” 450
2. Containment cost: “Activities and associated costs that enable a
company to reasonably prevent an outage from spreading, worsening
or causing greater disruption.”450
3. Recovery cost: “Activities and associated costs that relate to bringing
the organization’s networks and core systems back to a state of
readiness.”450
4. Ex-post response cost: “All after-the-fact incidental costs associated
with business disruption and recovery.”450
5. Equipment cost: “The cost of new equipment purchases and repairs,
including refurbishment.”450
6. IT productivity loss: “The lost time and related expenses associated with
IT personnel downtime.”450
7. User productivity loss: “The lost time and related expenses associated
with end-user downtime.”450
8. Third-party cost: “The cost of contractors, consultants, auditors and
other specialists engaged to help resolve unplanned outages.” 450
“In addition to the above process-related activities, most companies experience
opportunity costs associated with the data center outage,” claims the Ponemon
report.450 This can result in lost revenue, business disruption and average
contribution.450 Accordingly, Ponemon’s cost framework includes the following
categories450:
• Lost revenues: The total revenue loss from customers and potential
customers because of their inability to access core systems during the
outage period.
• Business disruption (consequences): The total economic loss of the
outage, including reputational damages, customer churn and lost
business opportunities.
Figure 24 presents the activity-based costing framework used in the Ponemon
research, which consists of 10 discernible categories.450 As shown below, the
four internal activities or cost centers include detection, containment, recovery
and ex-post response.450 “Each activity generates direct, indirect and
opportunity costs, respectively,” notes Ponemon.450 “The consequence of the
unplanned data center outage includes equipment repair or replacement, IT
productivity loss, end-user productivity loss, third parties (such as consultants),
lost revenues and the overall disruption to core business processes. Taken
together, we then infer the cost of an unplanned data center outage,” argues

337
 ANDREW W. PEARSON

Ponemon.450

Figure 24: Activity-Based Cost Account Framework

Source: Ponemon Institute

Ponemon’s “benchmark instrument was designed to collect descriptive

information from IT practitioners and managers of data center facilities about
the costs incurred either directly or indirectly as a result of unplanned
outages.”450 The survey design relied “upon a shadow costing method used in
applied economic research. This method does not require subjects to provide
actual accounting results, but instead relies on broad estimates based on the
experience of individuals within participating organizations.”450
“The benchmark framework in Figure 24 presents the two separate cost streams
used to measure the total cost of an unplanned outage for each participating
organization,” claims Ponemon.450 “These two cost streams pertain to internal
activities and the external consequences experienced by organizations during or
after experiencing an incident,” says Ponemon.450 Its benchmark methodology
“contains questions designed to elicit the actual experiences and consequences
of each incident.”450 Ponemon believes that, “This cost study is unique in
addressing the core systems and business process-related activities that drive a
range of expenditures associated with a company’s incident management
response.”450
For Ponemon, the process worked as such450:
“Within each category, cost estimation is a two-stage process.

338
 THE A.I. CRUISE LINE

First, the survey requires individuals to provide direct cost

estimates for each cost category by checking a range variable.
A range variable is used rather than a point estimate to
preserve confidentiality (in order to ensure a higher response
rate). Second, the survey requires participants to provide a
second estimate for both indirect cost and opportunity cost,
separately. These estimates are calculated based on the
relative magnitude of these costs in comparison to a direct cost
within a given category. Finally, we conduct a follow-up
interview to obtain additional facts, including estimated
revenue losses as a result of the outage.”
In total, the benchmark instrument contained “descriptive costs for each one of
the five cost activity centers.”450 Within each cost activity center, respondents
were asked to estimate the cost in three different ranges — direct cost, indirect
cost and opportunity cost.450 These are defined as follows450:
• Direct cost – the direct expense outlay to accomplish a given activity.
• Indirect cost – the amount of time, effort and other organizational
resources spent, but not as a direct cash outlay.
• Opportunity cost – the cost resulting from lost business opportunities
as a consequence of reputation diminishment after the outage.
Table 19 summarizes the frequency of companies and separate data centers
participating in the benchmark study.450 A total of 16 industries are represented
in the sample.450 Six data center organizations were rejected from the final
sample for incomplete responses to the survey, which resulted in a final sample
of 63 separate data centers.450

Data
Industries Companies Rejected Final sample
centers
Co-location 4 11 1 10
Communications 1 2 0 2
Consumer
products 2 2 0 2
eCommerce 7 7 1 6
Education 2 3 0 3
Financial services 6 11 2 9
Healthcare 5 5 0 5
Hospitality 1 1 0 1
Industrial 4 6 0 6
Media 1 2 0 2
Public sector 4 5 1 4

339
 ANDREW W. PEARSON

Research 1 1 0 1
Retail 5 5 0 5
Services 5 7 1 6
Transportation 1 1 0 1
Total 49 69 6 63
Table 19: Tally of cost studies by recruited company & data center
Source: Ponemon Institute

Table 20 summarizes the participating “data center size according to total square
footage and the duration of both partial and complete unplanned outages.” 450
According to Ponemon, “The average size of the data center in this study is
14,090 square feet and the average outage duration is 95 minutes.” 450

Data center square

Key stats
footage Duration in minutes
Average 14,090 95
Maximum 55,000 415
Minimum 1,505 13
Table 20: Key statistics on data center size and duration of the outage
Source: Ponemon Institute

$2,409,991
Maximum $1,733,433
$1,017,746
$740,357
Mean $690,204
$505,502
$648,174
Median $627,418
$507,052
$70,512
Minimum $74,223
$38,969

$- $500,000 $1,000,000 $1,500,000 $2,000,000 $2,500,000 $3,000,000

2016 2013 2010

Figure 25: Key statistic on data center outages

Comparison of 2010, 2013, and 2016 results, Source:
Ponemon Institute

340
 THE A.I. CRUISE LINE

Figure 25 reports key statistics on the cost of unplanned outages as reported in

2010, 2013 and 2016.450 The cost range, median and mean all consistently show
substantial increases in cost over time.450
As the Ponemon report shows, “the maximum cost has more than doubled over
six years from just over $1 million to $2.4 million (a 34 percent increase since the
last study). Both mean and median costs increased since 2010 with net changes
of 38 and 24 percent respectively. Even though the minimum data center outage
cost decreased between 2013 and 2016, this statistic increased significantly over
six years, with a net change of 58 percent.”450
Figure 26 reports the cost structure on a percentage basis for all cost activities
for FY 2010, 2013 and 2016.450 As the Ponemon reports shows, “the cost mix has
remained stable over the past five years. Indirect cost represents about half and
opportunity loss represents 12 percent of total cost of outages.” 450

60% 52% 51%

50%
50%
38% 35% 36%
40%
30%
20% 12% 12% 12%
10%
0%
Direct cost Indirect cost Opportunity cost

2010 2013 2016

Figure 26: Percentage cost structure of unplanned data center outages

Comparison of 2010, 2013 and 2016 results
Source: Ponemon Institute

Table 21 summarizes the cost of unplanned outages for all 63 data centers. It
should be noted, however, that cost statistics are derived from the analysis of
only one unplanned outage incident.450

341
 ANDREW W. PEARSON

Cost categories Total Mean Median Minimum Maximum

Third parties $625,401 $9,927 $6,970 $1,551 $27,600
Equipment $597,114 $9,478 $8,865 $1,249 $67,783
Ex-post
activities $530,964 $8,428 $11,566 $36,575
Recovery $1,334,151 $21,177 $17,570 $1,900 $58,171
Detection $1,682,856 $26,712 $22,813 $877 $69,100
IT productivity $3,898,440 $61,880 $56,789 $6,994 $125,600
End-user
productivity $8,706,159 $138,193 $124,551 $15,600 $456,912
Lost revenue $13,141,737 $208,599 $197,500 $26,591 $755,810
Business
disruption $16,125,669 $255,963 $201,550 $15,750 $812,440
Total cost $46,642,491 $740,357 $648,174 $70,512 $2,409,991

Table 21: Cost summary for unplanned outages

Source: Ponemon Institute

256.0
Business disruption 238.7
179.8
208.6
Lost revenue 183.7
118.1
138.2
End-user productivity 140.5
96.2
61.9
IT productivity 53.6
42.5
26.7
Detection 23.8
22.3
21.2
Recovery 22.0
20.9
8.4
Ex-post activities 9.6
9.5
9.5
Equipment 9.7
9.1
9.9
Third parties 8.6
7.0
0.0 50.0 100.0 150.0 200.0 250.0 300.0

2016 2013 2010

Figure 27: Comparison of activity cost categories

Comparison of 2010, 2013 and 2016 results, $1,000 omitted
Source: Ponemon Institute

342
 THE A.I. CRUISE LINE

Figure 27 shows that the Ponemon study reveals a relatively consistent pattern
across nine cost categories over five years (three studies).450 “The cost associated
with business disruption, which includes reputation damages and customer
churn, represents the most expensive cost category. Least expensive involves
the engagement of third parties such as consultants to aid in the resolution of
the incident.”450

25%
UPS system failure 24%
29%
22%
Cyber crime 18%
2%
22%
Accidental/human error 22%
24%
11%
Water, heat or CRAC failure 12%
15%
10%
Weather related 12%
12%
6%
Generator failure 7%
10%
4%
IT equipment failure 4%
5%
0%
Other 0%
2%

0% 5% 10% 15% 20% 25% 30% 35%

2016 2013 2010

Figure 28: Root causes of unplanned outages

Comparison of 2010, 2013 and 2016 results
Source: Ponemon Institute

Figure 28 analyzes the sample of 63 data centers by the primary root cause of
unplanned outage.450 The “other” category refers to incidents where the root
cause could not be determined.450 As the chart shows, 25 percent of companies
cite UPS system failure as the primary root cause of the incident. 450 Twenty-two
percent cite accidental or human error and cyberattack as the primary root
causes of the outage.450 For human error, this is the same percentage as was
found in 2013, indicating that there has been no progress in reducing what
should be an avoidable cause of downtime. 450 The 22 percent figure for
cybercrime represents a 20 percent increase from 2013 and a 167 percent
increase from 2010.450 IT equipment failure represents only four percent of all
outages studied in this research.450

343
 ANDREW W. PEARSON

$1,000,000

$800,000

$600,000

$400,000

$200,000

$-
2010 2013 2016

Partial unplanned outage Total unplanned outage Overall average cost

Figure 29: Cost for partial and total shutdown

Comparison of 2010, 2013 and 2016 results
Source: Ponemon Institute

Figure 29 compares costs for partial unplanned outages and complete

unplanned outages.450 All three studies show complete outages are more than
twice as expensive as partial outages.450

Total cost of unplanned outages

Financial services 994

Communciations 970
Healthcare 918
eCommerce 909
Co-location 849
Research 807
Consumer products 781
Industrial 761
Retail 759
Education 648
Transportation 645
Services 570
Hospitality 514
Media 506
Public sector 476

Figure 30: Distribution of total cost for 15 industry segments

$1,000 omitted
Source: Ponemon Institute

344
 THE A.I. CRUISE LINE

Figure 30 provides the total cost of unplanned outages for the 15 industry
segments included in the benchmark sample.450
Analysis by industry is limited because of a small sample size; however, it is
interesting to see wide variation across segments ranging from a high of
$994,000 (financial services) to a low of $476,000 (public sector).450

160
140
120
100
80
60
40
20
-
2010 2013 2016

Partial unplanned outage Total unplanned outage Overall average cost

Figure 31: Duration for partial and total shutdown (measured in minutes)
Comparison of 2010, 2013 and 2016 results
Source: Ponemon Institute

Figure 31 compares the average duration (minutes) of the event for partial and
complete outages.450 In the 2016 study, complete unplanned outages, on
average, last 66 minutes longer than partial outages. 450 “It is also interesting to
note a U-shape relationship in duration over six years — wherein unplanned
outages on average decreased 15 minutes between 2010 and 2013 and
increased 11 minutes between 2013 and 2016.”450 One possible explanation for
the increase in duration time is the rise of cyberattacks, which are difficult to
detect and contain.450

$20,000
$15,000
$10,000
$5,000
$0
Minimum Median Mean Maximum

2010 2013 2016

Figure 32: Total cost per minute of an unplanned outage

Comparison of 2010, 2013 and 2016
Source: Ponemon Institute

345
 ANDREW W. PEARSON

Figure 32 reports the minimum, median, mean and maximum cost per minute of
unplanned outages computed from 63 data centers. 450 It shows the most
expensive cost of an unplanned outage is over $17,000 per minute. On average,
the cost of an unplanned outage per minute is nearly $9,000 per incident. 450

$995
IT equipment failure $959
$750
$981
Cyber crime (DDoS) $822
$613
$709
UPS system failure $678
$688
Water, heat or CRAC $589
$517
failure $489
$528
Generator failure $501
$463
Accidental/human $489
$380
error $298
$455
Weather related $436
$395
$- $200 $400 $600 $800 $1,000 $1,200
2016 2013 2010

Figure 33: Total cost by primary root causes of unplanned outages

Comparison of 2010, 2013 and 2016 results, $1,000 omitted
Source: Ponemon Institute

Figure 33 reports the average cost of outage by primary root cause of the
incident.450 As shown below, IT equipment failures result in the highest outage
cost, followed by cybercrime.450 The least expensive root cause appears to be
related to weather followed by accidental/ human errors. 450

Unused Capacity
As Zoltán Sebestyén and Viktor Juhász claim in their paper The Impact Of The
Cost Of Unused Capacity On Production Planning Of Flexible Manufacturing
Systems451, “capacity is one of the most important measures of resources used
in production. Its definition and analysis are therefore one of the key areas of
production management.” Sebestyén and Juhász believe that, “The use of
conventional parameters often leads to wrong decisions.451
For Sebestyén and Juhász, “There are three aspects of the problem of

346
 THE A.I. CRUISE LINE

conventional capacity measures: the absence of economic content, quantity

based approach, and the unduly high emphasis laid on technical processes.”451
Sebestyén and Juhász believe that, “If capacity measures could side step the
problems discussed above, i.e. if they could include the value of resources, and
could refer to the costs of unused capacity, then better decisions could be made
in a number of cases.”451 “Changes in the nature of production, and the
enhanced significance of auxiliary processes made calculations necessary for
production and service systems where processes are difficult to quantify,” argue
Sebestyén and Juhász.451 “The appearance of activity-based costing (ABC) solved
precisely these problems, because the main goal of the method is to analyse and
differentiate the overhead costs associated with capacity maintenance and
operation and support processes.”451 “When using ABC for the determination of
cost data, the results are also appropriate for performing capacity usage
calculations,” claim Sebestyén and Juhász.451
“The ABC calculation attempts to approximate an ideal situation where the
overhead is incorporated in the product only to the extent it actually exploits its
resources,” say Sebestyén and Juhász.451 “In the past, due to the small ratio of
overheads, it was sufficient to use a single cost driver for overhead allocation.
The presently observed growth of the ratio of overheads makes it important to
allocate them accurately to different products. Different products use resources
to a different extent, therefore, several cost drivers are used,” the writers
contend.451 The ideal case say Sebestyén and Juhász “would be to use a separate
cost driver for every cost, but this is impossible to realize. Costs must be collected
in cost centers, and a characteristic cost driver must be assigned to each cost
group in the center.”451
“To perform a capacity calculation, we have to know the ratio of the costs of the
machine, equipment, plant or division depending on, and independent of the
output,” claim Sebestyén and Juhász.451 Their work builds upon the work of
Cooper and Kaplan in their Activity-Based System: Measuring the Cost of
Resource Usage.452 “These costs are closely related to the resource where they
appear. The operated resources can be divided into groups on the basis whether
they are provided according to their usage needs or in advance, without the prior
knowledge of these needs.”451 452 “There is no capacity problem associated with
resources provided on the basis of direct needs, because these are provided on
the basis of known or estimated needs,” claim Sebestyén and Juhász.451
“Resources provided in advance are those that are made available independent
of direct needs. Even though needs are forecasted by different methods to some
degree, still there is the constraint of availability prior to the time when the need
actually appears,” add Sebestyén and Juhász.451
For Sebestyén and Juhász, “There are three typical groups of resources provided
in advance. One of them is investment. The company invests in machines,
equipment, or buildings that present costs to be paid in advance and that are

347
 ANDREW W. PEARSON

expected to operate for quite a number of years.”451 “Another cost type is the
contract costs. In this case, the company signs a contract for the future use of a
service. The third and last typical example is that of workforce. This includes
those employees who are paid a fixed wage.” 451
“The cost of unused capacity can be calculated when the fix cost of the resource,
the actual resource usage and the effective capacity are known,” contend
Sebestyén and Juhász.451 The determination of the group of resources allocated
in advance, the collection of their fixed costs, and the measurement of the actual
capacity usage require the development of the management information
system.453 For Cooper and Kaplan, “The analysis of the cost of unused capacity is
based on the following simple formula452:

Activity Availability = Activity Usage + Unused Capacity.

“The cost of capacity is the entire cost paid beforehand to obtain the resource
under consideration,” explain Sebestyén and Juhász.451 This consists of the costs
of capacity rightfully used in the operation – also called exploited – and the cost
of unnecessarily allocated, that is, unused capacity, as shown in Figure 34.452 453
454

Figure 34: Interpretation of the cost of unused capacity

Source: Zoltán Sebestyén and Viktor Juhász, The Impact Of The Cost Of Unused
Capacity On Production Planning Of Flexible Manufacturing Systems

“The separation into two parts of capacity costs can be appropriately done by
linear approximation,” claim Sebestyén and Juhász.451 “There also exist

348
 THE A.I. CRUISE LINE

theoretical models for the description of capacity utilization costs in terms of a

non-linear function, but they are seldom used in practice,” they add. 451 The unit
fix cost of the resource in question is derived by “Dividing the costs of availability
of the resource during a given period by the quantity of the resource available
during the same period.”451 This is just the slope of the linear function shown in
Figure 34.451 “Knowing the quantity of the used resource and the unit cost it is
easy to calculate the cost of unused capacity,” argue Cooper and Kaplan.452 “If ui
is the capacity, hi is the actual production, and Fi is the total fix cost of the
resource, then the unused capacity is given, according to the formula below” 452:

Streaming Analytics
As The Cluetrain Manifesto364 points out, “Real-time marketing is the execution
of a thoughtful and strategic plan specifically designed to engage customers on
their terms via digital social technologies.” Adding to that description, Wikipedia
notes that real-time marketing is455:
“Marketing performed ‘on-the-fly’ to determine an
appropriate or optimal approach to a particular customer at a
particular time and place. It is a form of market research
inbound marketing that seeks the most appropriate offer for a
given customer sales opportunity, reversing the traditional
outbound marketing (or interruption marketing) which aims to
acquire appropriate customers for a given 'pre-defined' offer.”
Real-time marketing can be inexpensive compared to the cost of traditional paid
media. “Expensive research, focus groups, and awareness campaigns can be
replaced with online surveys, blog comments, and tweets by anyone or any
business,” add Macy and Thompson in their book The Power of Real-Time Social
Media Marketing.456 Just to be clear, the expense of real-time marketing might
be low compared to running through traditional media channels, but setting up
an IT operation that can hit a level of personalization that will wow a customer
is anything but cheap, or simple.
In his article How Real-time Marketing Technology Can Transform Your
Business23, Dan Woods’ amusing comparison of the differing environments that
marketers face today as compared to what their 1980s counterpart faced is
highly instructive as today’s marketing executives don’t have time for a market
research study in his sort of figurative first-person-shooter game. “The data
arrives too late and isn’t connected to the modern weapons of marketing. The
world is now bursting with data from social media, web traffic, mobile devices,

349
 ANDREW W. PEARSON

and tripwires of all kinds,” Woods warns.23

Today, most large companies have massive amounts of data pertaining to
consumer behavior coming at them constantly, from all angles. The challenge is
to make sense of the data in time to matter, to understand how consumer
attitudes and behaviors are changing and how they are being changed by
marketing and advertising efforts; to grab the treasure and avoid the pitfalls of
unleashing a Pandora’s box full of marketing furies.
The challenge in understanding the modern consumer is making sense of all of
the customer data, coming in from these vast unstructured sources. 23 Some of
this information explains the broad fluctuations in mass opinion, while other
evidence clarifies what consumers might be doing on a company website.23
Others still explain what consumers have done en masse or as individuals. 23 Still
other data can be collected after a customer trip in the form of surveys, whether
they are mobile, physical, or verbal surveys.
In his article When do you need an Event Stream Processing platform? 457, Roy
Schulte states that:
“An event is anything that happens. An event object (or
‘event,’ event message, or event tuple) is an object that
represents, encodes, or records an event, generally for the
purpose of computer processing. Event objects usually include
data about the type of activity, when the activity happened
(e.g., a time and date stamp), and sometimes the location of
the activity, its cause, and other information. An event
stream is a sequence of event objects, typically in order by time
of arrival.”
Large cruise line companies typically have three kinds of event streams:
• Copies of business transactions, such as customer orders, bank deposits
or withdrawals, customer address changes, call data records, advance
shipping notices, or invoices.457 These are generated mostly internally,
and reflect the operational activities of the company.457
• “The second are information reports, such as tweets, news feed articles,
market data, weather reports, and social media updates, including
Facebook and LinkedIn posts.”457 According to Schulte, “most of these
sources are external to the company, but may contain information that
is relevant to a decision within the company.”457
• “The third, and fastest growing, kind of event stream contains sensor
data coming from physical assets.”457 Generally known as IoT data, this
includes “GPS-based location data from vehicles or smart phones,
temperature or accelerometer data from sensors, RFID tag readings,
heart beats from patient monitors, and signals from supervisory control
and data access (SCADA) systems on machines.”457

350
 THE A.I. CRUISE LINE

The reason for performing analytics on one or more event streams is to obtain
information value from the data.457 As Schulte explains, “A stream analytics
application converts the raw input data (base events), into a form, derived
events, that is better suited for making decisions. The derived events
are complex events, which means that they are events that are abstracted from
one or more other events.457
Stream analytics are executed in one of two ways, push-based, continuous
intelligence systems, which recalculate as new data arrives without being asked
to or pull-based systems that run when a person enters a request, or a timer
sends a signal to produce a batch report. Event Stream Processing (ESP)
platforms are mostly relevant in highly demanding, push-based systems, but
they are occasionally used for pull-based analytics on historical data.457
When people think of ESP, they usually think of push-based continuous
intelligence systems, which ingests ongoing flows of event data and provide
situation awareness, while also supporting near-real-time, sense-and-respond
business processes.457 “Continuous intelligence systems typically refresh
dashboards every second or minute, send alerts, or implement hands-free
decision automation scenarios,” Schulte explains. “They may be used to monitor
a data source, such as Twitter, or a business operation, such as a customer
contact center, supply chain, water utility, telecommunication network, truck
fleet, or payment process,” adds Schulte.457
Schulte explains that457:
“ESP platforms are software subsystems that process data in
motion, as each event arrives. The query is pre-loaded, so the
data comes to the query rather than the query coming to the
data. ESP platforms retain a relatively small working set of
stream data in memory for the duration of a limited time
window, typically seconds to hours — just long enough to
detect patterns or compute queries. The platforms are more
flexible than hardwired applications because the query can be
adjusted to handle different kinds of input data, different time
windows (e.g., one minute or one hour instead of ten minutes)
and different search terms.”
According to Schulte, continuous intelligence applications are best implemented
on ESP platforms if457:
• A high volume of data (thousands or millions of events per second).
• Frequently recalculated results (every millisecond or every few
seconds).
• Multiple simultaneous queries are applied to the same input event
stream.

351
 ANDREW W. PEARSON

Schulte gives the example of Twitter’s ESP platforms, Storm and Heron as
powerful streaming platforms.457 These DWs are “used to monitor Twitter, which
averages about 6,000 tweets per second. A simple query might report the
number of tweets that included the word ‘inflation’ in the past ten minutes.
However, at any one time, there may be thousands of simultaneous queries in
effect against Twitter, each looking for different key words or different time
windows.”457
“In high volume scenarios, ESP platform applications can scale out vertically
(multiple engines working in parallel on the same step in a processing flow)
and/or horizontally (split the work up in a sequence or pipeline where work is
handed from one engine to the next while working on the same multistep event
processing query (i.e., an event processing network),” notes Schulte.457
Schulte notes that, “On-demand analytics are pull-based applications that
support ad hoc data exploration, visualization and analysis of data.” 457 On-
demand analytics can be used with historical event data to build analytical
models.457 In this context, “historical means stored event streams that are hours,
weeks or years old.”457 Schulte explains that the “analytical models can be used
for either of two purposes:
• “To design rules and algorithms to be used in real-time continuous
intelligence applications (see above), or
• “To make one-time, strategic, tactical and long-term operational
decisions.”457
The most common tool for on-demand analytics with historical data is a data
discovery product like Alteryx, Qlik, Tableau, SAS, Tibco, etc., etc., however,
“companies occasionally use ESP platforms to run analytics on historical event
streams by re-streaming the old event data through the ESP engine.” 457 “This is
particularly relevant when developing models for subsequent use in real-time,
continuous intelligence ESP applications.”457
ESP platforms are not the only type of software optimized for high performance
analytics on event stream data. Some stream analytics products like First
Derivatives KDB+, Interana Platform, Logtrust Platform, One Market Data
OneTick, Quartet ActivePivot, and Splunk Enterprise combine analytics and
longer term data storage in one product.457 “These products typically provide on-
demand, pull-based analytics, but some are also used for continuous, push-
based continuous intelligence. They ingest and store high volume event streams
very quickly, making the ‘at rest’ data immediately available for interactive
queries, exploration and visualization,” explains Schulte.457
For a real-time platform to work, data must be gathered from multiple and
disparate sources, which can include Enterprise Resource Planning (ERP),
Customer Relationship Management (CRM), Social CRM (SCRM) platforms,
geofencing applications (like Jiepang and Foursquare), Over-The-Top services

352
 THE A.I. CRUISE LINE

(like WhatsApp and WeChat), mobile apps, augmented reality apps, and other
mobile and social media systems. This data must be collected and then
seamlessly integrated into a data warehouse that can cleanse it and make it
ready for consumption.231 As the authors’ state in Mobile Advertising:
“The analytical system must have the capability to digest all the
user data, summarize it, and update the master user profile. This
functionality is essential to provide the rich user segmentation
that is at the heart of recommendations, campaign and offer
management, and advertisements. The segmentation engine
can cluster users into affinities and different groups based on
geographic, demographic or socio-economic, psychographic,
and behavioral characteristics.”38
In his article Real-Time Stream Processing as Game Changer in a Big Data World
with Hadoop and Data Warehouse36, Kal Wähner states that:
“Stream processing is designed to analyze and act on real-time
streaming data, using ‘continuous queries’ (i.e. SQL-type
queries that operate over time and buffer windows). Essential
to stream processing is Streaming Analytics, or the ability to
continuously calculate mathematical or statistical analytics on
the fly within the stream. Stream processing solutions are
designed to handle high volume in real time with a scalable,
highly available and fault tolerant architecture. This enables
analysis of data in motion.”
As a batch processing framework, Hadoop can’t handle the needs of real time
analytics. As the first open source distributed computing environment, Hadoop
has garnered a lot of attention recently, but it is not necessarily the best platform
for real-time analytics of dynamic information.458
One recent development in stream processing methods is the invention of the
“live data mart”, which “provides end-user, ad-hoc continuous query access to
this streaming data that’s aggregated in memory,” explains Wähner.36 “Business
user-oriented analytics tools access the data mart for a continuously live view of
streaming data”36 and a “live analytics front ends slices, dices, and aggregates
data dynamically in response to business users’ actions, and all in real time,”
adds Wähner36
For a cruise operator, streaming data could be coming in from facial recognition
and geo-location software, fraud or anti-money laundering solutions, slot and
table games systems, patron card on the casino floor and campaign management
databases, redemption systems, social media feeds, IoT data, as well as
wearables and employee/labor data sets.
Stream processing excels when data must be processed fast and/or

353
 ANDREW W. PEARSON

continuously. Many different frameworks and products are available on the

market already, however the number of mature solutions with good tools and
commercial support today is quite small.
Apache Storm is a good, open source framework, but it suffers from its open
source nature and custom coding is required because of limited developer tools.
The typical “commercial solution vs. open source” questions must be answered;
do I want a pre-built product that will require limited — and sometimes not so
limited implementation costs — or do I want to start with a solid solution and be
required to customize everything?
As Wähner explains, a stream processing solution has to solve several different
challenges, including36:
1. Processing massive amounts of streaming events (filter, aggregate, rule,
automate, predict, act, monitor, alert).
2. Real-time responsiveness to changing market conditions.
3. Performance and scalability as data volumes increase in size and
complexity.
4. Rapid integration with existing infrastructure and data sources: Input
(e.g. market data, user inputs, files, history data from a DWH) and
output (e.g. trades, email alerts, dashboards, automated reactions).
5. Fast time-to-market for application development and deployment due
to quickly changing landscape and requirements.
6. Developer productivity throughout all stages of the application
development lifecycle by offering good tool support and agile
development.
7. Analytics: Live data discovery and monitoring, continuous query
processing, automated alerts and reactions.
8. Community (component/connector exchange, education/discussion,
training/certification).
9. End-user ad-hoc continuous query access.
10. Alerting.
11. Push-based visualization.

Augmented and Virtual Reality

Not just the stuff of science fiction anymore, Augmented Reality (AR) is now a
part of our everyday lives. In his article CrowdOptic and L’Oreal to make history
by demonstrating how augmented reality can be a shared experience 459, Tarun
Wadhwa states that augmented reality works by “displaying layers of computer-
generated information on top of a view of the physical world.” It is “a technology
that alters the perception of reality by distorting it, allowing escape from it, and
enhancing it — all at the same time.”459

354
 THE A.I. CRUISE LINE

Webopedia.com adds that augmented reality or AR is460:

“A type of virtual reality that aims to duplicate the world's
environment in a computer. An augmented reality system
generates a composite view for the user that is the combination
of the real scene viewed by the user and a virtual scene
generated by the computer that augments the scene with
additional information. The virtual scene generated by the
computer is designed to enhance the user's sensory perception
of the virtual world they are seeing or interacting with. The goal
of Augmented Reality is to create a system in which the user
cannot tell the difference between the real world and the virtual
augmentation of it. Today Augmented Reality is used in
entertainment, military training, engineering design, robotics,
manufacturing and other industries.”
According to Gartner’s Top 10 Strategic Technology Trends 2017 461, Augmented
reality (AR) and virtual reality (VR) will “transform the way individuals interact
with each other and with software systems creating an immersive
environment. For example, VR can be used for training scenarios and remote
experiences.”
AR enables a blending of the real and virtual worlds, which “means businesses
can overlay graphics onto real-world objects.”461 Immersive experiences with AR
and VR are reaching tipping points in terms of price and capability but will not
replace other interface models.”461 In the future, AR and VR are expected to
expand beyond visual immersion and they might include all of the human
senses461, although this is a very complicated thing to pull off as smell-o-vision
tried and failed to do in the entertainment business last century.
According to its press release Gartner Says Augmented Reality Will Become an
Important Workplace Tool462, “Augmented reality is the real-time use of
information in the form of text, graphics, audio and other virtual enhancements
integrated with real-world objects.” Tuong Huy Nguyen, principal research
analyst at Gartner, states that “AR leverages and optimizes the use of other
technologies such as mobility, location, 3D content management and imaging
and recognition. It is especially useful in the mobile environment because it
enhances the user's senses via digital instruments to allow faster responses or
decision-making.”462
Gartner believes “AR technology has matured to a point where organizations can
use it as an internal tool to complement and enhance business processes,
workflows and employee training.”462 Gartner contends that “AR facilitates
business innovation by enabling real-time decision-making through virtual
prototyping and visualization of content.”462
According to Deloitte, Wearable AR devices can “allow users to access

355
 ANDREW W. PEARSON

standardized sets of instructions for a particular task in real time, triggered by

environmental factors and overlaid on the user’s field of vision.” 463 Research has
shown that overlaying 3D instructions over a real-life process can reduce the
error rate for an assembly task by 82 percent, with a particularly strong impact
on cumulative errors due to previous assembly mistakes. 463
“AR allows for improved senses and memory through the capture and
enhancement of the user’s perspective. By recording video/audio, capturing
images and removing elements that obscure the senses, AR technology allows
users’ eyes to act as cameras, and can enhance the senses in ways not available
naturally, such as night vision or the ability to zoom in on far-away objects,” says
Deloitte.463
AR uses location-based data for navigation, overlaying digital maps and
directions on real-world environments.462 Through the lens of an AR device, a
user can receive visual guidance based on GPS technology. 462 AR services
generally fall into one of two categories — “location-based or computer vision.
Location-based offerings use a device's motion sensors to provide information
based on a user's location. Computer-vision-based services use facial, object and
motion tracking algorithms to identify images and objects.”463
Mr. Nguyen claims AR’s benefits include the “potential to improve productivity,
provide hands-on experience, simplify current processes, increase available
information, provide real-time access to data, offer new ways to visualize
problems and solutions, and enhance collaboration.” 463
Augmented reality has many potential applications in the cruise industry as well
and the following ideas might seem a little like science fiction, but they are
certainly within the realm of technical possibilities, and today there is no
question that they would take the concept of personalization to a whole new
level.
In his article Augmented Reality and Hospitality…the Next Generation of
Hotels?464, Matt S-J lays out a very interesting scenario for AR in a hospitality
environment, whether that is for a hotel, a cruise line, or an integrated resort. If
a cruise line provided its front desk staff with a pair of AR glasses that connected
to an EDW that provided real-time patron information, the staff would be
empowered to greet and interact with a patron on a truly personal level. The
clerks could know all of the customer’s past history and, perhaps even if these
were well-known VIPs, the recent news headlines associated with them. This
type of engagement would bring the concept of customer service to a whole new
level, a level that would be unlike anything these patrons had ever witnessed
before, even if they were high-level celebrities.464
A guest who had taken a cruise in the past would immediately be identified and
all of his or her preferences and necessary patron information could appear on
the AR glasses’ virtual screen.464 “The guest could be checked in before they even

356
 THE A.I. CRUISE LINE

reached the door. The extent goes further as restaurants could identify guests’
allergies or preferences, orders would be recognized by dish then linked to the
table and guest images shown to see who has ordered what, so the food would
always be served to the correct person.”464
Birthday or anniversary greetings could be offered up without having to research
a patron’s profile.464 Staff who interact with VIP guests could be informed of
sensitive topics to avoid.464 Many of these things can be achieved through
excellent staff, but they all require research, time and a good long memory,
which not everyone possesses.464
For the cruise patron, AR could enhance his or her onboard experience
considerably. By simply downloading the cruise line’s AR app onto his mobile
phone, the patron could be checked in virtually and then be given personalized
directions to his room, where cruise line staff members could greet him. A free
bottle of champagne or Chateau Laffite wine could be awaiting him in his suite. 464
The cruise line’s general manager could even appear in a video to offer a
personalized greeting on the television.464
Continuing with the AR journey, a patron could go to one of the integrated
resort’s restaurants and, when seeing an appetizing meal being brought out from
the kitchen, he could whip out his mobile phone, snap a picture of the meal,
quickly scan it on the app, and then discover that it is a dish of beef wellington,
and then, potentially, place an order for it.464 If interested, the patron could even
pay for the dish on his mobile device, possibly utilizing patron points, should he
chose to redeem those.
The AR app could also help with cruise line maintenance. As a user scans his or
her room, the app could take notice of any minor maintenance issues. 464 These
issues would not be highlighted for the user, but would be relayed to the
appropriate maintenance departments so that they could be fixed. 464 This, of
course, does raise privacy issues, but they are probably nothing a good corporate
lawyer’s legalese couldn’t overcome.
So where is AR going? In his article Augmented reality: expanding the user
experience465, John Moore claims that “app creators have begun to engage more
of a mobile device’s sensors–accelerometers and gyroscopes, for example.
Augmented reality apps that use detailed animations are also in the works. The
objective: inject augmented reality technology in a wider range of apps to boost
the user experience.”459
Pokémon Go was the first location-based augmented reality game that hit it big.
Despite mixed reviews, the mobile app quickly became a global phenomenon
and it was one of the most used and profitable mobile apps of 2016, having been
downloaded more than 500 million times worldwide.466 It certainly revealed the
enormous potential of AR and it proved, without a doubt, that the barriers to AR
technology were limited and easily scaled by humans, at least those seeking out

357
 ANDREW W. PEARSON

little dueling pocket monsters.

Internet of Things (IoT)

As previously mentioned the Internet of Things is “the network of physical
objects that contain embedded technology to communicate and sense or
interact with their internal states or the external environment.” 29 Technology
costs are down, broadband’s price has dropped, while its availability has
increased. There is a proliferation of devices with Wi-Fi capabilities and censors
built into them, and smart phone penetration is exploding; all of these individual
technological advances were good for the IoT, together they have created a
perfect storm for it.467 Simply put, IoT is the concept of basically connecting any
device with an on and off switch to the Internet, including cell phones, coffee
makers, washing machines, headphones, lamps, wearable devices and almost
anything else imaginable.29
According to its article Gartner Says the Internet of Things Installed Base Will
Grow to 26 Billion Units By 2020468, Gartner claims that, “The Internet of Things
(IoT), which excludes PCs, tablets and smartphones, will grow to 26 billion units
installed in 2020 representing an almost 30-fold increase from 0.9 billion in
2009.” Gartner predicts IoT product and service suppliers will generate
incremental revenue exceeding $300 billion, mostly in services, in 2020. It will
result in $1.9 trillion in global economic value-add through sales into diverse end
markets.”468
Today, it is almost impossible to understand how all-encompassing the Internet
of Things will be in our daily lives in the not-too-distant future. From such life-
changing technology as Google’s driverless cars — which could help optimize
traffic, thereby reducing traffic congestion, as well as making people more
productive — sensors that can help regulate room temperature thereby saving
energy, IoT is definitely here to stay.
“The growth in IoT will far exceed that of other connected devices. By 2020, the
number of smartphones tablets and PCs in use will reach about 7.3 billion units,”
said Peter Middleton, research director at Gartner 468 “In contrast, the IoT will
have expanded at a much faster rate, resulting in a population of about 26 billion
units,” Middleton adds.468
For a cruise line, IoT applications can be utilized in the following ways:
• Internet of things:
o Smart lighting
o Waste management
• Smart Metering:
o Smart grid
o Tank level

358
 THE A.I. CRUISE LINE

o Water flow
o Silos stock calculation
o Water leakages
• Security & Emergencies:
o Perimeter access control
o Liquid presence
o Explosive and hazardous gases alerts
• Retail:
o Supply chain control
o NFC payment
o Intelligent shopping applications
o Smart product management
• Inventory optimization
• Logistics:
o Quality of shipment conditions
o Item location
o Storage incompatibility detection
o Fleet tracking
• Industrial Control:
o Smart Warehouse
o M2M applications
o Indoor air quality
o Temperature monitoring
o Indoor location tracking
o Vehicle auto-diagnosis
• Video analytics:
o Object detection
o Slip fall analysis
o People counting
• Swimming pool remote measurement
IoT isn’t a standalone technology and when combined with wearable technology
that is equipped with AR, personalized interactions with the physical world can
be created, which I discuss next.
IoT is faced with the typical problems of new technologies, a lack of standards as
the big and small players jockey for position, although there is a movement in
place to create a vendor-independent protocol that will allow devices to connect
with each other under the guise of a common service layer.
Currently, IoT’s growing pains are being tackled and security issues are being
addressed. The addition of edge analytics, which can reduce network and
connectivity costs, is also circumventing the need for cloud integration.
Improving computer processing power and memory in semiconductors and

359
 ANDREW W. PEARSON

modules increases by the month. This should allow IoT devices to add an ML
component, which could help IoT devices realize the potential of ambient
intelligence, allowing them to grow smarter over time. However, heavy duty
number crunching power is still needed when performing intensive predictive
and prescriptive analytics.
In his article The Data of Things: How Edge Analytics and IoT Go Hand In Hand469,
Gadi Lenz explains that, although IoT data has similar characteristics to Big Data,
it is much more complicated. IoT data is:
• “Messy, noisy, and sometimes intermittent because sensors are often
deployed in the field. IoT data is ultimately collected by sensors sitting
somewhere — for example, a sensor could be deployed on a telephone
pole or streetlight. Sensors often cut in and out.
• Often highly unstructured and sourced from a variety of sensors (fixed
and mobile).
• Dynamic — ‘data in motion’ as opposed to the traditional ‘data at rest’.
• Sometimes indirect — we cannot measure a certain relevant quantity
directly, for example, using a video camera with video analytics to count
people in a certain area.”469
The idea of collecting all of this sensor information and bringing it into one
centralized computing station is not viable over the long term, particularly as the
volume of IoT devices increases exponentially.469 “Bringing such a large amount
of data into a relatively small number of data centers where it is then analyzed
in the cloud, simply [sic] not scale.”469 The cost, too, would be prohibitive.469
“With so many devices producing so much data, a correspondingly large array of
analytics, compute, storage and networking power and infrastructure is
essential. Though analytics will be necessary to the growth and business value of
IoT, the traditional approach to analytics won’t be the right fit,” argues Lenz.469
Edge analytics addresses these problems. A cruise operator can “harness the
smartness of the myriad of smart devices and their low cost computational
power to allow them to run valuable analytics on the device itself.” 469 As Lenz
explains, “Multiple devices are usually connected to a local gateway where
potentially more compute power is available (like Cisco’s IOx), enabling more
complex multi-device analytics close to the edge.”469
Distributed IoT analytics would work in three ways, “simple” analytics would be
done on the smart device itself, more complex multi-device analytics on the IoT
gateways, and finally the high computational computing — the Big Data
analytics, if you will — would connect to and run on the cloud.469 “This
distribution of analytics offloads the network and the data centers by creating a
model that scales. Distributing the analytics to the edge is the only way to
progress,” advises Lenz.469

360
 THE A.I. CRUISE LINE

As the DHL Trend Research and Cisco Consulting Services paper Internet of
Things in Logistics470 explains:
“With the advent of IoT, Internet connections now extend to
physical objects that are not computers in the classic sense
and, in fact, serve a multiplicity of other purposes. A shoe, for
example, is designed to cushion the foot while walking or
running. A streetlight illuminates a road or sidewalk. A forklift
is used to move pallets or other heavy items. None of these
have traditionally been connected to the Internet — they did
not send, receive, process or store information. Nonetheless,
there is information latent in all of these items and their use.
When we connect the unconnected — when we light up “dark
assets” — vast amounts of information emerge, along with
potential new insights and business value.”
A connected shoe can reveal the number of footfalls in a given period of time, or
the force with which the foot strikes the ground.470 A connected street light can
understand traffic patterns, and “provide information to drivers or city officials
for route planning and to optimize the flow of traffic.” 470 A connected forklift can
be fitted with predictive asset maintenance alerts that can warn a warehouse
manager of an impending mechanical problem.470
In The A.I. Cruise Line, cameras immediately pick up a customer once he or she
enters or even when he or she boards a cruise line bus at the China-Macau
border gate. In this case, it makes sense for the analytics to be done inside the
camera itself, rather than having the data sent back to a centralized server as
that can be both inefficient and it risks bottlenecking. 469 Lenz adds, “Edge
analytics is all about processing and analyzing subsets of all the data collected
and then only transmitting the results.” 469 So, the systems is essentially
discarding some of the raw data and potentially missing some insights, but it
should be a calculated loss as analyzing everything is just not productive in most
cases.469
“Some organizations may never be willing to lose any data, but the vast majority
can accept that not everything can be analyzed. This is where we will have to
learn by experience as organizations begin to get involved in this new field of IoT
analytics and review the results,” adds Lenz.469 Overall, the potential for cruise
line to understand its business processes is enormous.
However, some trade-off must be considered with edge analytics. Lenz notes
that, “Edge analytics is all about processing and analyzing subsets of all the data
collected and then only transmitting the results.” 469 Some of the raw data is
discarded and potentially some insights are lost.469 “The question is, Can we live
with this ’loss’ and, if so, how should we choose which pieces we are willing to
‘discard’ and which need to be kept and analyzed?”469

361
 ANDREW W. PEARSON

It is also important to learn the lessons of past distributed systems. “For

example, when many devices are analyzing and acting on the edge, it may be
important to have somewhere a single ‘up-to-date view,’ which in turn, may
impose various constraints. The fact that many of the edge devices are also
mobile complicates the situation even more.” 469 Although incredibly powerful
devices in their own right, mobile phones and tablets will never reach the
capacity and compute technology of EDWs.

Wearables
Wearable products include smart watches, activity trackers, smart jewelry, head-
mounted optical displays, and earbuds. According to wearabledevices.com471:
“The terms ‘wearable technology’, ‘wearable devices’, and
‘wearables’ all refer to electronic technologies or computers
that are incorporated into items of clothing and accessories
which can comfortably be worn on the body. These wearable
devices can perform many of the same computing tasks as
mobile phones and laptop computers; however, in some cases,
wearable technology can outperform these hand-held devices
entirely. Wearable technology tends to be more sophisticated
than hand-held technology on the market today because it can
provide sensory and scanning features not typically seen in
mobile and laptop devices, such as biofeedback and tracking of
physiological functions.”
In general, wearable technology includes some form of communications
capability that allows the wearer to access real-time information.471 “Data-input
capabilities are also a feature of such devices, as is local storage. Examples of
wearable devices include watches, glasses, contact lenses, e-textiles and smart
fabrics, headbands, beanies and caps, jewelry such as rings, bracelets, and
hearing aid-like devices that are designed to look like earrings.”471
Wearable technology isn’t just for items that can be put on and taken off with
ease, there are also more invasive and permanent versions of the concept as
implanted devices such as micro-chips or even smart tattoos can be considered
wearables. Ultimately, whether a device is worn on or incorporated into the
body, “the purpose of wearable technology is to create constant, convenient,
seamless, portable, and mostly hands-free access to electronics and
computers.”471
In its Adoption of IoT for Warehouse Management472, Israel Gogle argues that
wearable devices and augmented reality are some of the best technologies to
help improve the performance of human operators.
John Bermudez, VP of Product Management at Infor, explains that, “In our
innovation lab, we are looking into options like providing workers with wearable

362
 THE A.I. CRUISE LINE

video cameras that can upload information to the warehouse management

system or with smart glasses. This augmented reality solution will give the
operator a visual confirmation on a small screen in front of his eye that he is
picking the right thing.”472
“Wearable devices will add a new layer of visibility that does not exist now. It will
work in route management, showing wearers where to go via the glasses, and
pick and pack verification, where bar code scans or RFID readings in real time can
be used to ensure correct pick and order management,” added Douglas Bellin,
Global Lead for Manufacturing and Energy Industries at Cisco Systems.472
Another advantage of wearables is their ability to collect information that can
also be used for such things as employee safety.472 This has a doubly positive
effect, not only does it ensure the employees’ wellbeing, but it can also save the
company money, minimizing losses due to injured employees and lost
productivity, maybe even saving lives.472
“Our wearables platform serves as a real-time warning system. It analyzes a vast
amount of information gathered from wearable sensors embedded in personal
protective equipment, such as smart safety helmets and protective vests, and in
the workers’ individual smartphones,” said Asaf Adi, Senior Manager of IoT and
Wearables at IBM Research.472 “Information from the sensors and smart
protective equipment feeds directly to the worker’s smartphone, which can then
immediately process and analyze the personal data,” explains Gogle. 472
By tracking a worker’s pulse rate, his movement, his body temperature, even,
potentially, his hydration level, sensors can continuously monitor a worker’s
physical condition.472 The noise level and/or an employees’ location in relation
to moving machinery and forklifts can also be monitored to guard against
accidents.472 Alerts can also be sent out in cases where sensors detect a worker
that has fallen or fainted in the warehouse. 472

Robotics
In their paper Progress on robotics in hospitality and tourism: a review of the
literature473, Ivanov et al. believe that, “The concept of the robot is not
particularly old, only being coined in 1920 by Karel Čapek in his play R.U.R—
Rossum’s Universal Robots474, and it took several decades before the concept
was incorporated fully into popular culture.”
By the 1950s, Hollywood movies depicting both malevolent and benevolent
robots broadly disseminated the concept of the robot and inspired development
of physical robots and the robotics industry.473 While Ivanov et al. admit that
there might be a colloquial understanding of what a robot is, there is also a more
technical and industry-accepted definition473, which is, “A robot is defined by the
International Organization for Standardization475 as an “actuated mechanism

363
 ANDREW W. PEARSON

programmable in two or more axes with a degree of autonomy, moving within

its environment, to perform intended tasks.”473
According to Ivanov et al., “The incorporation of robotics came relatively late to
the industries involved in travel, tourism and hospitality, probably since many of
the services provided require sophisticated reactions to the needs of the
customer.”473 “While some automobile factories were largely staffed by robots
by the mid-1990s, it was only in 2015 that a hotel predominantly staffed by
robots opened (the Henn na Hotel in Japan)” add Ivanov et al. 473 “At present,
robots are used in hotels for such tasks as checking guests in, vacuuming floors,
delivering things to guests, concierge services, and other common chores.”473
Robots also prepare drinks, entertain guests, guide guests through a property
and offer guests information.476
In his Ultimate transformation: How will automation technologies disrupt the
travel, tourism and hospitality industries? 477, Stanislav Ivanov argues that
“Companies from the travel, tourism and hospitality industry have started
adopting robots, artificial intelligence and service automation technologies
(RAISA) in their operations.”477 “Self-check-in kiosks, robotic pool cleaners,
delivery robots, robot concierges, chatbots, etc., are used increasingly by
tourism companies and transform the ways they create and deliver services.” 477
Ivanov believes that “in the future companies will divide into two large segments
– high-tech tourism companies offering standardized cheap robot-delivered
services, and high-touch companies, which rely on human employees.” 477
Every generation has witnessed technological breakthroughs that revolutionized
the tourism industry.477 In the middle of the 19th century, train travel allowed
tourists a feasible way to visit far off lands and the first organized trips
emerged.477 “In the first half of the 20th century, the car added a great deal of
flexibility to the selection of tourist routes and contributed to the
democratisation of travel,” says Stanislav Ivanov. 477 The introduction of the jet
engine in the second half of the 20th century made long haul destinations
accessible and affordable for the working middle class.477 At the end of the 20th
and beginning of the 21st century, the Internet and social media contracted the
world and changed the rules of the business of and competition among travel,
tourism and hospitality companies.477 Today, the tourism industry faces “a new
revolution, one more powerful, transformative and with longer term
implications than the previous changes. Tourism is entering the robotics era.”477
According to Ivanov, “While during the previous tourism revolutions humans
played a central role in the system, both as customers/tourists and as service
providers/employees in tourism companies, the new realities would induce
changes to both the demand and the supply sides of the tourism system.” 477 “The
advances in robotics, artificial intelligence and service automation technologies
(RAISA)478 479 480 allow companies from various sectors to use RAISA in order to
decrease costs, streamline operations, eliminate waste, and improve

364
 THE A.I. CRUISE LINE

productivity and efficiency,” says Ivanov.477 Robots already greet guests at hotel
receptions, act as waiters in restaurants and room service, clean floors and
swimming pools, cut grass at hotels’ green areas, provide information at airports
kiosks, and even cook food in automated kitchens.481 482
Currently, Ivanov believes that, “robots seem quite clumsy in their interactions
with humans and navigation through the premises of hotels, restaurants and
airports.”477 However, with advances in robotics, robots should “become much
more capable of serving guests and implementing various tasks beyond the 3D
(dirty, dull and dangerous) tasks which human beings do not want to implement,
hence widening their potential application and adoption by tourist
companies.”477
Researchers have applied AI in forecasting tourism arrivals, demand, and
expenditure483 484 485 as well as overall hotel occupancy486, identification of
destination attributes487, sentiment analysis of online reviews 488 489, analysis of
the impact of online reviews on hotel performance490, analysis of hotel employee
satisfaction491, market segmentation492, forecasting waste generation rates in
hotels493, or their energy demand494, to name just a few areas of interest in the
field. The application of facial recognition systems at airports has been studied
by Del Rio et al.495, while Nica et al.496 developed a chatbot for recommendations
in tourism.
RAISA technologies can work 24/7, 365 days a year.477 “Unlike human employees
who can work 40-60 hours a week depending on their job position, legal
regulations, health and mental condition and work urgency, RAISA technologies
are available 24/7, 168 hours a week,” says Ivanov. 477 “A hotel would need at
least 5 receptionists to have a non-stop working reception, something that could
be achieved with one kiosk only.”477
“RAISA technologies could implement various tasks and expand their scope with
software and hardware upgrades,” adds Ivanov. 477 A kiosk only needs a new
software package or upgrade in order to be used for an alternate purpose from
its original design.477 “A new block answers and set of rules would allow a
chatbot to be able to provide relevant answers to customer queries,” explains
Ivanov.477 On the other hand, human employees need training that could be both
significant and time consuming depending on the new tasks they have to
implement.477
RAISA technologies could do routine work repeatedly. 477 “Unlike human
employees, robots and kiosks can perform the same task numerous times,
without complaints and they do not need to be motivated to do it,” says
Ivanov.477 RAISA technologies don’t strike, spread rumors, or discriminate
against customers or employees, quit their job without notice, show negative
emotions, shirk from work, ask for pay increases, or get ill.477 This makes the life
of a company managers much easier. In fact, it could be argued that “every strike

365
 ANDREW W. PEARSON

of human employees is a step towards the automation of tasks they currently

perform.”477
On the other hand, RAISA technologies lack creativity.477 “At the current state of
technological development service robots, kiosks or chatbots cannot invent new
ways to deliver services to customers, regardless whether the innovation is in
the form of a new service or a new way to deliver an existing service,” says
Ivanov.477
RAISA technologies won’t be completely independent of human supervision any
time soon.477 RAISA technologies lack the human, personal touch so necessary in
the hospitality industry.477 Kiosks follow a service protocol to script, without ever
deviating from it because of unique requests. 477 “Service robots and chatbots
can adapt their responses on the basis of the specific interactions they have with
humans (customers or employees),” but they have no independent thought. 477
“Chatbots and robots can provide answers to questions that include specific key
words, which they recognise in order to activate a particular predetermined set
of answers.”477 Robot navigation through a hotel or IR property also requires
navigation markers.497
People understandably think robots will threaten their jobs.477 A recent study by
Frey & Osborne498 revealed that 47% of jobs in the US could be lost due to
computerization. The travel, tourism and hospitality industry should consider
and address these very real and understandable fears.477
RAISA can “change the way companies operate and influence all the aspects of
their value chain – marketing, operations, facilities design and maintenance,
human resource management, financial management and supply chain
management.”477 Additionally, the use of RAISA technologies will influence how
potential customers perceive a company, its brand, and the quality of its
services, as well as how a company competes or cooperates with other
companies.477
According to Ivanov, “One of the most important impacts of RAISA technologies
on travel, tourism and hospitality companies is on the design of their
premises.”477 Ivanov adds that, “In the future, the premises of hotels, restaurants
and airports, among others, would be used by a wide variety of mobile robots
(wheeled, legged, flying or underwater) such as security robots, robot guides,
robot waiters, companion/sex robots, pet robots, room service deliver robots,
robotic vacuum cleaners / lawnmowers / pool cleaners, entertainment robots,
delivery drones, etc.”477 It doesn’t matter if “these mobile robots belong to the
customers or to the hospitality companies, the latter would need to ensure the
robot-friendliness / robot-inclusiveness of their premises.”497
In their paper Toward a framework for robot-inclusive environments499, Tan,
Mohan & Watanabe define “robot-inclusiveness of an environment in which a
robot operates as how much the design of the environment takes into account

366
 THE A.I. CRUISE LINE

the robot therein, i.e. whether it helps the robot fulfil its tasks.” They add, “It
depends on the design of the premises where the robot needs to operate in,
their cleanliness, tidiness, signage, lightning, noise, physical barriers (e.g. doors,
doorsteps, stairs), presence of people and dynamic of the environment,
presence/lack of predetermined routes for robot movement, presence/lack of
(artificial) landmarks and sensors to help robot navigation, etc.”499 “When the
hospitality premises are more robot-inclusive, the same task can be performed
by a less intelligent robot and vice versa: an environment that is not robot-
inclusive would require a more intelligent robot to navigate through it,” say
Ivanov and Webster.497 Therefore, a hospitality company that invests in the
robot-friendliness of its facilities, can use less intelligent (and cheaper!) robots
operationally, which means that more types of customer-owned robots would
be able to use them.477 Therefore, the robot-friendliness of hospitality facilities
could actually be a new competitive advantage for travel, tourism and hospitality
companies.477
The adoption of RAISA will unquestionably have both positive and negative
implications for human resources at travel, tourism and hospitality
companies.477 On the plus side, employees won’t have to perform tedious and
repetitive tasks, instead focusing on more creative and revenue generating
activities.477 For this reason, RAISA technologies are actually enhancing rather
than replacing employees lives.477 “Human employees improve their
productivity and can implement more activities/tasks for the same amount of
time,” says Ivanov.477 Additionally, as Ivanov points out, “RAISA would trigger
changes in the required skills of employees – they would need more technical
skills in order to be able to operate with and maintain the new technology, and
communication and social skills to interact with the customers.” 477
“RAISA would solve some of the problems with hiring and firing of employees,
especially seasonal personnel and immigrants – robots and kiosks can be
leased/turned on during periods of high demand, and returned/turned off
during off-season without the need to deal with the bureaucracies of labour
laws,” notes Ivanov.477 “Moreover, due to the high employee turnover, travel,
tourism and hospitality companies need to constantly train employees, which
takes both time and money.”477 The cruise line industry is often seen as a sin
industry as well, so keeping employees is more important in this industry than
the general tourism and travel industry as a whole.
Ivanov points out that, “On the negative side, the use of RAISA technologies leads
to changes in the number of employees in various departments, usually
decreasing the number of employees whose tasks can easily be automated (e.g.
reception, information provision, cleaning of common areas, menu ordering and
food delivery, etc.).”477 Furthermore, RAISA have the potential to eliminate or at
least significantly decrease the number of entry-level jobs available.477 This will
significantly hinder the job opportunities for some of society’s most vulnerable

367
 ANDREW W. PEARSON

socio-economic groups, i.e., people with low education or students looking for
part-time jobs.477 In extreme cases, there has been an emergence of zero-
employee properties (or automated hotels/motels/hostels) in Japan, Korea,
Bulgaria and other countries.500 501 Of course, the wheels, or perhaps the gears,
of progress can’t be held back, but it is only natural that these changes concern
human employees, who may perceive RAISA technologies as a threat for their
jobs and therefore resist their usage.477
From a financial perspective, RAISA technologies can reduce labor costs.
Additionally, “RAISA can lead to increased sales – some customers, for example,
may make an order for room service out of curiosity in seeing the robot
delivering the order.”477 Admittedly, a successful, ongoing operation can’t be
built on the back of one-time curiosity purchases, but RAISA technologies are
associated with other expenses, including502:
• Acquisition costs – e.g. for purchasing a robot or kiosk, for purchasing a
chatbot/payment for its development;
• Installation costs – for a robot or kiosk. It might be virtually zero for a
chatbot;
• Maintenance costs – they include electricity consumption of the
robot/kiosk, spare parts, periodic maintenance, repair works, etc., but
could be very low for a chatbot;
• Software update costs;
• Costs for adapting the premises to facilitate mobility for robots – e.g.
removing any barriers for robot’s movement within a hotel;
• Costs for hiring specialists to operate and maintain the
robots/kiosks/chatbots;
• Costs for training human employees to use them;
• Insurance costs for the robots/kiosks, insurance for damages caused by
a robot, etc.
“The adoption of RAISA technologies allows the integration of the information
systems of suppliers and travel, tourism and hospitality companies,” explains
Ivanov.477 This started at the beginning of the century, when tourism websites
introduced back-to-back xml connections.477 “These connections allow, for
example, the inventory of hotels, rooms and their availability from one website
to be visualised in another website. When a customer makes a booking in the
second website, the booking goes directly to the first website without the need
for human intervention.”477 This integration, however, was mostly on the
website level.477 Today’s RAISA technologies allow much deeper integration –
“for example a booking made by a customer through a travel chatbot of one
company (e.g. a tour operator or online travel agency) could be automatically
registered into the booking system of that company’s supplier (e.g. hotel
chain).”477

368
 THE A.I. CRUISE LINE

“RAISA allows the implementation of automated pricing of products based on

sets of rules and real-time data on buyer behaviour of customers,” explains
Ivanov.477 At the extreme end, AI allows for personalised pricing, i.e., separate
pricing for every single customer based on his/her willingness to pay, which
should lead to revenue maximization.477 Ivanov contends that, “The dropping
cost of RAISA solutions compared to human employees and the increases in their
productivity mean that we can expect lower prices for mass ‘high-tech’
hospitality products (e.g. automated hostels and restaurant) and higher prices
for exclusive ‘high-touch’ products delivered by human employees.”477
RAISA will allow “automated communications with customers via chatbots, voice
assistants, kiosks, robots.”477 “On the one hand, this means that companies could
communicate with more customers, and do this in numerous languages and at
any time,” says Ivanov.477 “On the other hand, the lack of human involvement in
these automated communications may frustrate the customers, especially when
they do not receive answers to their specific questions and cannot contact a
human employee to help,” he concludes.477 Overriding it all, “a company that
adopts RAISA may boast positive word-of-mouth due to its image of an
innovative high-tech company.”477 However, as Ivanov & Webster warn, “it may
also suffer negative publicity and be perceived as a company that puts profits
before humans.”502 “Furthermore, customers may feel frustrated if they need to
communicate with a chatbot or robot when they have a complaint. Therefore,
even high-tech tourism/hospitality companies may need to keep human contact
with their customers, especially in emotionally charged service situations like
handling of complaints,” recommends Ivanov.477

383 Databricks. Conquer the AI Dilemma by Unifying Data Science and Engineering.
https://databricks.com/company/newsroom/press-releases/databricks-conquers-ai-
dilemma-with-unified-analytics (Accessed 8 October 2018).
384 IBM Software Thought Leadership White Paper. Descriptive, predictive, prescriptive:
Transforming asset and facilities management with analytics. October 2013.
https://static.ibmserviceengage.com/TIW14162USEN.PDF
385 Talend. Definitive Guide to Data Governance. Talend.com.
https://www.talend.com/resources/definitive-guide-data-governance/ (Accessed 18
August 2019).
386 https://en.wikipedia.org/wiki/Data_curation (Accessed 10 April 2019).
387 Crandall, R., Markland, R., 1996. Demand management: today’s challenge for service
industries. Production and Operations Management 5 (1), 106–120.
388 Taylor, G., 1980. How to match plant with demand: a matrix for marketing.
International Journal of Tourism Management (March), 56–60.

369
 ANDREW W. PEARSON

389 Klassen, K., Rohleder, T., 2002. Demand and capacity management decisions in
services: how they impact on one another. International Journal of Operations &
Production Management 5 (6), 527–548.
390 Bozarth, C., Handfield, R., 2006. Introduction to Operations and Supply Chain
Management. Pearson-Prentice Hall, Upper Saddle River, NJ.
391 Davis, M., Heineke, J., 2005. Operations Management: Integrating Manufacturing and
Services, fifth ed. McGraw-Hill/Irwin, New York.
392 Porter, M., 1996. What is strategy? Harvard Business Review (November–December),
61–78.
393 Schroeder, R., 2006. Operations Management: Contemporary Concepts and Cases,
third ed. McGraw-Hill/Irwin, New York.
394 Lawson, F., Baud-Bovy, M., 1977. Tourism and Recreational Development: A
Handbook of Physical Planning. The Architectural Press, Oxford, UK, pp. 183–186.
395 Mathieson, A., Wall, G., 1982. Tourism: Economic, Physical, and Social Impacts.
Longman, New York.
396 Schmenner, R., 1986. How can service businesses survive and prosper? Sloan
Management Review 27 (3), 21–32.
397 Wall, G., 1982. Cycles and capacity: incipient theory or conceptual contradiction?
Tourism Management (September), 188–192.
398 Saveriades, A., 2000. Establishing the social carrying capacity for the tourist resorts
of the east coast of the Republic of Cyprus. Tourism Management 21, 147–156.
399 Muller, C.C., 1999. A simple measure of restaurant efficiency. Cornell Hotel and
Restaurant Administration Quarterly 40 (June (3)), 31–37.
400 Kandampully, J., 2000. The impact of demand fluctuation on the quality of service: a
tourism industry example. Managing Service Quality 10 (1), 10–18.
401 Kilby, T., Fox, G., 1998. Retailer Operations Management. Wiley, New York.
402 Adenso-Dia´z, B., Gonza´les-Torre, P., Garcia, V., 2002. A capacity management model
in service industries. International Journal of Service Industry Management 13 (3), 286–
302.
403 Simón, F., Narangajavana, Y., Marque´s, D., 2004. Carrying capacity in the tourism
industry: a case studyofHengistbury Head. TourismManagement25 (2), 275–283.
404 Easton, F., Pullman, M., 2001. Optimizing service attributes: the seller’s utility
problem. Decision Sciences 32 (2), 251–275.
405 Sill, B., 2000. Capacity management: engineering the balance between customer
satisfaction, employee satisfaction and company profit. The Consultant, 2nd quarter, 2–
11.
406 Lawson, F., 1995. Hotels & Resorts: Planning, Design and Refurbishment. The
Architectural Press, Oxford, UK.
407 Roubi, S., Litteljohn, D., 2004. What makes hotel values in the UK? A hedonic
valuation model. International Journal of Contemporary Hospitality Management 16 (3),
175–181.
408 Brown, J., Dev, C., 1999. Looking beyond RevPAR: productivity consequences of hotel
strategies. Cornell Hotel and Restaurant Administration Quarterly 40 (2), 23–33.
409 Standards Australia, 1998. Construction and fit-out of food premises. AS/NZS 1234.
Sydney, Australia.
410 U.S. Food and Drug Administration [USDA], 2005. Equipment, utensils and linen. In:
Food code (Chapter 4).

370
 THE A.I. CRUISE LINE

411 Center for Disease Control and Prevention (US), 2001. Vessel sanitation program
construction guidelines. Retrieved June 6, 2008 from http://www.cdc.gov/nceh/
vsp/pub/
412 Kimes, S., McGuire, K., 2001. Function-space revenue management: a case study from
Singapore. Cornell Hotel and Restaurant Administration Quarterly (December), 33–46.
413 Kimes, S., Robson, S., 2004. The impact of restaurant table characteristics on meal
duration and spending. Cornell Hotel and Restaurant Administration Quarterly
(November), 333–346.
414 Robson, S., 1999. Turning the tables: the psychology of design for high-volume
restaurants. Cornell Hotel and Restaurant Administration Quarterly 40 (3), 56–63.
415 Wanhill, S., 2003. Economic aspects of developing theme park. In: Fyall, A., Garrod,
B., Leask, A. (Eds.), Managing Visitor Attractions: New Directions.
ButterworthHeinemann, Oxford, UK, pp. 39–57.
416 Hanlon, J.P., 1989. Hub operations and airline competition. Tourism Management 10
(2), 111–124.
417 Dickson, D., Ford, R., Laval, B., 2005. Managing real and virtual waits in hospitality
and service organizations. Cornell Hospitality Quarterly 46 (52), 52–68.
418 Brown, D.M., Hoover, L.W., 1990. Productivity management in food service: past
accomplishments—a future alternative. Journal of the American Dietetic Association 90
(7), 973–979.
419 Junko, D., 2004. Changes in social strata on the Cunard’s Queen Sisters. In:
Proceedings from 10th Asia Pacific Tourism Association Conference, Sasebo, Nagasaki,
Japan.
420 Jones, P., Lockwood, A., 2002. The Management of Hotel Operations. Continuum,
London.
421 Creed, P.G., 2001. The potential of foodservice systems for satisfying consumer
needs. Innovative Food Science & Emerging Technologies 2, 219–277.
422 Liu, J., Var, T., 1980. The use of lodging ratios in tourism. Annals of Tourism Research
7 (3), 406–427.
423 Carey, K., 1989. Tourism development in LDCs: hotel capacity expansion with
reference to Barbados. World Development 17 (1), 59–67.
424 Gu, Z., 2003. Analysis of Las Vegas strip cruise line hotel capacity: an inventory model
for optimisation. Tourism Management 24 (3), 309–314.
425 Thompson, G.M., 2002. Optimizing a restaurant’s seating capacity: use dedicated or
combinable tables? Cornell Hotel and Restaurant Administration Quarterly 43 (4), 48–57.
426 Berdichevskiy, V.C., Kapsekin, V.I., 1988. Design of Catering. High Education, Kiev,
Ukraine.
427 Pullman, M., Thompson, G., 2003. Strategies for integrating capacity with demand in
service networks. Journal of Service Research 5 (3), 169–183.
428 Schnitzler, H., 1982. Purchasing for a new hotel. Cornell Hospitality Quarterly 23 (2),
83–91.
429 Wie, B.W., 2005. A dynamic game model of strategic capacity investment in the cruise
line industry. Tourism Management 26 (2), 203–217.
430 Liu, J., Var, T., 1980. The use of lodging ratios in tourism. Annals of Tourism Research
7 (3), 406–427.
431 Verma, R., Plaschka, G., Louviere, J., 2002. Understanding customer choices: a key to
successful management of hospitality services. Cornell Hotel and Restaurant
Administration Quarterly 43 (6), 15–24.
371
 ANDREW W. PEARSON

432 Sasser, E., 1976. Match supply and demand in service industries. Harvard Business
Review 54 (6), 133–140.
433 Huete, L., Segarra, J., 1997. Port Aventura. IESE Case #697-054-1. University of
Navarra, Barcelona-Madrid.
434 Sill, B., 1991. Capacity management: making your service delivery more productive.
Cornell Hotel and Restaurant Administration Quarterly 31 (4), 76–88.
435 Bozarth, C., Handfield, R., 2006. Introduction to Operations and Supply Chain
Management. Pearson-Prentice Hall, Upper Saddle River, NJ.
436 Vollman, T., Berry, W., Whybark, D.C., Jacobs, F.R., 2005. Manufacturing Planning and
Control for Supply Chain Management. McGraw-Hill/Irwin, New York.
437 Lindenmeier, J., Tscheulin, D., 2008. The effects of inventory control and denied
boarding on customer satisfaction: the case of capacity-based airline revenue
management. Tourism Management 29 (1), 32–43.
438 Sanchez, J., Satir, A., 2005. Hotel yield management using different reservation
modes. International Journal of Contemporary Hospitality Management 17 (2), 136–146.
439 Kandampully, J., 2000. The impact of demand fluctuation on the quality of service: a
tourism industry example. Managing Service Quality 10 (1), 10–18.
440 Kimes, S., Barrash, D., Alexander, J., 1999. Developing a restaurant revenue-
management strategy. Cornell Hotel and Restaurant Administration Quarterly (October),
18–29.
441 Getz, D., Nilsson, P.A., 2004. Responses of family businesses to extreme seasonality
in demand: the case of Bornholm, Denmark. Tourism Management 25, 17–30.
442 Kimes, S., 2000. Revenue management on the links. Cornell Hotel and Restaurant
Administration Quarterly (February), 120–127.
443 Sundarapandian, V. (2009). “7. Queueing Theory”. Probability, Statistics and
Queueing Theory. PHI Learning. ISBN 8120338448.
444 Thompson, G.M., 1998a. Labour scheduling, part 1: forecasting demand. Cornell
Hotel and Restaurant Administration Quarterly 39 (5), 22–31.
445 Kleinrock, L., 1975. Queuing Systems: Theory, vol. 1. Wiley, New York.
446 Thompson, G.M., 1998b. Labour scheduling, part 2: knowing how many on-duty
employees to schedule. Cornell Hotel and Restaurant Administration Quarterly 39 (6), 26–
37.
447 Thompson, G.M., 1999a. Labour scheduling, part 3: developing a workforce schedule.
Cornell Hotel and Restaurant Administration Quarterly 40 (1), 86–96.
448 Thompson, G.M., 1999b. Labour scheduling, part 4: controlling workforce schedules
in real time. Cornell Hotel and Restaurant Administration Quarterly 40 (3), 85–96.
449 Hur, D., Mabert, V., Bretthauer, K., 2004. Real-time work schedule adjustment
decisions: an investigation and evaluation. Production and Operations Management
Journal 13 (4), 322–339.
450 Ponemon Institute. Cost of Data Center Outages January 2016 Data Center
Performance Benchmark Series. January 2016.
https://www.vertiv.com/globalassets/documents/reports/2016-cost-of-data-center-
outages-11-11_51190_1.pdf (Accessed 1 November 2019).
451 Sebestyén, Zoltán and Juhász, Viktor. (2003). The Impact Of The Cost Of Unused
Capacity On Production Planning Of Flexible Manufacturing Systems. Department of
Industrial Management and Business Economics Budapest University of Technology and
Economics H–1521 Budapest, Hungary. October 20, 2003.
https://www.researchgate.net/publication/254407614_The_impact_of_the_cost_of_un
372
 THE A.I. CRUISE LINE

used_capacity_on_production_planning_of_flexible_manufacturing_systems (Accessed
31 October 2019).
452 Cooper, R., Kaplan, R. S. (1992). Activity-Based System: Measuring the Cost of
Resource Usage. Accounting Horizons. September, 1992. pp. 1–13.
https://pdfs.semanticscholar.org/acc7/63b83b5b321c404730c782468db3bf71a4da.pdf
(Accessed 24 November 2019).
453 Copper, R., Kaplan, R. S. (1998). Cost & Effect: Using Integrated Cost Systems to Drive
Profitability and Performance, Harvard Business School Press, 1998.
454 Kololtai, T. Sebestyén, Z. (2002). New Directions of Developing a Production Planning
Model Based on the Cost of Unused Capacity Approach. Proceedings of the 12th
International Conference on Flexible Automation and Intelligent Manufacturing 2002,
Dresden, pp. 654– 663.
455 https://en.wikipedia.org/wiki/Real-time_marketing
456 Macy, B. a. (2011). The Power of Real-Time Social Media Marketing. New York:
McGraw Hill, 2011.
457 Schulte, Roy. (May 23, 2017). When do you need an Event Stream Processing
Platform? Logtrust.com. https://www.logtrust.com/need-event-stream-processing-
platform/ (accessed 30 August 2017).
458 Deng, Lei, Gao, Jerry, Vuppalapati, Chandrasekar. March 2015. Building a Big Data
Analytics Service Framework for Mobile Advertising and Marketing. Online:
https://www.researchgate.net/profile/Jerry_Gao/publication/273635443_Building_a_Bi
g_Data_Analytics_Service_Framework_for_Mobile_Advertising_and_Marketing/links/5
508de220cf26ff55f840c31.pdf
459 Wadhwa, T. (2013, June 3). CrowdOptic and L’Oreal to make history by demonstrating
how augmented reality can be a shared experience. Retrieved from Forbes.com:
http://www.forbes.com/sites/tarunwadhwa/2013/06/03/crowdoptic-and-loreal-are-
about-to-make-history-by-demonstrating-how-augmented-reality-can-be-a-shared-
experience/
460 http://www.webopedia.com/TERM/A/Augmented_Reality.html
461 Gartner. October 18, 2016. Gartner’s Top 10 Strategic Technology Trends for 2017.
http://www.gartner.com/smarterwithgartner/gartners-top-10-technology-trends-2017/
462 Gartner. 2014, January 14). Gartner believes augmented reality will become an
important workplace tool. Retrieved from Gartner.com:
http://www.gartner.com/newsroom/id/2649315
463 Deloitte. (2013). Augmented Government, Transforming Government Services
Through Augmented Reality. Retrieved from Deloitte Development LLC:
http://www.deloitte.com/assets/Dcom-
UnitedStates/Local%20Assets/Documents/Federal/us_fed_augmented_government_06
0613.pdf
464 S-J, M. (2013, January 22). Augmented reality and hospitality...the next generation of
hotels? Retrieved from Chocolate pillow: chocolatepillow.com/augmented-reality-and-
hospitality-the-next-generation-of-hotels/
465 Moore, J. (2012). Augmented reality: expanding the user experience. Retrieved from
Digital Innovation Gazette:
http://www.digitalinnovationgazette.com/uiux/augmented_reality_app_development/
#axzz2z8tFKN00
466 https://en.wikipedia.org/wiki/Pokémon_Go
467 Morgan, J. (2014, May 13). A Simple Explanation of 'the Internet of Things'. Retrieved
from Forbes.com: http://www.forbes 373
 ANDREW W. PEARSON

468 Gartner. (2013, December 12). Gartner Says the Internet of Things Installed Base Will
Grow to 26 Billion Units By 2020. Retrieved from Gartner.com:
http://www.gartner.com/newsroom/id/2636073
469 Lenz, Gadi. September 22, 2015. Datanami.com The Data of Things: How Edge
Analytics and IoT Go Hand In Hand. https://www.datanami.com/2015/09/22/the-data-
of-things-how-edge-analytics-and-iot-go-hand-in-hand/
470 DHT Trend Research, Cisco Consulting Services. 2015. Internet of Things in Logistics.
http://www.dpdhl.com/content/dam/dpdhl/presse/pdf/2015/DHLTrendReport_Interne
t_of_things.pdf
471 http://www.wearabledevices.com/what-is-a-wearable-device/
472 Gogle, Israel. Internet of Things. May 2016. Adoption of IoT for Warehouse
Management. A & S Internation. Asmag.com.
http://www.cisco.com/c/dam/en/us/solutions/collateral/industry-solutions/iot-
warehouse.pdf
473 Ivanov, S., Gretzel, U., Berezina, K., Sigala, M., & Webster, C. (2019) Progress on
robotics in hospitality and tourism: a review of the literature. Journal of Hospitality and
Tourism Technology (forthcoming).
474 NPR. (2011) Science Diction: The Origin of the Word 'Robot'. Retrieved from
https://www.npr.org/2011/04/22/135634400/science-diction-the-origin-of-the-word-
robot (Accessed 4 December 2019).
475 https://www.iso.org/home.html (Accessed 4 December 2019).
476 Ivanov, S., Webster, C. & Berezina, K. (2017). Adoption of robots and service
automation by tourism and hospitality companies. Revista Turismo & Desenvolvimento,
27/28, 1501-1517.
477 Ivanov, S. (2019). Ultimate transformation: How will automation technologies disrupt
the travel, tourism and hospitality industries? Zeitschrift für Tourismuswissenschaft 11(1),
(forthcoming).
478 Bhaumik, A. (2018). From AI to Robotics: Mobile, Social, and Sentient Robots. Boca
Raton, FL: CRC Press.
479 Miller, M. R., & Miller, R. (2017). Robots and Robotics: Principles, Systems, and
Industrial Applications. McGraw-Hill Education.
480 Russell, S. J., & Norvig, P. (2016). Artificial intelligence: a modern approach. Harlow:
Pearson Education Limited.
481 Ivanov, S. (2018). Tourism beyond humans – robots, pets and Teddy bears. Paper
presented at the International Scientific Conference “Tourism and Innovations”, 14-15th
September 2018, College of Tourism – Varna, Varna, Bulgaria.
482 Webster, C., & Ivanov, S. (2019). Robotics, artificial intelligence, and the evolving
nature of work. In George, B., & Paul, J. (Eds.). Business Transformation in Data Driven
Societies, Palgrave-MacMillan (forthcoming).
483 Law, R. (2000). Back-propagation learning in improving the accuracy of neural
network-based tourism demand forecasting. Tourism Management, 21(4), 331-340.
484 Palmer, A., Montano, J. J., & Sesé, A. (2006). Designing an artificial neural network for
forecasting tourism time series. Tourism Management, 27(5), 781-790.
485 Sun, S., Wei, Y., Tsui, K. L., & Wang, S. (2019). Forecasting tourist arrivals with machine
learning and internet search index. Tourism Management, 70, 1-10.
486 Law, R. (1998). Room occupancy rate forecasting: a neural network approach.
International Journal of Contemporary Hospitality Management, 10(6), 234-239.

374
 THE A.I. CRUISE LINE

487 Santos Silva, M., Albayrak, T., Caber, M., & Moutinho, L. (2016). Key destination
attributes of behavioural intention: An application of neural networks. European Journal
of Tourism Research 14, 16-28.
488 Kirilenko, A. P., Stepchenkova, S. O., Kim, H., & Li, X. (2018). Automated sentiment
analysis in tourism: Comparison of approaches. Journal of Travel Research, 57(8), 1012-
1025.
489 Ye, Q., Zhang, Z., & Law, R. (2009). Sentiment classification of online reviews to travel
destinations by supervised machine learning approaches. Expert Systems with
Applications, 36(3 Part 2), 6527-6535.
490 Phillips, P., Zigan, K., Silva, M. M. S., & Schegg, R. (2015). The interactive effects of
online reviews on the determinants of Swiss hotel performance: A neural network
analysis. Tourism Management, 50, 130-141.
491 Tian, X., & Pu, Y. (2008). An artificial neural network approach to hotel employee
satisfaction: The case of China. Social Behavior and Personality: an international journal,
36(4), 467-482.
492 Kim, J., Wei, S., & Ruys, H. (2003). Segmenting the market of West Australian senior
tourists using an artificial neural network. Tourism Management, 24(1), 25-34.
493 Azarmi, S. L., Alipour, H., & Oladipo, A. A. (2017). Using artificial neural network and
desirability function to predict waste generation rates in small and large hotels during
peak and lean seasons. In Proceedings of the 7th Advances in Hospitality & Tourism
Marketing & Management (AHTMM) Conference, Famagusta, Cyprus, 10-15 July 2017
(pp. 539-547). Eastern Mediterranean University and Washington State University.
494 Casteleiro-Roca, J.L., Gómez-González, J.F., Calvo-Rolle, J.L., Jove, E., Quintián, H.,
Martín, J.F.A., Perez, S.G., Diaz, B.G., Calero-Garcia, F., & Méndez-Perez, J.A. (2018).
Prediction of the Energy Demand of a Hotel Using an Artificial Intelligence-Based Model.
In International Conference on Hybrid Artificial Intelligence Systems (pp. 586- 596). Cham:
Springer.
495 del Rio, J. S., Moctezuma, D., Conde, C., de Diego, I. M., & Cabello, E. (2016).
Automated border control e-gates and facial recognition systems. Computers & Security,
62, 49-72.
496 Nica, I., Tazl, O. A., & Wotawa, F. (2018). Chatbot-based tourist recommendations
using model-based reasoning. In Felfernig, A., Tiihonen, J., Hotz, L., & Stettinger, M. (Eds.)
Proceedings of the 20th International Configuration Workshop, 27-28 September 2018,
Graz, Austria, pp. 25-30.
497 Ivanov, S., & Webster, C. (2017). Designing robot-friendly hospitality facilities.
Proceedings of the Scientific Conference “Tourism. Innovations. Strategies”, 13-14
October 2017, Bourgas, Bulgaria, pp. 74-81.
498 Frey, C. B., & Osborne, M. A. (2017). The future of employment: how susceptible are
jobs to computerisation? Technological Forecasting and Social Change, 114, 254-280.
499 Tan, N., Mohan, R. E., & Watanabe, A. (2016). Toward a framework for robot-inclusive
environments. Automation in Construction, 69, 68-78.
500 Höykinpuro, R. (2015). Front stage with no front-stage employees: customer
perceptions of self-service hotels. In Ropo, A., Salovaara, P., Sauer, E., & De Paoli, D. (Eds.).
(2015). Leadership in spaces and places. Cheltenham: Edward Elgar Publishing, pp. 163-
180.
501 Kim, H., & Kim, B. G. (2018). A qualitative approach to automated motels: a rising
issue in South Korea. International Journal of Contemporary Hospitality Management,
30(7), 2622-2636.
375
 ANDREW W. PEARSON

502 Ivanov, S., & Webster, C. (2018). Adoption of robots, artificial intelligence and service
automation by travel, tourism and hospitality companies – a cost-benefit analysis. In
Marinov, V., Vodenska, M., Assenova, M. & Dogramadjieva E. (Eds) Traditions and
Innovations in Contemporary Tourism, Cambridge: Cambridge Scholars Publishing, pp.
190-203.

376
 CHAPTER SEVEN: THE A.I. CRUISE
LINE
“You can’t win in the digital era with industrial-age technology.”
~Adobe

Overview
In their article Data-Driven Transformation: Accelerate at Scale Now503,
Gourévitch et al. argue that “Data-driven transformation is becoming a question
of life or death in most industries” and more so for the cruise industry than many
others. “Most CEOs recognize the power of data-driven transformation. They
certainly would like the 20% to 30% EBITDA gains that their peers are racking up
by using fresh, granular data in sales, marketing, supply chain, manufacturing,
and R&D,” Gourévitch et al. claim.503 What’s not lost on these CEOs is the fact
that today the top five companies with the highest market capitalization
worldwide are all data-driven, tech companies–Apple, Alphabet, Microsoft,
Amazon, and Facebook.503 Five years ago, there was only one of these tech
companies in the top in the top five (Apple), whereas ten years ago there was
only one in the top ten (Microsoft).503
CEOs are correct in worrying about how their organizations are going to handle
a tenfold increase in company data when their managers are already
complaining about a lack of data skills and overburdened IT systems. 503
“
Transformations should start with pilots that pay off in weeks or months,
followed by a plan for tackling high-priority use cases, and finishing with a
program for building long-term capabilities,” recommend Gourévitch et al.503
“It starts with small-scale, rapid digitization efforts that lay the foundation for
the broader transformation and generate returns to help fund later phases of
the effort,” Gourévitch et al. advocate.503 “In the second and third phases,
companies draw on knowledge from their early wins to create a roadmap for
companywide transformation, ‘industrialize’ data and analytics, and build
systems and capabilities to execute new data-driven strategies and
processes.”503
In terms of infrastructure and data transformation, Gourévitch et al. advise
asking the following questions: “Can our current infrastructure support our
future data value map? Should we make or buy? Should we go to the cloud? Do
we need a data lake? What role should our legacy IT systems play in our data

377
 ANDREW W. PEARSON

transformation? The company should design a data platform (or data lake) that
can accommodate its product map and should use that platform to progressively
transform its legacy systems.”503
“To progressively transform its legacy system,” is an important idea here
because it is imperative that companies don’t bite off more than they can chew
when they decide to embrace the data driven culture. The Japanese have a
concept known as Kaizen — continuous incremental improvement — and it is an
idea that should be kept in mind when a company steps into the data driven and
AI world.
While the company architects its transformation roadmap, it needs to begin
industrializing its data and analytics.503 As Gourévitch et al. explain, “This means
setting up a way to standardize the creation and management of data-based
systems and processes so that the output is replicable, efficient, and reliable.” 503
Digital systems need to have all the attributes of industrial machinery, including
reliability and consistency.503
For analytics, a flexible open architecture that can be updated continuously and
enhanced with emerging technologies works best.503 “Rather than embracing an
end-to-end data architecture, companies should adopt a use-case-driven
approach, in which the architecture evolves to meet the requirements of each
new initiative,” advise Gourévitch et al.503 “The data governance and analytics
functions should collaborate to create a simplified data environment; this will
involve defining authorized sources of data and aggressively rationalizing
redundant repositories and data flows,” recommend Gourévitch et al.503
To prepare an organization for a digitized future, a company “needs to move on
four fronts: creating new roles and governance processes, instilling a data-
centric culture, adopting new ways of working, and cultivating the necessary
talent and skills.”503
Change starts at the top and senior leaders need to buy into and adopt data-
driven objectives and instill a data driven culture in every department
throughout the organization.503 Gourévitch et al. recommend that top
management “set up data councils to extend the work to all sectors of the
organization and to carry it out more effectively. The company should promote
data awareness by using data champions to disseminate data-driven
practices.”503
“Not everyone needs to become steeped in data analytics or learn to code in
order for digital transformation to work. However, everyone does need to adopt
a less risk-averse attitude,” advise Gourévitch et al.503 The writers recommend
companies embrace the software company model that utilizes a test-and-learn
philosophy that accepts failure and is constantly changing – and learning.503
Companies can also foster the desired cultural change through organizational

378
 THE A.I. CRUISE LINE

moves, “such as creating internal startup units where employees can focus on
experimentation or co-locating data labs within operating units.”503 “The
company can also promote the new culture by using cross-functional teams that
share data across silos, thereby encouraging openness and collaboration
throughout the organization,” advise Gourévitch et al.503
For any data-based transformation to succeed, a company needs talent with the
right skills to execute data-driven strategies and manage data-based
operations.503 Cruise lines should start by assessing its current employees and
defining their future needs.503 “The company should create an inventory of the
talents and skills that its employees will need, and it should identify where the
gaps are in the current workforce,” say Gourévitch et al.503
Companies should retrain current employees, hire new talent, or use a
partnership to get the right capabilities. “To recruit people with digital skills, the
company may need to rethink the value proposition it offers — work,
opportunity, rewards, career path, and so on — in relation to what tech
companies offer,” advise Gourévitch et al.503
Cruise line executives should be inspired by the idea of using data to make better
decisions, to create stronger customer bonds, and to digitize all sorts of
processes to improve performance. They should also be motivated “by fear that
they won’t be able to keep up with competitors who are ahead of them in data-
driven digital transformation.”503
However, some caution is due; sweeping, companywide change to go digital can
easily lead to counterproductive overreaching.503 In this case, the contest will
not necessarily be won by making huge bets.503 As Gourévitch et al. conclude,
“The winners will be agile, pragmatic, and disciplined. They will move fast and
capture quick wins, but they will also carefully plan a transformation roadmap to
optimize performance in the functions and operations that create the most
value, while building the technical capabilities and resources to sustain the
transformation.”503
As I mentioned at the beginning, one philosophy being adopted now is the
“experience is the new brand”25 mantra. Cruise lines need to take this philosophy
to heart and, throughout the rest of this book, I will reveal ways for them to do
just that.

Text
Text use cases break down into several different areas, including chatbots, NLP,
sentiment analysis, augmented search, and language translation.
According to Amazon, “Lex is a service for building conversational interfaces into
any application using voice and text. Amazon Lex provides the advanced deep

379
 ANDREW W. PEARSON

learning functionalities of automatic speech recognition (ASR) for converting

speech to text, and natural language understanding (NLU) to recognize the intent
of the text, to enable you to build applications with highly engaging user
experiences and lifelike conversational interactions.”504 Amazon Lex contains the
same deep learning technology that powers Amazon Alexa that are now
available to any developer. They will enable users to quickly and easily build
sophisticated, natural language, conversational chatbots.504
Amazon Transcribe “is an automatic speech recognition (ASR) service that makes
it easy for developers to add speech-to-text capability to their applications.” 505
Using the Amazon Transcribe API, brands can analyze audio files stored in
Amazon S3 and have the service return a text file of the transcribed
speech. Brands can also send a live audio stream to Amazon Transcribe and
receive a stream of transcripts in real time.
Amazon Transcribe can currently transcribe customer service calls and generate
subtitles on audio and video content. The service can transcribe audio files
stored in common formats, like WAV and MP3, with time stamps for every word
so that you can easily locate the audio in the original source by searching for the
text. Because it is an AI-based technology, Amazon Transcribe is continually
learning and improving to keep pace with the evolution of language.
Amazon Comprehend is “a natural language processing (NLP) service that uses
machine learning to find insights and relationships in text. No machine learning
experience required.”506
“There is often a treasure trove of potential sitting in a company’s unstructured
data. Customer emails, support tickets, product reviews, social media, even
advertising copy represents insights into customer sentiment that can be put to
work for a business,” says Amazon.506 Machine learning is “particularly good at
accurately identifying specific items of interest inside vast swathes of text (such
as finding company names in analyst reports).” 506 “Amazon Comprehend can
learn the sentiment hidden inside language (identifying negative reviews, or
positive customer interactions with customer service agents), at almost limitless
scale.”506
Amazon Comprehend “identifies the language of the text; extracts key phrases,
places, people, brands, or events; understands how positive or negative the text
is; analyzes text using tokenization and parts of speech; and automatically
organizes a collection of text files by topic.”506
In his article Turning Up Your Brand’s Voice to Reach the Most Advanced
Customers507, Pini Yakuel explains why digital assistants are unique to every
other channel when it comes to personalization — they cut through the noise.
“Yes, emails can be personalized, just like paid search and social ads, but they
share their real estate with thousands of other pieces of content. When there
are 20 personalized messages asking for your attention, which one will

380
 THE A.I. CRUISE LINE

consumers go for?” ask Yakuel.507

“Like any other marketing channel, the key to winning with digital assistants lies
in the deep knowledge of the purchase lifecycle customers go through. Delving
into the desire to know, go, do and buy that consumers have will deliver
success,” says Yakuel.507
“Micro-moments are defined as intent-rich moments when a person turns to
their device to act on a need through the conversational nature of queries to
digital assistants. Analyzing these intent-rich moments and acting upon them
might be the gate to showing up as the preferred answer,” says Yakuel. 507
Brands should always keep in mind “that retention through digital assistants
works so well because consumers want to make life easier for themselves every
chance they get.”507 “For example, smart fridges that know when we're running
out of food and talk to digital assistants are already out there. Digital assistants
will likely broaden their reach across the broad spectrum of consumer needs to
anticipate upcoming purchases,” says Yakuel.507
“Whether by diving into data or by directly speaking with existing and past
customers, marketers should always learn what questions and online behavior
drives folks toward their brands. Analyze their pain points, and focus on creating
content that uses their phrases and makes their lives easier quickly,” advises
Yakuel.507 “The better you get at this, the more likely you are to be the digital
assistant’s chosen option,” he concludes.507

Chatbots
In its 14 Powerful Chatbot Platforms508, Maruti Tech lists some of the best
chatbot publishing and development platforms for brands to use. According to
Maruti Tech, a “chatbot publishing platform is a medium through which the
chatbot can be accessed and used by the users.”508 “A chat bot development
platform, on the other hand is a tool/ application through which one can create
a chatbot,” says Maruti Tech.508 These chatbot platforms let users add more
functionality to a bot by creating a flow, machine learning capabilities, API
integration, etc.508 These chatbot platforms are simple to use, and users don’t
need to have deep technical knowledge or programming skills as many come
with drag-and-drop functionality.508 (Please note, there aren’t 14 platforms as
several have been discontinued or acquired, which is a testament to how quickly
this space can change.)

Chatfuel
Calling itself “the leading bot platform for creating AI chatbots for Facebook,”509
Chatfuel (chatfuel.com) claims that 46% of Messenger bots run on its
platform.509 No coding is required with Chatfuel, which “provides features like
adding content cards and sharing it to your followers automatically, gathering

381
 ANDREW W. PEARSON

information inside Messenger chats with forms.”509 Chatfuel also uses AI to script
interactive conversations.509
The platform is completely free for anyone to build a bot, but after the bot
reaches 100K conversations/month users have to subscribe as a premium
customer.509 Chatfuel’s client list includes multinational companies like Adidas,
MTV, British Airways, and Volkswagen.

Botsify
“Let your bot chat like a human” is Botsify’s (botsify.com) tagline510 and it is
another popular Facebook Messenger chatbot platform that uses a drag and
drop template to create bots.509 Botsify offers features like easy integrations via
plugins, Smart AI, Machine learning and analytics integration.508 Botsify’s
platform allows seamless transition from a bot to a human. 508 First bot is free,
but any others are charged for thereafter. 508

Flow XO
According to its website (flowxo.com)511, “Flow XO is a powerful automation
product that allows you to quickly and simply build incredible chatbots that help
you to communicate and engage with your customers across a wide range of
different sites, applications and social media platforms.” It is the only chatbot
platform to provide over 100 integrations.508 It boasts an easy to use and visual
editor.508 Flow XO’s platform allows users to build one bot and implement it
across multiple platforms.508 In terms of pricing, users are limited to a certain
number of conversations, surpassing that requires a subscription.508

Motion.ai
Recently purchased by Hubspot, Motion.ai was a chatbot platform that helps
users to visually build, train, and deploy chatbots on FB Messenger, Slack,
Smooch or any business’ website.508 Motion.ai lets users diagram a conversation
flow like a flowchart to get a visual overview of the outcomes of a bot query. 508
The bot can be connected to a messaging service like Slack, Facebook
Messenger, and go.508 Motion.ai allows Node.js deployment directly from its
interface along with several other integrations.508

Chatty People
This platform has predefined chatbots with templates for e-commerce,
customer support, and F&B businesses.508 When users select the e-commerce
chatbot, he or she can simply add products, Q&A information as well as some
general settings.508 The platform even includes PayPal and Stripe API
integration.508
According to Maruti Tech, “The chatbot platform’s simplicity makes it ideal for
entrepreneurs and marketers in smaller companies.”508 While its technology

382
 THE A.I. CRUISE LINE

makes it suitable for enterprise customers, users can make a simple bot
answering customer service questions or integrating it with Shopify to
potentially monetize one’s Facebook fan pages. (Chatty People was recently
acquired by MobileMonkey in 2018 and can now be found at
mobilemonkey.com.)

QnA bot
Microsoft has created QnA (qnamaker.ai) bot for the same reason as its name
suggests, i.e., for answering a series of user questions.508 The URL FAQ page must
be shared with the service and the bot will be created in a few minutes using the
information on the FAQ page and the structured data.508
Furthermore, the bot can be integrated with Microsoft Cognitive Services to
enable the bot to see, hear, interpret and interact in more human ways. 508 QnA
Maker seamlessly integrates with other APIs and can scale to be a know-it-all
part of a bigger bot.508

Recast.ai
In January 2019, Recast.AI was integrated into the SAP portfolio and
renamed SAP Conversational AI.512 This bot building platform enables users to
train, build and run their bots.508 By creating and managing the conversation
logic with Bot Builder, SAP Conversational AI’s visual flow management interface
and API lets users build bots that understand predefined queries as well as
quickly set up responses.508 Messaging metrics and bot analytics tools are also
included.508

BotKit
Their rather alliterate tagline is, “Building Blocks for Building Bots” and it is a
toolkit that gives users a helping hand to develop bots for FB Messenger, Slack,
Twilio, and more.508 “BotKit can be used to create clever, conversational
applications which map out the way that real humans speak,” says Maruti
Tech.508 “This essential detail differentiates from some of its other chatbot
toolkit counterparts,” BotKit adds.508
“BotKit includes a variety of useful tools, like Botkit Studio, boilerplate app
starter kits, a core library, and plugins to extend your bot capabilities. Botkit is
community-supported open-source software that is available on GitHub,” says
Maruti Tech.508 Online, the company can be found at: botkit.ai.

ChatterOn
On its website, ChatterOn (chatteron.io), claims it can help users build a chatbot
in five minutes. ChatterOn is a bot development platform which gives users the
required tools to build Facebook Messenger chatbots without any coding.508 The
platform helps users “build the bot flow (each interaction with a user has to have

383
 ANDREW W. PEARSON

a goal that the user has to be taken to the next chat) and setup the AI by entering
a few examples of the expected conversation between the user and bot.” 508
India’s first full stack chatbot development platform, ChatterOn is, according to
the company, “far superior in ease of development and functionalities than its
international counterparts.”508 “All the bots on ChatterOn’s platform are
powered by a proprietary self-learning contextual AI,” claims Maruti Tech.508

Octane AI
According to its website (octaneai.com), Octane AI “enables Shopify merchants
to increase revenue with a Facebook Messenger bot that customers love.”513
Octane AI has pre-built features that make it easy for users to add content,
messages, discussions, showcase merchandise, and much more to their bot. 508
According to Octane AI, convos are conversational stories that can be shared
with an audience. It’s as easy as writing a blog post and the best way to increase
distribution of a company’s bot, at least according to Octane AI. 508 The platform
also integrates with all of the popular social media channels as well as provides
real-time analytics.508

Converse AI
The Converse AI (converse.ai) platform has been built to handle a wide range of
use cases and integrates seamlessly with Facebook Messenger and Workplace,
Slack, Twilio, and Smooch. Some of its features include508:
• A complete UI that allows easy, code-free builds.
• Integration with multiple platforms, including complete user, request
and conversation tracking.
• Inbuilt NLP parsing engine, that includes the ability to easily build
conversation templates.
• Can converse while using both plain text and rich media.
• An inbuilt query and analytics engine allows for easy tracking and drill
down that helps brands understand how users are engaging with the
service.

GupShup
The leading smart messaging platform that handles over 4 billion messages per
month, GupShup (gupshup.io) has processed over 150 billion messages in
total.508 “It offers APIs for developers to build interactive, programmable, Omni-
channel messaging bots and services as well as SDKs to enable in-app and in web
messaging,” says Maruti Tech.508 “Unlike plain-text messages, GupShup’s
innovative smart-messages contain structured data and intelligence, thus
enabling advanced messaging workflows and automation,” add Maruti Tech. 508
In conclusion, chatbot platforms are essential for the development of

384
 THE A.I. CRUISE LINE

chatbots.508 With the availability of such platforms, Maruti Tech argues, anyone
can create a chatbot, even if they don’t know how to code. However, to make
an intelligent chatbot that works seamlessly, AI, machine learning and NLP are
required.508
Chatbots will undoubtedly revolutionize the future of industries by their rich
features.508 They will reduce human errors, “provide round the clock availability,
eliminate the need for multiple mobile applications and make it a very seamless
experience for the customer.508

Augmented Search
In his article How to use AI for link building and improve your search rankings 514,
Kevin Rowe claims that “AI’s applications in the search engine optimization (SEO)
world are continuing to expand to new horizons.” Besides the Y Combinator-
backed RankScience, which uses thousands of A/B tests to determine how best
to positively influence search engine rankings, it is unlikely that a complete
handling of SEO by AI will catch on any time soon. 514 While no software exists
that leverages AI to build links, brands can still use multiple types of software for
various stages of the link building process.514 These include514:
• Data collection. NLP tools can be used to determine if the sites are
contextually relevant and keyword relevant.
• Site analysis. AI can determine if a site will predictably have an impact
on rankings.
This means that AI can be used to augment, automate or automatize processes,
claims Rowe.514 Link building can’t specifically be a fully autonomous process,
but AI can be leveraged to augment human processes, which can help find
bloggers and influencers, as well as improve the quality of sites that are
approached for links, advises Rowe.514
Rowe believes to leverage existing AI in a link-building campaign, brands must
first look at websites as a whole, including the multiple contributors or people
on staff at these websites.514 “These can be good link-building opportunities
through sponsored or contributed content,” explains Rowe.514 Brands should
find industry publications or other informative sites that appeal to the brand’s
target audience.514 Brands should look for the following items514:
• Frequent publication: is new content often being published on the site?
• Last publish date: Has there been any new content in the last month?
• User experience and design: Is the design up-to-date and easy to use?
Secondly, brands should identify important industry blogs and influencers. 514
These sites usually have less people on staff than standard publications,
however, they just might have a wider reach.514
Rowe believes that, “Text processing analytics like Watson Analytics can be used

385
 ANDREW W. PEARSON

to find influencers and blog content that hits a brand’s target market.” 514 “For
instance, someone might not always say, ‘I am interested in polymer
manufacturing,’ online, but using AI tools that can predict related text patterns
and speech, you might be able to find more influencers who haven’t directly used
the terms you’re looking for,” says Rowe.514
Things to look for include514:
• Comments and social shares on posts: Do the influencer’s posts get a
lot of engagement?
• Last publish time and frequency: Is content published actively
consistently?
• User experience and design: Is it up-to-date and easy to use?
• Social platform: Does the influencer have a large social media following
on the platforms that are preferred by the brand’s industry players?
• Reputation: Sometimes, individual influencers or blogs might have a
strong opinion about hot topics that you might not want to be
associated with for either political or religious reasons.
Once a list of publications, influencers and blogs have been compiled, it’s time
for the hard part, i.e., determining if they will have an impact on your target
keyword rankings.514 Rowe calls this “the powerful part of AI — the part that can
improve the impact of the links.”514 “AI can process data from multiple sources
to identify likely variables or variable clusters that correlate with ranking in
Google,” claims Rowe.514

Image
Facial Recognition
Facial recognition technology is the capability of identifying or verifying a person
from a digital image or a video frame from a video source by comparing the
actual facial features of someone on camera against a database of facial images,
or faceprints, as they are also known.
As previously mentioned, rapid advancements in facial recognition technology
have reached the point where a single face can be compared against 36 million
others in about one second.92 A system made by Hitachi Kokusai Electric and
reported by DigInfo TV shown at a security trade show recently was able to
achieve this blazing speed by not wasting time on image processing. 92
Using edge analytics, the technology takes visual data directly from the camera
to compare the face in real time. 92 The software also groups faces with similar
features, so it can narrow down the field of choice very quickly. The usefulness
to the company’s security enforcement is pretty obvious, but it can be used by
multiple departments; facial recognition technology can be set up to send alerts

386
 THE A.I. CRUISE LINE

to clerks, managers, or just about anyone needing to identify customers.

As customers enter an area, “security cameras feed video to computers that pick
out every face in the crowd and rapidly take many measurements of each one’s
features, using algorithms to encode the data in strings of numbers,”515 as
explained in the Consumer Reports article Facial Recognition: Who’s Tracking
Who in Public.281 The faceprints are compared against a database, and when
there’s a match, the system alerts the VIP department or sales people. Faceprints
could also be used to allow people to purchase tickets or as part of a boarding
system.
In some sense, facial recognition technology is becoming second nature to
consumers, especially in Asia. Worldwide, consumers are used to tagging
themselves in photos on Facebook, Snapchat, Picasa, and/or WeChat. In 2015,
Google launched a photo app that helped users organize their pictures by
automatically identifying family members and friends.281
This technology is moving so fast that privacy advocates are having trouble
keeping up with it. In this regard, today’s facial recognition technology is
reminiscent of the World Wide Web of the mid-1990s.281 Back then, few people
would have anticipated that every detail about what we read, watched, and
bought online would become a commodity traded and used by big business and
sometimes, more sinisterly, hacked and used by nefarious individuals for
criminal purposes.281
According to his article Qantas have seen their Future. It’s Facial Recognition516,
Chris Riddell explains how Qantas is taking facial recognition technology to a
whole new level. According to Riddell516:
“Qantas have just started a brand-new programme of trialing
facial recognition to enable them to monitor passengers from
the very moment they check in, all the way through to the gate
when they board the plane. They’re also going to be monitoring
everything in between, including what café you’re getting your
coffee at, and where you are shopping for that last minute pair
of jeans. They’ll also know what electrical gadgets you were
playing with at the tech shop, and whether you were too busy
trying free shots of cognac to buy that gift for your other half
that you promised, but then ‘forgot’”
Riddell sees this as “a big retail play by the red kangaroo and it is pushing the
national airline into very new and unchartered territory. Qantas are
exceptionally interested in the movement of people through the terminals, and
how they spend their time.”516
Riddell adds that, “Qantas will want to know what people are doing, how long
they are doing it, which shops they are spending the most time, and which shops

387
 ANDREW W. PEARSON

they spend the least time in. By combining that with the incredible amount of
data from frequent flyer programme and passenger information they collect,
they’ll be catapulting themselves into the world of hyper intelligent retail.” 516
Of course, Qantas are not alone in wanting to capture all this customer data,
explains Riddell, every major airline is doing it.516 “The truth is though, few are
using the data they hoard with any level of real sophistication for the customer,”
claims Riddell.516
“All airlines know who you work for, who you book travel through, where you go
on holiday, where you travel for work and for how long you are away,” says
Riddell.516 “They also know what food you like, what food you are allergic to, and
who you bank with. They also know where you live, and who lives there with
you, whether you’ve got children, and how old they are. The list goes on…. If
you’ve linked other loyalty programmes to your frequent flyer account, they also
know a whole lot about your shopping habits,” adds Riddell. 516
All of this data helps a business understand its customer down to a macro level,
which is information that is more critical than ever.516 For a business like Qantas
this data helps them deliver services and experiences that are relevant, personal
and predictive.516
Next up, Riddell believes “will be the delivery of experiences in real-time as you
are in an airport retail store. Facial Recognition technology will be able to deliver
you services based on how you feel at the exact moment it matters. This is the
future, and it’s called emotional analytics.”516
As CB Insights reports in its What’s Next in AI? Artificial Intelligence Trends74,
“Academic institutions like Carnegie Mellon University are also working on
technology to help enhance video surveillance.” “The university was granted a
patent around ‘hallucinating facial features’ — a method to help law
enforcement agencies identify masked suspects by reconstructing a full face
when only periocular region of the face is captured. Facial recognition may then
be used to compare ‘hallucinated face’ to images of actual faces to find ones
with a strong correlation.”74
However, CB Insights warns that the tech is not without glitches. The report
states that “Amazon was in the news for reportedly misidentifying some
Congressmen as criminals”74 — although perhaps there’s a predictive element in
the technology that we’re unaware of as well?
“‘Smile to unlock’ and other such ‘liveness detection’ methods offer an added
layer of authentication,” claims CB Insight.74 For example, “Amazon was granted
a patent that explores additional layers of security, including asking users to
perform certain actions like ‘smile, blink, or tilt his or her head.’”74 These actions
can then be combined with ‘infrared image information, thermal imaging data,
or other such information’ for more robust authentication.”74

388
 THE A.I. CRUISE LINE

In his article Machine Learning and AI: If Only My Computer Had a Brain Wired
for Business, Michael Klein writes that, fifty-nine percent of fashion cruise lines
in the U.K. are using facial recognition to identify V.I.P clients and provide them
with special service.63 “The technology also enables cruise lines to track
customer sentiment and gauge how customers respond to in-store displays, how
long they spend in the store and traffic flow in each of their retail locations,” says
Klein.63
“But that’s not the only way cruise lines are taking advantage of facial
recognition and its AI technology. They’re using the technology, which is typically
employed in airports, for added security,” notes Klein. 63 For example, Saks “has
leveraged facial recognition technology to match the faces of shoppers caught
on security cameras with that of past shoplifters. From this perspective, AI can
serve the dual purpose of preventing losses while improving the customer
experience — and that ultimately helps cruise lines boost sales.”63

Image Search
“If a picture is worth a thousand words, visual search — the ability to use an
image to search for other identical or related visual asset — is worth thousands
of spot-on searches — and thousands of minutes saved on dead-end queries,”
says Brett Butterfield in his Adobe blog See It, Search It, Shop It: How AI is
Powering Visual Search.517 In the article, Butterfield explains how visual search
can become a big part of a buyer’s shopping future. With visual search, you don’t
need to try and guess the brand, style, and/or retail outlet something was
purchased on, you can simply snap a picture of the item you like, upload the
image, and immediately find exactly the same sneakers or ones like them, and
purchase them, all rather seamlessly.517
“That spot-it/want-it scene is common, and good for business. It could be a shirt
on someone walking down the street, an image on Instagram, or a piece of
furniture in a magazine — somewhere, your customer saw something that made
them want to buy one, and now they’re on a mission to find it,” explains
Butterfield.517
“While it’s a seemingly simple task, in many cases the path from seeing to buying
is a circuitous and friction-filled route that leads to a subpar purchase — or no
purchase at all. Just one in three Google searches, for example, leads to a click
— and these people come to the table with at least a sense of what they’re
searching for,” notes Butterfield.517
“Visual search is all about focusing your attention toward a target,” says Gina
Casagrande, senior Adobe Experience Cloud evangelist, “and helping you find
what you’re looking for that much faster. You also get the added benefit of
finding things you didn’t even know you were looking for.” 517

389
 ANDREW W. PEARSON

“Like text-based search, visual search interprets and understands a user’s input
— images, in this case — and delivers the most relevant search results possible.
However, instead of forcing people to think like computers, which is how the
typical text search works, visual search flips the script,” adds Butterfield. 517
“Powered by AI, the machine sees, interprets, and takes the visual cues it learns
from people. After applying metadata to the image, AI-powered visual search
systems can dig through and retrieve relevant results based on visual similarities,
such as color and composition,” explains Butterfield. 517 Visual search is another
technology that can facilitate better, more frictionless retail experiences that can
help buyers find what they want faster.517
“One early adopter of visual search is Synthetic, Organic’s cognitive technology
division, an Omnicom subsidiary,” says Butterfield.517 “Synthetic’s Style
Intelligence Agent (SIA) — powered by Adobe Sensei — uses AI to help
customers not just find specific clothing items, but also find the right accessories
to complete their new look.”517
To use SIA, customers simply upload an image, either from a website, from real
life or even from an ad in a magazine and from there, “Adobe Sensei’s Auto Tag
service extracts attributes from the image based on everything from color, to
style, to cut, to patterns.”517 SIA’s custom machine-learning model then kicks in,
correlating those tags with a massive catalog of products. 517 “SIA then displays
visually similar search results as well as relevant recommendations — items with
similar styles, cuts, colors, or patterns, for example.” 517 Just as importantly, SIA
then “uses these visual searches to build a rich profile for that customer’s
preferences and tastes — a much deeper profile than what could be built from
text-based searches alone.”517 Here you are getting customer preferences on
steroids, an enormous of amount of personalized data that can then be used in
customer marketing.
“This is where visual search goes beyond just search and becomes a true
shopping consultant,” says Casagrande, “and a superior, more sophisticated way
to search for what you want and what you didn’t know you wanted.” 517
“In delivering such a simple, seamless experience, AI-powered visual search
removes the friction from traditional search-and-shop experiences,” says
Butterfield.517 “No longer do customers have to visit multiple cruise lines or sites
and strike out. They can now find virtually anything, anywhere, even without
knowing exactly where to find it,” he adds.517 This is another important moment
for marketers because if brands invest in visual search, they can propel their
brand up the Google rankings and get a solid leg up on the competition.
Several retailers currently “use visual search to make the distance between
seeing and buying virtually nonexistent — within their own brand experience.”517
“Macy’s, for example, offers visual search capabilities on its mobile app, which

390
 THE A.I. CRUISE LINE

allows customers to snap a photo, and find similar products on Macys.com. It’s
‘taking impulse buying to new heights,’ one source says.”517
Frictionless image search is just the beginning.517 “The value of visual search
technology grows as the customer returns to the site,” says Casagrande. 517 “On
that next visit, it’s a more personalized, powerful targeted search. Just being able
to pick up where I left off and get to that product that much faster helps reduce
friction, and has been shown to increase conversions and order rate.” 517
“Visenze, which builds shopping experiences using AI, is already seeing these
benefits,” says Butterfield.517 For example, “the company saw a 50 percent
increase in conversion among clients such as Nike and Pinterest that
implemented visual search technology.”517
“In the United States, Amazon and Macy’s have been offering this feature for
some time,” says Visenze CEO Oliver Tan. “Consumers are crying out for a
simpler search process,” claims Tan.517 If brands don’t have that, their customers
will move on to other companies that do.517
Though the benefits of visual search are clear, there’s still a gap in between
customer expectations and delivery.517 “Our current iteration of visual search
gets us maybe 70 percent of the way there,” says Casagrande. 517
“Keep in mind, as more data and content become available the algorithms will
get smarter, and the visual search experience will only continue to get better,”
says Casagrande.517
Another interesting use of AI is what Pinterest is doing with its visual search
technology. According to Lauren Johnson’s Adweek article Pinterest Is Offering
Brands Its Visual Search Technology to Score Large Ad Deals 518, “The visual search
technology is Pinterest’s version of AI and human curation that lets consumers
snap a picture of IRL things and find similar items online. Taking a picture of a
red dress for example, pulls up posts of red dresses that consumers can browse
through and shop,” states Johnson.117
“The idea is to give people enough ideas that are visually related so that they
have a new way to identify and search for things,” said Amy Vener, retail vertical
strategy lead at Pinterest.117 “From a visual-discovery perspective, our
technology is doing something similar where we’re analyzing within the image
the colors, the shapes and the textures to bring that to another level of
dimension,” Vener adds.117
Utilizing the technology, someone who points his or her phone’s camera at a
baby crib will receive recommendations for similar baby products. 117
“Eventually, all of Target’s inventory will be equipped with Pinterest’s
technology to allow anyone to scan items in the real world and shop similar items
through Target.com,” states Johnson. “Target is the first retailers to build
Pinterest’s technology into its apps and website, though the site also has a deal

391
 ANDREW W. PEARSON

to power Bixby, Samsung’s AI app that works similarly.”117

“We’re now in a place where we’re using Pinterest as a service to power some
visual search for other products,” Vener said. “I think there’s an opportunity for
cruise lines to be a little more of a prominent player when it comes to visual
discovery.”117
Although these examples are mostly in the retail business, it is important for
cruise line executives to be aware of what is happening in a closely related
business.

Video
In his article The Future of Video Advertising is Artificial Intelligence519, Matt
Cimaglia sees a video advertising world that is completely different to the
current one. He describes it as such: “Meanwhile, somewhere in another office,
in that same year, a different team is creating a different digital video. Except
they're not shooting a single video: They're shooting multiple iterations of it. In
one, the actor changes shirts. In another, the actor is an actress. In another, the
actress is African American. After finishing the shoot, this agency doesn't pass
the footage off to a video editor. They pass it off to an algorithm.”519
Cimaglia states that, “The algorithm can cut a different video ad in milliseconds.
Instead of taking one day to edit one video, it could compile hundreds of videos,
each slightly different and tailored to specific viewers based on their user
data.”519 “As the video analytics flows in, the algorithm can edit the video in real-
time, too — instead of waiting a week to analyze and act on viewer behavior, the
algorithm can perform instantaneous A/B tests, optimizing the company's
investment in a day,” claims Cimaglia.519
Cimaglia believes this is what is happening right now.519 Cimaglia contends, “We
are witnessing a moment in video marketing history, like moments experienced
across other industries disrupted by the digital revolution, where human editors
are becoming obsolete.”519 This is the evolution of advertising — personalized
advertising, i.e., tailoring content to individuals rather than the masses 519;
surgically striking relevant offers to a market of one, rather than blasting a
shotgun of offerings to the uninterested many.
“Savvy agencies are turning to artificial intelligence for help making those new,
specialized creative decisions,” says Cimaglia.519 “It's the same logic that's long
overtaken programmatic banner and search advertising, machine learning and
chatbots: There are some things computers can do faster, cheaper and more
accurately than humans,” contends Cimaglia.
“In this future of data-driven dynamic content, viewers' information is siphoned
to AI that determines aspects of the video based on their data,” explains

392
 THE A.I. CRUISE LINE

Cimaglia.519
Cimaglia sees advertising being tailored towards individuals.519 “The options for
customization extend beyond user data, too. If it's raining outside, it could be
raining in the video,” easily done by the agency plugging in a geolocating
weather script.519 Similarly, if a user is watching the video at night, the video
could mirror reality and be a night scene filled with cricket sounds. 519 For
Cimaglia, “This is a logical progression for a society already accustomed to
exchanging their privacy for free services.”519 The video could also be in multiple
languages thanks to tools like Amazon Polly.
Cimaglia believes that “this customization model of video production is more
effective than the current model of creating a single video for the masses.”519 He
rightfully questions the current preoccupation in investing tremendous
resources in single, groundbreaking commercials.519 Currently, “It's all about
producing a multimillion-dollar, 30-second mini-film that screens during the
Super Bowl, gets viewed on YouTube 10 million times and wins a Cannes Lion,”
claims Cimaglia, but what really does that gain you? It’s less about the viewer
and more about stroking the already inflated egos of a select creative set, who
are doing nothing more than delivering a one-size-fits-all product to millions of
prospects.519
Cimaglia believes there is a place for this in a one-size-fits-all advertising product,
but making them “the centerpiece of a multimillion-dollar campaign is foolhardy
in an era when companies are sitting on more customer information than ever
before — and when AI is even taking over in that arena.”519 “Personalization is
the way of the future, but, unfortunately, most companies simply don't know
what to do with their stores of customer data,” laments Cimaglia. 519 However,
the companies that do will surely reap huge financial rewards.
Once cruise lines have created their content, the next question is where to put
it. In its Top 5 Programmatic Advertising Platforms for 2020 and Beyond 520,
Indrajeet Deshpande specifically lists out SmartyAds, TubeMogul, Simpli.fi,
MediaMath, and PubMatic as their top platforms for 2020.
“SmartyAds offers a full-stack programmatic advertising ecosystem to cater to
brand, agency, publisher, and app developer needs,” says Deshpande.520 The
platform “ecosystem consists of a Demand-Side Platform (DSP), Data
Management Platform (DMP), Supply-Side Platform (SSP), and an ad
exchange.”520 “SmartyAds also offers a white-label solution that allows brands
to build their product on top of it,” adds Deshpande.520
The platform enables cruise line advertisers “to run display, in-app, video, and
native advertising campaigns across desktop and mobile devices. Features such
as ad creation tools, robust analytics, and rich targeting capabilities help
advertisers with inventory selection, ad placement, and campaign
management.”520 “SmartyAds’ header bidding solution allows publishers to run

393
 ANDREW W. PEARSON

pre-bid auctions to sell the ad inventory at a premium price. App developers can
use the platform to run in-app ads in formats such as rewarded video, native,
banner, and playable ads,” says Hollander.520
TubeMogul’s programmatic advertising software, which is a part of the Adobe
Advertising Cloud, “allows brands and agencies to plan, run, and optimize their
advertising efforts.”520 “TubeMogul gives brands access to premium ad inventory
through the real-time bidding (RTB) process. Brands can also import their direct
deal or programmatic reserves inventory into the software,” explains
Deshpande.520
“You can run desktop video, mobile video, connected TV, display, native, and
digital out-of-home (DOOH) ads through the platform,” says Deshpande.520
“TubeMogul’s programmatic TV buying solution can be used to access the
inventory, which is not usually available through traditional media buying.”520
“Simpli.fi Programmatic Marketing Platform is a localized programmatic
platform that allows advertisers to buy ad inventory on multiple RTB ad
exchanges,” says Deshpande.520
“Advertisers can build audiences on various criteria such as device, OS used,
browser, geography, intent-based search data, etc.,”explains Deshpande.520
Users can import their CRM data to enhance audience targeting.520 “The
audience can be further optimized on the search and contextual keywords, IP
data, frequency capping, CRM data, etc.,” says Deshpande.520 Look-alike and
search-alike modeling can also be used to expand the audience base.520
MediaMath’s omnichannel programmatic marketing platform enables
marketers integrate data sets from first- and third-party sources, which can be
broken into segments before activation.520 Advertisers can also discover and
connect with their most valuable audience by utilizing the MediaMath Audience
feature.520
“The omnichannel DSP takes care of omnichannel ad campaigns that include
mobile, display, video, audio, native, and Digital Out of Home (DOOH) ads. The
Creative Management and Dynamic Creative features enable marketers to
deliver a consistent brand experience through all mediums,” notes
Deshpande.520
With PubMatic, advertisers can access a high-quality ad inventory through the
platforms private marketplace.520 “The media buyer console helps advertisers
plan and manage programmatic direct campaigns across different ad formats
and channels,” says Deshpande.520 “PubMatic’s fraud-free program refunds a
credit amount in case an ad fraud is detected.”520 Advertisers can also utilize the
PubMatic real-time analytics feature to optimize their campaigns.520

394
 THE A.I. CRUISE LINE

Audio
In his article AI’s role in next-generation voice recognition521, Brian Fuller notes
that “speech is a fundamental form of human connection that allows us to
communicate, articulate, vocalize, recognize, understand, and interpret. But
here’s where the complexity comes in: There are thousands of languages and
even more dialects.” “While English speakers might use upwards of 30,000
words, most embedded speech-recognition systems use a vocabulary of fewer
than 10,000 words. Accents and dialects increase the vocabulary size needed for
a recognition system to be able to correctly capture and process a wide range of
speakers within a single language,” states Fuller.521
Today, the state of speech-recognition and AI still has a long way to go to match
human capabilities.521 Fuller claims that, “With the continually improving
computing power and compact size of mobile processors, large vocabulary
engines that promote the use of natural speech are now available as an
embedded option for OEMs.”521
“The other key to improved voice recognition technology is distributed
computing,” says Fuller.521 We’ve gotten to this amazing point in voice-
recognition because of cloud computing, but there are limitations to cloud
technology when real-time elements are needed.521 Things are improving
radically but this is a very tricky world to operate in because user privacy,
security, and reliable connectivity are difficult to get to work in concert.521 “The
world is moving quickly to a new model of collaborative embedded-cloud
operation — called an embedded glue layer — that promotes uninterrupted
connectivity and directly addresses emerging cloud challenges for the
enterprise,” says Fuller.521
As Fuller explains521:
“With an embedded glue layer, capturing and processing user
voice or visual data can be performed locally and without
complete dependence on the cloud. In its simplest form, the
glue layer acts as an embedded service and collaborates with
the cloud-based service to provide native on-device processing.
The glue layer allows for mission-critical voice tasks — where
user or enterprise security, privacy and protection are required
— to be processed natively on the device as well as ensuring
continuous availability. Non-mission-critical tasks, such as
natural language processing, can be processed in the cloud
using low-bandwidth, textual data as the mode of bilateral
transmission. The embedded recognition glue layer provides
nearly the same level of scope as a cloud-based service, albeit
as a native process.”

395
 ANDREW W. PEARSON

Fuller believes that, “This approach to voice recognition technology will not only
revolutionize applications but devices as well.”521

Voice Activated Internet

In his article 2019 Predictions from 35 voice industry leaders522, Bret Kinsella
quotes Jason Fields, Chief Strategy Officer of Voicify, who claims that “2019 is
going to be the year voice and IVA’s are integrated into brands overall CX
strategy.”
In her article Voice search isn’t the next big disrupter, conversational AI Is523,
Christi Olson explains the importance of being what she calls ‘position zero’ in
the search rankings. She says523:
“When you type a query into a search engine, hundreds of
options pop up. It’s different with voice. When people engage
in a voice search using a digital assistant, roughly 40 percent of
the spoken responses today (and some say as many as 80%)
are derived from ‘featured snippet’ within the search results. In
search speak, that’s position zero. When you are that featured
snippet in an organic search, that’s what the assistant is going
to default to as the spoken response. Siri, Google, Cortana and
Alexa don’t respond with the other ten things that are a
possibility on that search page. Just the one.”
Understanding this, it’s clear why position zero is so important; “while you might
be number two in the text-based searches, you’re getting little to no traffic if
people are engaging with intelligent agents and listening to the spoken
response.”523
The opportunity here is for companies to reverse-engineer the process to ensure
they get position zero, so they can win the search race and therefore gain the
traffic. But how? “It goes back to the best practices of organic search, basic SEO,
and having a solid strategy,” argues Olson.523 “It’s embracing schema markup
and structured data within your website, so you are providing search engines
with signals and insights to be included in the knowledge graph. It’s claiming your
business listings so that the data is up-to-date and correct. It’s understanding the
questions people are asking and incorporating that question and conversational
tone into your content,” says Olson.523 “Simply put: It’s understanding the
language your customers are using so that you can provide value and answers in
their own words and phrases. So, let’s conclude with that,” Olson adds. 523
“Conversational AI for voice-assisted search is different from text-based search.
If you look at the top 80 percent of queries, text-based searches typically range
between one to three words. When we (at Microsoft, my employer) look at our
Cortana voice data, the voice searches coming in range from four to six words.
That’s substantially longer than a text-based search,” says Olson.523

396
 THE A.I. CRUISE LINE

It means that people are engaging with the digital assistant conversationally,
asking questions and engaging in almost full sentences. 523 “Given this insight,
there’s an opportunity to think about the questions your customers are now
asking. Think about what their need is in the way that your customers naturally
talk, not in marketer speak or marketing terms. Then, provide value back to them
in that manner,” recommends Olson.523
“With conversational AI, we’re going back to being able to create an emotional
connection through more meaningful conversations with our customers to build
relationships,” says Olson.523 “Brands will be able to differentiate themselves by
adding emotional intelligence to IQ through these conversations,” concludes
Olson.523
Amazon Polly is a service that turns text into lifelike speech, allowing users “to
create applications that talk, and build entirely new categories of speech-
enabled products.524 Amazon Polly is a text-to-speech service that uses advanced
deep learning technologies to synthesize speech that sounds like a human
voice.”524
Amazon Polly contains dozens of lifelike voices across a wide range of languages,
allowing users to select the ideal voice and build speech-enabled applications
that work in many different countries.524 At Intelligencia, we use it to quickly
create videos in multiple languages. Some of the Polly voices sound a little stilted
and machine-like, but there is usually one in the series of specific languages who
does a passable job.

Voice Search
In its article The Next Generation of Search: Voice525, seoClarity argues that
brands should take voice search seriously because it is becoming a zero-sum
game. seoClarity states that525:
“Because of the rise in voice search, Google has recognized the
increasing need to improve the experience for consumers
conducting these searches. Instead of simply displaying a list of
10 blue links, Google increasingly provides a single direct
answer to queries. This makes sense since voice searches are
often conducted when our hands and eyes are otherwise
occupied (for instance, while driving). A standard SERP result
would not be helpful in such situations. Rather, having the
answer (which Google believes to be the best answer for the
query) read out aloud provides immediate gratification and a
much better user experience. Therefore, Google’s response of
creating the Answer Box is no accident.”
“Now, and for the foreseeable future, Google’s Answer Box is the golden ticket
in the organic search rankings sweepstakes,” says seoClarity. 525 “In addition to it

397
 ANDREW W. PEARSON

being the only answer to voice search queries, it is the result that appears above
all other results on the SERP, ‘ranking zero’, seoClarity notes. 525 “Capturing the
Google’s Answer Box can mean a dramatic increase in traffic to your website,
credibility and overall brand awareness,” they add. 525
Google’s Answer Box, or “featured snippet block,” is the summary of an
answer.525 “Not only is the Google Answer box at the very first spot, above
standard organic results, but also has a unique presentation format that
immediately sets it apart from the remainder of the page. This instantly increases
the credibility and authority of the brand providing the answer to the user’s
query. Consequently, Google’s Answer Box may be the only search result viewed
by the user,” says seoClarity.525 Perhaps more importantly, it is the only answer
read in response to a voice search.525 “Not only does Google’s Answer Box
dominate the SERP, it also boosts organic traffic, leverages mid- to long-range
keywords, and focuses on the searcher’s intent,” notes seoClarity.525 “Given the
great importance of the Answer Box, brands should be focused on delivering the
best search experience rather than worrying about any specific tactic to trick the
algorithms,” argues seoClarity.525
“It’s valuable to think about the shopper’s journey. Shoppers at different stages
of their journey are searching for different things. So, it is crucial that brands
provide content that meet shoppers’ needs wherever they are in their journey.
When you are able to capture Google’s top result for searches along the
shopper’s journey, you will maximize your brand’s credibility and authority,”
argues seoClarity.525
“Voice search users tend to use specific, long-tail search phrases. Instead of
inquiring about a term or phrase, voice searchers typically ask proper questions,”
says seoClarity.525 “For example, when looking for places to dine out, desktop
users might type ‘Italian restaurant.’ However, when using voice search, they’re
more likely to ask, ‘where’s the nearest Italian restaurant?’”525
Voice searchers tend to use language that’s relevant to them.525 “When speaking
to their device, queries are more conversational, leaving it to the search engine
to decipher the actual intent,” says seoClarity.525
Voice searches are more targeted in the awareness and consideration phase. 525
Many voice searches have local intent — “as much as 22 percent of voice queries
inquire about local information such as directions, restaurants, shopping, local
services, weather, local events, traffic, etc.,” says seoClarity.525 “The remainder
of queries is distributed between non-commercial queries like personal assistant
tasks, entertainment, and general searches,” note seoClarity.525 “This makes
local the biggest commercial intent among voice searches. As a result, you should
incorporate new strategies to position your business in local voice search,”
argues seoClarity.525
Voice search is still messy and complex.525 “Google’s RankBrain algorithm

398
 THE A.I. CRUISE LINE

leverages artificial intelligence to discover contextual connections between

searches,” says seoClarity.525 Google “tries to understand ‘intent’ based on
context of the search (such as location, time of day, device used, previous
searches, connected data from email and other assistant sources) instead of just
plainly matching words from on a page.”525 However, the machine is learning
and training, so, “instead of trying to keep up with Google’s algorithms, it is
essential to understand what your audience needs and focus your optimization
to your end user, not on chasing the latest algorithm shifts.” 525
seoClarity recommends brands to “build a more effective content marketing
strategy to win the Answer Box by optimizing for topics that reflect the intent of
your audience instead of just optimizing for keywords.” 525 When brands focus
their content strategies on the intent of the audience, it better addresses the
real needs of the customer.525 Additionally, the created content can solve
challenges and answer most commonly asked brand questions. 525 By targeting
the awareness and consideration phases of the customer journey, brands can
capture their audience early in the customer journey.525
Brands should optimize to short attention spans.525 It is essential to connect with
customers at the right moment.525 SeoClarity says that, “Google outlines the
following moments that every marketer should know: I-want-to-know moment;
I-want-to-go moment; I-want-to do moment; I want-to-buy moment.”525
Always create a FAQ page as it can provide answers to common questions that
users may have.525 “By figuring out what questions your customers are asking,
you can create the type of content that they are most likely to find useful,” says
seoClarity.525
“Answer the five W’s & H — Be sure to answer the essential questions that
everyone asks when collecting information or solving a problem: Who, What,
Where, When, Why. And don’t forget the all-important How,” says seoClarity.525
“The data also showed some other important trigger words including Best, Can,
Is, and Top,” says seoClarity.525 Brands should also “Explain steps to complete
tasks — Focus on content that details steps and how to complete tasks that
relate to your product or service and also other explanations specifically for your
product or service. “How to” and “What Is” contain significant lead over other
trigger words.”525
Other things seoClarity recommends are, “Highlight the best options for
customers. Create buying guides that help aid the decision making process in list
and bullet point type of format to demonstrate the best options for
customers.”525
“Focus on structuring content in a way that matches consumer intent — Use
formats that work for your customers and structure the content to intent,”
recommends seoClarity.525 “Consider using tables, ordered lists, bullet points,
and video. Use schema markup — Always use the best SEO practices by placing

399
 ANDREW W. PEARSON

your keywords and key phrases in your header, metadata, URL structures, and
alt tags,” says seoClarity.525

One of the most important recommendations seoClarity offers is for brands to

produce in-depth content.525 “In your SEO efforts, you must never forget that
content is the most important thing. Be sure to create relevant content that
provides in-depth answers to the questions your target audience asks,” says
seoClarity.525

According to seoClarity526, nearly 20% of all voice search queries are triggered by
only 25 keywords (see Table 22), which include “how”, “why”, or “what”, as well
as adjectives like “best” or “easy”. The top ten are listed below, others included
“Why”, “Who”, “New”, “Recipe”, “Good”, “Homes”, “Make”, “Does”, “Define”,
“Free”, “I”, “List”, “Home”, “Types”, and “Do.”

Trigger Words Count % of Total

How 658,976 8.64%
What 382,224 5.01%
Best 200,206 2.63%

The 75,025 0.98%

Is 53,497 0.70%

Where 43,178 0.57%

Can 42,757 0.56%

Top 42,277 0.56%

Easy 31,178 0.41%

When 27,571 0.36%

Table 22: Voice search words

Search: Dialogtech.com526

Programmatic Advertising
“When it comes to advertising,” the Adobe Sensei Team believes that, “the
promise of AI is that customers will receive the most relevant ads, while allowing
brands to drive awareness, engagement, conversions, and loyalty.” 26 This should
result in happier customers and less wasted ad spend.26 “With AI, advertisers can
budget, plan, and more effectively spend limited ad dollars,” claim the Adobe

400
 THE A.I. CRUISE LINE

Sensei Team.26
The Adobe Sensei team provides the following example: “Cynthia is a travel and
hospitality media buyer trying to determine the best mix of search advertising
for her global hotel brand.”26 “She knows that with millions of keywords,
multiple search engines, and different audience segments to consider, coming
up with the right bid amount for each combination, as well as determining how
to allocate her budget across her campaigns to most efficiently meet her goals is
simply too much for her to handle alone.” 26 Cynthia “turns to her media buying
platform to help her make sense of the data.”26 The Adobe Sensei Team sees the
process working as follows for Cynthia26:
"With AI leading the way, she reviews a forecast simulation to
see how an increase or decrease in budget will impact her
clicks, revenue, conversions, and other metrics. Once she
selects her budget, she reviews AI-powered ad spend
recommendations to see how to best allocate her advertising
budget. She clicks on her preferred allocation. Later, as her ad
campaigns are running, she accesses model accuracy
performance reports so she can see how actual performance
numbers compare with AI-generated forecasts, allowing her to
make any necessary adjustments along the way. Once her
campaigns have run, she’s thrilled to see that they delivered 99
percent of the clicks that were forecast, and actual revenue
was five percent higher than forecasted. Now that Cynthia has
a clear picture of what worked during her search ad
campaigns, she checks the performance of her display and
video campaigns. Again, she calls on AI to report on awareness
and performance while letting her demand-side platform (DSP)
guide automated budget allocation so she can stay focused on
strategic media planning and buying.
The Adobe Sensei Team believes that with the help of AI, brands “can keep up
with changing customer preferences, navigate mountains of data, and make
adjustments multiple times per day if needed to make sure” 26 budgets are
allocated “most effectively across channels like search, display, and video, or
even within a specific channel.”26
In her article Experts Weigh in On the Future of Advertising, Giselle Abramovich
believes that AI can help build a media-buying platform that allows a marketer
to input goals “and a transparent algorithm does the rest, executing buys and
optimizing every millisecond.”189 The ad could dynamically change the tone of
the voiceover based on the preferences of the viewer. 189 Abramovich believes
that, “The convergence of AI with human creativity and insight will transform
advertising, and we’re just beginning to see what’s possible.” 189

401
 ANDREW W. PEARSON

One of the companies delving into AI head-first is Citi, which recently launched
its “Welcome What’s Next” campaign.189 “[AI] is allowing us to create custom ads
that meet people where they are. For example, if you’re looking at the weather,
it’s serving up the ad in a customized way so it’s relevant to what you’re looking
at,” says Jennifer Breithaupt, global consumer CMO at Citi.189 “It integrates with
a consumer’s path online and provides a more seamless way to experience the
ad.”189
AI is already helping Citi surpass its advertising benchmarks, Breithaupt adds.189
“For example, the financial giant has realized a 10%+ lift in video completion
rates versus standard, non-customized ads as a result of AI.”189
“But what’s going to be crucial to the success of AI is structuring it in a
transparent manner that involves a partnership between parties,” says
Breithaupt.189 “In other words, above all as advertisers, it’s crucial we’re clearly
defining the value exchange and providing consumers with the opportunity to
make an informed choice about their participation.”189
In his article How AI is Driving a New Era of TV Advertising527, Varun Batra states
that in November 2017, eMarketer reported that 70% of U.S. adults “second
screen” while watching TV. Although that sounds pretty discouraging for brands
that spend millions on TV spots, one should consider that this is reported, not
observed, behavior.527 “No doubt we all second screen, but we don’t do so all
the time. That begs the question: how does a brand know consumers paid
attention to its $5 million Super Bowl ad rather than their mobiles?” asks
Batra.527
“Using AI, data scientists have been able to map multiple devices to the same
individual and household, as well as to connect online behavior with offline
behavior, such as watching the Super Bowl via a connected TV and engaging with
a smartphone during commercials,” explains Batra.527 Brands “can determine
when consumers second screen during the commercials by counting the number
of bid requests from their devices,” says Batra.527
“Of course, AI can’t tell us if an inactive device meant the consumer watched the
ad or went to the kitchen for another beer, but if we track ad requests across
millions of household, we can get a lot of insight into a creative’s ability to
captivate consumers,” says Batra.527
“AI can also help determine the impact of an ad on consumer behavior, thanks
to that same ability to link online and offline behavior. For example, if we know
that a particular household was presented with a TV ad for a ‘one-day-only sale’
on GM pickup trucks, and a mobile device associated that household shows up
at the local dealer on sale day, then we can assume the ad had an impact,”
explains Batra.527 “The connection becomes more compelling when the behavior
is seen across all households that see the ad,” he adds. 527

402
 THE A.I. CRUISE LINE

“Marketers will continue to see new opportunities to improve their campaigns

as TV becomes more digitized,” argues Batra.527 He adds that, “As of 2017, there
are nearly 133 million connected TV users in the US and will grow to at least 181
million in” 2018.527 That means the online and offline behavior of 55% of the
population can now be tied to ad-views, which is obviously a huge number.527
“Many programmatic companies allow marketers to incorporate TV inventory
into their multi-channel programmatic campaigns. These connected TVs are
targeted using first- and third-party data sets, just as if they were laptops and
tablets,” says Batra.527 “Marketers can create surround-sound marketing, hitting
consumers with messages on their laptops, mobile devices, and TVs,” he adds.527
“AI is more precisely transforming the very segments we use to pinpoint
consumers who are in the market for a particular product,” says Batra. “Machine
learning excels at sifting through massive amounts of observed online and offline
user behavior to discover distinct signals that indicate purchase intent,” explains
Batra.527 AI can also make sub-millisecond decisions to remove a consumer from
a targeting segment as soon as he or she stops sending in-market signals.527
Humans just don’t have the capacity to do this, so the models do it completely
autonomously and at a scale far beyond human capability. 527
“Through numerous applications of machine learning, we’ve learned that there
is a host of common — and often non-intuitive — behaviors that people engage
in before they exhibit the signals of being in the market,” says Batra. 527 For
example, “in the classic digital marketing use case, airlines will retarget
consumers who search for flights to Las Vegas. In a machine-learning use case,
airline marketers would target consumers who look at wedding chapels, an early
signal that they’ll soon look for a flight to the city.”527 “In other words, the
machine predicts who an airline’s future customer will be, giving the airline the
opportunity to get a jump on its competition,” notes Batra.527
Batra believes that, “Television has always been a powerful awareness tool,
enabling brands to reach millions of consumers quickly and effectively.” 527 He
concludes that, AI can only enhance “that power by predicting the right people
to receive a TV ad, gauging its effectiveness, and assessing its impact on online
and offline consumer behavior.”527

The Customer Journey

As I wrote in chapter one, successful marketing is about reaching a customer
with an interesting offer when he or she is primed to accept it. Knowing what
might interest the customer is half the battle to making the sale and this is where
customer analytics comes in. Customer analytics has evolved from simply
reporting customer behavior to segmenting customers based on their
profitability, to predicting that profitability, to improving those predictions

403
 ANDREW W. PEARSON

(because of the inclusion of new data), to actually manipulating customer

behavior with target-specific promotional offers and marketing campaigns.
A cruise line needs a single view of the customer in real-time that will enable its
marketers to deliver personalized experiences whenever the customer is primed
to receive them. Data can come from transactional systems, CRM systems, app
impressions, operational data, facial recognition software, wearables, iBeacons,
clickstream data, amongst a whole host of other data systems.
As Dan Woods explains in his amusing comparison of the different environments
that marketers face today when compared with what their 1980’s counterparts
saw.23 Today, stealthy marketers are forced to use “email campaigns, events,
blogging, tweeting, PR, eBooks, white papers, apps, banner ads, Google Ad
Words, social media outreach, search engine optimization.”23 Woods didn’t
include SMS, but that is still an integral part of today’s marketing chain.
In practice, all these channels should work in concert; an email campaign can
promote a sale at an event, which can be blogged and tweeted about through
social media. PR can also promote the event through its typical channels.
Coupons for the event can be disseminated through the company’s mobile app
and SMS messaging services. Banner ads will appear on the company’s website,
while Google ads and SEO will drive buyers and potential buyers to the cruise
line’s website or its social channels. Hopefully, viral marketing then kicks in, with
customers and potential customers sharing on Facebook, Instagram, Pinterest,
Twitter, Weibo, WeChat, etc., etc. Of course, influencer marketing can also help
the viral marketing process.
Seen through the lens of the Cruise line Engagement and Loyalty Platform, all of
these activities can increase personalization to the point where it will be recognized
by the customer.
Lovelock and Wirtz’s “Wheel of Loyalty” concept and its three sequential steps
— building a foundation for loyalty, creating loyalty bonds, and identifying and
reducing factors that result in churn76 should be kept in mind when building up
the foundation of The A.I. Cruise Line’s CRM system. The most important part of
the second step is the cross-selling and bundling of products and a real-time
stream processing recommendation engine can certainly help with that. One
thing that should be kept in mind regarding loyalty programs is the fact that
cruise lines should try to foster relational behaviors and gain long-term benefits
from a loyalty program, and to do this cruise lines need to consider offering both
social and economic rewards.101

Listening
In the Listening part cruise lines should define and look out for triggers such as
photos, hashtags, keywords, likes, video views, etc., etc.

404
 THE A.I. CRUISE LINE

In their article Artificial Intelligence and Visual Analytics: A Deep-Learning Approach

to Analyze Hotel Reviews & Responses528, Chih-Hao Ku et al. present a framework
to “integrate visual analytics and machine learning techniques to investigate
whether hotel managers respond to positive and negative reviews differently and
how to use a deep learning approach to prioritize responses.”
According to Chih-Hao Ku et al., “Online travel agencies (OTAs) such as
Booking.com, Expedia, and TripAdvisor provide a platform where users can share
subjective opinions, recommendations and ratings about their travel and
accommodation experiences.”528 Today, TripAdvisor, the largest travel platform
in the world529, has over 630 million reviews and opinions with an average of 455
million monthly unique visitors.530 At “TripAdvisor, a managerial response
becomes the final conversation on the review because only one registered
manager can create such last response. That means, managerial responses can
have a significant impact on other potential guests who plan to book a hotel.” 531
“In hospitality management, the customer-generated content such as hotel
ratings and reviews could be a valuable source for identifying the consumption
patterns and trends due to its active and real-time natures,” claim Chih-Hao Ku
et al.528 “Managerial responses to customer reviews may enhance existing
customers’ loyalty and turn unhappy customers into loyal customers,” says
Pantelidis.532
In 2015, Schuckert et al.533 conducted a content analysis study on 50 articles from
2004 to 2013 that were relevant to hospitality and tourism online reviews. The
researchers discovered that existing studies focused predominantly on analyzing
secondary data, discovering the relationship between online reviews and sales
including customer satisfaction, and opinion mining of online reviews. 534 They
further pointed out the limitations of studies of the time, which included the use
of simple variables such as overall ratings and the number of reviews535 for data
analyses.533 Furthermore, prior research has focused mostly on a stand-alone
fashion536, analyzing either online reviews or managerial responses, which
generate limited insights of the interrelated relationship between online reviews
and managerial responses.537
Chih-Hao Ku et al.’s study filled “the research gaps by taking dimensions such as
aspect ratings, types of travelers, and time to respond to reviews into our data
analyses.”528 “A novel approach to integrate deep-learning models and visual
analytics techniques”528 was then proposed. The overall results can be used to
improve customer relationship management, make self-improvements532 for
response management managers, and generate decision-making information for
travelers.528
Today, according to Mauri and Minazzi in their article Web Reviews Influence on
Expectations and Purchasing Intentions of Hotel Potential Customers,538 “tourists
usually read online reviews to plan their trip and decide where to stay.”

405
 ANDREW W. PEARSON

“Traditional word-of-mouth (WOM) communications, oral messages between

persons, have evolved into electronic word-of-mouth (eWOM) communications,
online messages between users,” say Mauri and Minazzi. 538 “The proliferation of
eWOM has been identified as a strong impact on consumers’ purchase
decision,”538 as well as on revisit intentions539, search behaviors540, and online
sales.534 Consumers tend to search and compare tourism and hospitality
products and services to reduce uncertainty541 and potential risks associated
with a purchase.542
“An average review rating is an important indicator leading to hotel sales,” argue
Chih-Hao Ku et al.528 In their article Pricing in a Social World: The Influence of Non-
Price Information on Hotel Choice543 Noone and McGuire “examined the relation
between online reviews and online hotel booking and found that higher average
review ratings lead to higher numbers of hotel bookings.” “However, the overall
ratings may not reveal customers’ real satisfaction, and more nuances of
response strategies should be further studied,” say Liu, Li and Thomas.544
Prior studies have revealed that a positive correlation between a hotel’s star
rating and hotel’s sales.545 Martin-Fuentes546 collected a sample of more than
14,000 hotels in 100 cities from Booking.com and TripAdvisor to examine the
star-rating classification system of hotels, room price, and user satisfaction
measure from user ratings. Martin-Fuentes’ analysis confirmed that hotel stars
indicate the overall quality of hotels and the hotel price is well corelated to hotel
stars and user satisfaction levels.546
Due to the explosion of online reviews over the past decade or two on social
media platforms of all kinds, managerial responses have become a new form of
CRM.547 Law et al.548 analyzed 111 hospitality-related articles from March to
August 2017 and concluded that hospitality CRM research has grown from a
marketing to a social CRM concept. They further point out that technology plays
an indispensable role in such processes and AI can generate new knowledge in
this rapidly growing field and thus foster strong customer relationships.548
Today, managing online reviews for hotels has become an important task for
hospitality management549 and researchers have urged hotel managers to
respond to online reviews proactively.542 Existing studies550 reveal that deficient
service does not necessarily lead to dissatisfaction, but rather improper
responses do and most customers recognize imperfect service.545 More recently,
Sparks et al.542 adopted an experimental approach based on Kardes’s consumer
inferences theory551 to examine organizational responses to negative eWOM
and found that a timely response yielded favorable customer inferences.
According to Andreassen’s service recovery theory 552, managerial responses to
negative reviews can identify service failures and enhance customer
satisfaction.553
Kim et al.554 collected online reviews and responses from 128 hotels in 45 states

406
 THE A.I. CRUISE LINE

in the U.S. and found that overall ratings and responses to negative comments
are the most salient predictor of hotel performance. As Sparks et al. contend,
“Seeking effective approaches to manage eWOM, especially negative ones, is a
widely recognized challenge for hospitality management.”542
Obviously “Understanding and responding to massive online reviews is a time-
consuming and exhausting task.”528 As Chih-Hao Ku et al. reveal, “A customer
review may contain both positive and negative information, which make the in-
depth analysis of online reviews even more challenging. 528 To automate the
analysis process with millions of data records, AI techniques almost have to be
used.528
One of the popular unsupervised learning algorithms is the k-means clustering
algorithm, which groups data points that are found to possess similar features
together.528 Zhang and Yu555, for example, “use a Word2Vec tool, a deep-learning
tool proposed by Google, k-means clustering algorithm, and ISODATA, a
clustering algorithm based on k-means, to conduct the experiments of sentiment
analysis on hotel reviews and found a slight performance improvement by using
Word2Vec together with ISODATA.” Chong et al.556 used a neural network
approach to investigate the use of online reviews, online promotion strategies,
and sentiments from user reviews to predict product sales and the researchers
did find a positive correlation.
For Chih-Hao Ku et al.’s study, London was selected because it is both an English-
speaking city and the most visited city in Europe and Northern America based on
the volume of visitors.528 Chih-Hao Ku et al. selected “43 Hilton hotels, up to 25
miles from London, in 2017 based on Hilton’s website. Only 4- and 5-stars hotels
were selected, because luxury hotels are more likely to provide better
experience and service to guests. Three Hampton hotels that have lower or no
star rankings are eliminated from our data analysis. This results in a total of 40
hotels used in this study.”528
Relying on an automatic web crawler, Chih-Hao Ku et al. collected user review,
manager response, and hotel rating data from TripAdvisor. 528 For each hotel,
Chih-Hao Ku et al. “collected hotel name, star, the number of excellent, good,
average, poor, and terrible reviews, an average of a price range, hotel address,
amenities, type of rooms, and description.” 528 For each hotel review, Chih-Hao
Ku et al. “collected the hotel name, review title, review content, manager
response, overall rating, aspect ratings, types of travelers, and review date.” 528
“A total of 91,051 reviews were collected. Among them, 70,397 reviews contain
managerial responses, resulting in an overall 77% response rate.” 528
To preprocess the collected data, Chih-Hao Ku et al. first joined two datasets –
hotel data and review data – to create a holistic view of the data, then keyword
extraction, value extraction and data classification were done on the data.528
Chih-Hao Ku et al. examined summary statistics and data visualizations containing

407
 ANDREW W. PEARSON

more than 20 dimensions such as sentiment, aspect rating, managerial response,

review time, response time, latitude, longitude, and type of traveler. 528
An Öğüt and Taş study534 compared hotels in London and Paris listed on booking
sites and revealed that hotel star ratings significantly affect room prices and
customer ratings; star ratings are highly correlated with room prices and
frequently used to rate hotel quality.534 Glauber Eduardo557 also discovered that
cleanliness, location, and facilities in 8,000 hotels worldwide are relevant to
hotel quality and price differentials. Another researcher, Ye et al.558, found that
a good reputation is related to a higher hotel price.
Chih-Hao Ku et al. believe their study breaks new ground in several ways.528 They
“developed a data crawler to collect data automatically and presented a novel
approach to integrating visual analytics and deep-learning models to gain
insights into various aspects of hotel review and response data.” 528 According to
Chih-Hao Ku et al. 528:
“The study result produces managerial, decision-making, and
technical contributions. First, hotel managers can prioritize
response orders and gain insights into online reviews and
responses to make self-improvement. Second, the overall
results also provide decision-making information for travelers
to select 4-star hotels and enjoy 5-star service and environment
based on our clustering analysis. Finally, we are among the first
to integrate visual analytics and deep-learning models to
analyze hotel reviews and responses. This can be justified by
our experimental results, which indicate our proposed
approach outperforms existing machine learning methods such
as NB, KNN, SVM, DKV, and RNN.”
Prior research, such as Berger, Sorensen, and Rasmussen’s559 emphasized that
consumers are likely to pay more attention to negative reviews when making
purchase decisions, specifically negative reviews could increase consumer
awareness of hotels.560 However, Chih-Hao Ku et al.’s results showed “that 72.5%
of hotels in our study have a neutral preference of response strategy, implying
that the hotel managers put an equal amount of effort to respond to customers’
positive and negative reviews.”528 As negative reviews reduce purchase
likelihood and sales559, Chih-Hao Ku et al.’s findings “suggests that managers
should provide detailed strategies to respond to negative and positive
reviews.”528
Although the Chih-Hao Ku et al. study was quite illuminating, there are several
limitations to it, the writers admit; 528 the “study only took the response time,
response rate, and hotel rating in response data analysis;”528 “The nuances of
response strategies need to be further studied, which should provide practical
decision-making information for hotel managers when they respond to positive

408
 THE A.I. CRUISE LINE

and negative reviews using different response strategies,” contend Chih-Hao Ku

et al.528 Another limitation was the small sample.528 Chih-Hao Ku et al. confess
that, “The performance of the deep-learning models needs to be tested in future
research by including more cities, hotel brands, and hotels. This will provide
additional insights into data analysis.”528
Not all eWOMs are of equal value, Chih-Hao Ku et al. admit.528 “After analyzing
the textual features of the hotel reviews and managerial responses, future
research can focus on the social network among those hotel questions and
answers.”528 Also, “The analytical framework applied in this study can be
expanded to include social network analysis, which can show how the structure
of social ties may influence the hotel reviews and manager’s response
strategy.”528
Chih-Hao Ku et al. see the potential for each hotel to have a small social network
that is based on the communication between customers. 528 “If one customer
answers another customer’s question, this indicates a one directional tie
between these two customers. Based on these conversation ties, a small social
network of each hotel can be generated.”528 This does propose the following
interesting question: “if the structural cohesion of social network will influence
the hotel reviews and response strategy.”528 “An equally intriguing question is
whether the structural holes in each network will cause a different impact on the
hotel reviews and response strategy,”528 but that’s for another time and another
study.
Hootsuite’s 14 of the Best Social Media Monitoring Tools for Business561 lists
some of the best tools for cruise lines to use for this listening step, including
Reddit, Streamview, Reputology, and Synthesio, Crowd Analyzer, amongst
others (see Table 23).
SERVICE DESCRIPTION
With a community of over 700 million users it makes sense to monitor what
people are posting on Instagram, especially if your audience falls in the 18 to
29 age range. With the Streamview for Instagram app you can monitor posts
Streamview
by location, hashtag, or username. The app within Hootsuite allows you to
for Instagram
monitor and engage with users that are posting in your area, or an area you
choose to follow. For example, you can use this tool during events to see what
is being posted and to engage with attended.

Manage and monitor all your favorite blogs and websites with Hootsuite
Hootsuite
Syndicator Pro. This tool provides a quick and easy way to view RSS feeds and
Syndicator
quickly share them to your social media channels, as well as rich filtering,
Pro
monitoring, and tracking tools. You can also track which stories you’ve shared.

Online (and offline) reputation management is extremely important and

surprisingly easy. The Reputology app lets you monitor and check major review
Reputology sites, such as Yelp, Google, Facebook reviews, so that you can engage with
reviewers and resolve any issues in a timely manner. You can track activity
across multiple storefronts and locations, and respond quickly via quick links.

409
 ANDREW W. PEARSON

SERVICE DESCRIPTION
Hootsuite Insights combines social media listening, analytics, and powerful
social media monitoring capabilities. It allows you to gain powerful real-time
Hootsuite insights about your brand, track influencers, stories, and trends, and visualize
Insights the metrics — all in one place. You can filter and tailor results by sentiment,
platform, location, and language, and engage directly from your stream to take
action on previously hidden results.

The name says it all; the Brandwatch app in Hootsuite lets you keep watch over
your brand through deep listening. You can identify key insights from more
Brandwatch than 70 million traffic sources across the web, including major social channels,
blogs, forums, news and review sites, and much more. This tool lets you make
real-time, informed decisions and take action on them.

Whether it’s a positive or negative online review, your response should be in

the same place as that review. The ReviewInc app for Hootsuite lets you view
ReviewInc
over 200 popular review sites across over 100 countries. Organize positive
reviews for sharing on social media sites, and resolve negative issues instantly.

Synthesio is a comprehensive social monitoring tool for finding the information

you need to gain deeper insights and better inform business decisions. The tool
Synthesio lets you monitor multiple mention streams at once, so you can listen to the
social media conversations most important to you. You can then analyze these
conversations and join them.

If you or your customers are based in the Middle East, Crowd Analyzer is an
invaluable analytics and social media monitoring tool. As the first Arabic-
Crowd
focused social media monitoring platform, Crowd Analyzer analyzes “Arabic
Analyzer
content in terms of relevancy, dialect and sentiment.” It not only monitors
major social networks, but also blogs, forums, and news sites.

If content marketing is an important aspect of your Facebook marketing

strategy, consider 76Insights. This social media monitoring tool measures the
76Insights resonance of your social media content and breaks down your resonance
score, which measures how much social media engagement someone receives
after publishing something.

Keyhole lets you see what’s being said about you on Twitter and Instagram in
real-time. You can monitor keywords, hashtags, URLs, and usernames, and see
Keyhole
historical as well as real-time data. One cool feature is the heat maps that show
you activity levels around the world.

Digimind lets you track keywords in news outlets and social media platforms
for mentions of your company in real-time. It also measures sentiment, so you
Digimind can gauge whether what is being said about you is good, bad, or “meh.” You
can also compare how your company is perceived online against your
competitors.

Google Alerts lets you monitor the web for mentions of your company, your
competitors, or other relevant topics. Just go to the Google Alerts page, type a
Google Alerts keyword or phrase in the search box, and provide your email address to receive
a notification every time Google finds results relevant to your alert criteria. You
can set alerts for specific regions and languages.

410
 THE A.I. CRUISE LINE

SERVICE DESCRIPTION
On top of all the social media monitoring tools mentioned above, Hootsuite Pro
provides social listening capabilities right in the dashboard. Monitor specific
Hootsuite keywords, hashtags, regions, and more. Stay on top of what people are saying
about your brand and listen to your customers and competitors to gain
competitive advantage.

Table 23: Hootsuite’s Social Media Monitoring Tools

Source: 14 of the Best Social Media Monitoring Tools for Business561

Cruise lines should also be listening to comment boards or short-term blogging sites
like Tumblr or social news aggregation sites like Reddit for comments about their
company and their properties. Cruise line customers are often happy to post
wonderful reviews about their visits and purchases, and this is gold for word-of-
mouth marketing so cruise lines should do their best to motivate their customer to
provide reviews.
Once all of this data is captured, geofencing, location-based services, and
location-based advertising offer some unique ways to reach consumers when
they are in the all-important “decision mode.”
Personally, I am not a fan of checking into locations for something as ethereal as
a virtual badge, but that is only my opinion, many others seem to enjoy the
gamification aspects of this process. There are plenty of people who are more
than happy to give away their personal information for a virtual badge or an
opportunity to win something of value, so why not take them up on it? A cruise
line that is able to offer highly specific advertisements to customers who might
just need a little extra nudge to make that purchase should find an investment
in geofencing applications highly profitable.
Geofencing marketing does raise the issue of privacy, but I believe the day will
come when mobile users who enter a mall, a hotel, a cruise ship, or a shopping
area will view messages that ask for their permission to accept a location-based
ad as harmlessly as they currently view television and radio advertising. It is very
possible that they will embrace this form of advertising because of its immediacy.
Geotrigger services will probably be the next iteration of geofencing
applications. These send targeted location-based messages to app users who
either enter or leave a geofenced area. Geotriggers can send the right message
to the right person at the right time in the right place, which should increase
their use as well as prove ROI positive for anyone who makes the investment in
it.
“A smart combination of listening to the online conversation already taking
place, learning what people want, and then providing what they are open to
receive from the brand constitutes the winning ticket,” advises Macy and
Thompson.456 Whether the engagement is through video, online polls, games,

411
 ANDREW W. PEARSON

photo sharing, e-mail, blogging, PowerPoint presentations, or podcasting,

“engagement strategies present an opportunity for brands to align content
creation for social media with a company's priorities and involve cross-functional
interaction and collaboration,” argue Macy and Thompson.456 “Social media
engagement can also be used for front-end campaigns and appearances to help
guide the conversation and generate buzz,” conclude Macy and Thompson. 456

Rules Engine
The Rules Engine step is a pretty straightforward concept to understand; cruise lines
are already creating considerable business rules for their businesses and these rules
should be extended to each company’s defined rewards program, their reward’s
economy, and the marketing of the loyalty program.
Rewards can be as simple as a reward for a visit, a points threshold reached, a
birthday or anniversary, loyalty card utilization, or reaching a spending tier. Reward
rules engine must contain the conditions of the loyalty program, i.e., if the
member activity of a member fulfills the conditions, the loyalty engine executes
the assigned rule actions, which could be giving the member a unique offer
based on his or her spend.
At the very beginning of the book, I spoke about the difference between doing
things in a rules-based analytics way and an AI-powered way and I believe it’s
worth repeating the AI-powered way here8:
• Provide warnings whenever a company activity falls outside the norm.
o AI-powered analytics: The AI analytics tool automatically
determines that the event is worthy of an alert, then fires it off
unaided.
• Conduct a root cause analysis and recommend action.
o AI-powered analytics: Your tool automatically evaluates what
factors contributed to the event and suggests a cause and an
action.
• Evaluate campaign effectiveness:
o AI-powered analytics: The AI analytics tool automatically
weights and reports the factors that led to each successful
outcome and attributes credit to each campaign element or
step accordingly.
• Identify customers who are at risk of defecting:
o AI-powered analytics: Your tool automatically identifies which
segments are at greatest risk of defection.
• Select segments that will be the most responsive to upcoming
campaigns:
o AI-powered analytics: Your tool automatically creates
segments based on attributes that currently drive the desired
response.

412
 THE A.I. CRUISE LINE

• Find your best customers:

o AI-powered analytics: Your tool automatically identifies
statistically significant attributes that high-performing
customers have in common and creates segments that include
these customers for the business to take action on.
These should become the new rules of a cruise line’s IT and marketing
departments.

Automation
One of the huge benefits of automating campaigns is that offers based on either
stated or inferred preferences of customers can be developed. Analysis can
identify which customers may be more responsive to a particular offer. The
result: more individualized offers are sent out to the cruise line’s customers and,
because these offers tap into a customer’s wants, desires, needs and
expectations, they are more likely to be used; more offers used mean more
successful campaigns, means higher returns on investments.
By understanding what type of customer is on its ships, why they are there, and
what they like to do while they are there, a cruise operator can individualize its
marketing campaigns so that they can be more effective, thereby increasing the
cruise line’s ROI.
Once the customer leaves the property, the marketing cycle begins anew. RFM
models can project the time at which a customer is likely to return and social
media should be checked for any comments, likes, or uploads left by the
customer.
All of a customer’s captured information can now become part of the Master
Marketing Profile that will be the basis for future marketing efforts. Combining
the daily, weekly and monthly Master Marketing Profiles will also allow the
cruise line to develop insightful macro views of its data, views that could help
with facilities, labor management and vendor needs as well.

Moderation
Moderating boards and UGC posts create a double whammy for cruise lines
because, as Rachel Perlmutter explains in her article Why You Need Social Proof on
Your Website,562 “People need to see that others also enjoy that product. It’s
what we call social proof: the idea that buyers are influenced by the decisions
and actions of others around them.”
Perlmutter offers the following reasons why it is so important to have UGC
review on a business’ website562:
• “Testimonials add credibility for the products and services you offer.
• People tend to trust online reviews when making purchases.

413
 ANDREW W. PEARSON

• Social proof earns better SEO: Adds more favorable language

surrounding your brand online.
• When sourcing opinions from your client base, you show that you care
about their experience with your brand, thus strengthening the
relationship you have with your clients.”562
Perlmutter states that businesses can gather testimonials in a variety of ways,
including sending surveys to new clients.562 Perlmutter also advises businesses
to encourage buyers to post on social media.562 Cruise Lines should use hashtags
to track customers’ responses to the company’s products and services so they
can be easily found and responded to accordingly.562 Instagram should be a big
part of a cruise line’s strategy because testimonials with images trump text
testimonials alone. Testimonials are powerful examples of social proof as well. 562
Cruise lines shouldn’t be afraid to send free gifts of cruises or food to people with
large followings on platforms like Instagram, YouTube, etc. “Whether you want
to call them social media influencers, bloggers, or local celebrities, consider
getting meatier’ testimonials from people who have already gained some
amount of trust online,” states Permitter.562 “Some may ask for a small fee to
review your product, but the return you get from their article, video post,
Instagram picture, or even just their words and name listed on your site will likely
be tremendous,” advises Perlmutter.562
It is important to get instant reactions, too. Perlmutter recommends that, “If you
host events, then you have the prime opportunity to gather testimonials from
attendees right on the spot.”562 Don’t be afraid to set up a camera right outside
the event space and asks participants to provide their opinions on the spot. 562
In her Digiday article How Facebook is wooing luxury brands563, Bethany Biron
writes that “Facebook is advocating for ‘digitally influenced sales,’ that
assist consumers with the discovery process while still driving them to e-
commerce sites and physical stores, said Narain Jashanmal, industry manager of
retail at Facebook. This concept has helped major cruise lines like Barneys break
out of the traditional retail rut and embrace e-commerce.”
In her article Engage Customers and Gain Advocates Through Social Media and
Social Networking564, Wendy Neuberger argues that: “Social commerce is about
making a retailer’s brand a destination. Cruise lines really need to listen to what
their customers are saying. Customers can provide valuable input and feedback
that can be used to make more informed assortment decisions, changes to
website features and enhancements to the shopping experience.”
“When customers feel their voice is being heard, they feel a stronger connection
to the retailer and are more likely to become advocates.”564 Neuberger claims it
is important for retailers to identify and engage with the key influencers for
several reasons, the two most important being: “to empower their advocacy or
capabilities, which helps build and foster a sense of community among brand

414
 THE A.I. CRUISE LINE

loyalists, and empowers those loyalists to better advocate on behalf of a brand,

product and/or service.”564
Neuberger recommends that retailers use the following social media platforms
(but this is applicable to cruise lines as well)564:
• Blogs: retailers can provide additional product or category information
here as well as post how-to information in the form of text, photos
and/or videos. Retailers should also provide space for customers to add
feedback and/or comments about their retailing experience.
• Micro-blogging: coupons, sales and promotions can be offered through
these channels. Retailers can “‘tweet press releases, provide exclusive
tips and tricks to customers, and ask for customer feedback,
suggestions or ideas for improvements. Some cruise lines even use
Twitter as a customer service mechanism.”564
• Co-Shopping: this is a form of social shopping and it enables two people
— a customer and sales associate or two shoppers in different locations
— to share a joint shopping session using live instant messaging such as
Skype, WeChat or any number of other OTT services.
• Widgets: these are tiny applications that can be embedded into a
website, blog or social network that are portable and relatively
inexpensive to create.
• Social Bridging: anyone who has signed onto a website using their
Facebook, Pinterest or Twitter account knows what social bridging is.
“This level of authentication provides enough credentials to participate
in the social elements of the site. Additional authentication is required
to complete a shopping transaction due to the sensitivity of the content
included in a shopper’s account. Social bridging can be used to drive
traffic and engage existing and new customers. It can access a user's
identity, their social graph, and stream activities such as purchases and
other social participation on the cruise line’s site” explains
Neuberger.564
• In-Store Kiosks and Flat Panels can be provided to enable customers to
use social networking tools from within a store.564
In her article Facebook wants to become the new mobile storefront, unveils new
ad tools for brands and cruise lines565, Tanya Dua states that Facebook is trying
again to establish itself as a true shopping outlet. Facebook “believes it can play
a unique role in the shopping world — helping people both discover
new products and make decisions when they're ready to buy.” 565
“In a sense, we are becoming something like the new storefront,” Martin Barthel,
Facebook's global head of retail and ecommerce strategy, says, adding that,
“People discover products and then we redirect these people to mobile or
desktop properties of the cruise lines we are working with.”565

415
 ANDREW W. PEARSON

“Facebook believes that because people spend so much time on its mobile app,
it can lay claim to being able to help marketers pitch their products before people
know they even want them (like TV) and then help people find products when
they know their ready to pull the trigger on purchases (like Google and
Amazon),” notes Dua.565
“Facebook wants to be a solution not just at the very bottom of the marketing
funnel for solutions like retargeting, we actually want to create new purchase
intent and consideration further up,” said Graham Mudd, product marketing
director at Facebook.565 “If you look at 20 to 30 years ago, that was actually done
through broadcast media but in a feed-based environment we have the
opportunity to do that in a much more relevant way,” adds Mudd. 565
Facebook might be onto something here as “its own research has shown that
shoppers increasingly rely on Facebook and Instagram to find and purchase
products.”565 Facebook claims that, “Mobile-first shoppers in the U.S. are 1.7
times more likely to get inspiration for gifts or shopping ideas on Facebook, and
2.5 times more likely to research gift or shopping ideas on Instagram.”565
In 2018 and beyond, video will drive more online sales. When Facebook surveyed
20,824 mobiles shoppers across 17 markets, 30% of them said they preferred to
discover new products via video.565 This is why the “company is enhancing its
dynamic ads feature, which allows brands and cruise lines to upload videos to
show-off their products catalogues, instead of just static images.” 565
According to Dua, “Dynamic ads automatically promote products to people who
have expressed some interest in a brand, whether on its website, in its app or
anywhere else on the internet. The new video feature in dynamic ads has already
been trailed by cruise lines like made.com.”565
“Facebook has also introduced overlays for dynamic ads, a product which
enables brands to add price tags and visuals into their dynamic ads, touting
discounts and other offers,” adds Dua.565
One other item that should be noted, Facebook will now let retailers and brands
target consumers on Facebook based on households, rather than just as
individuals.565 According to Dua, “Facebook will allow marketers to create a new
‘household audience,’ which enables marketers to target to family members in
the same household, with the idea being to inspire members of their audience's
household to purchase.”565 Facebook believes that advertisers will be able to
“measure the impact of these ads, including whether they influence household
members who didn't actually see the ads to make purchases.” 565
In her article WhatsApp For Business–What Does It Mean?566, Holly Turner
explains that in August 2017, “WhatsApp announced it was experimenting with
verified business accounts on the platform, which would offer brands the
opportunity to communicate with its users; a platform the average user checks

416
 THE A.I. CRUISE LINE

23 times a day and which boasts 1.2 billion monthly active users.”
“Businesses can gain a verified green checkmark icon to indicate the authenticity
of the account, assuring users its legitimacy, alongside opening the door to the
platform’s previously walled off garden,” Turner adds.566
“The platform could be following in the footsteps of other messaging apps such
as Facebook Messenger and Kik, implementing a chatbot function to enable
WhatsApp users to ask businesses questions, make purchases and receive
instant bot responses.”566
The opportunity is considerable. WhatsApp business accounts offer up one more
channel for brands to send out simple automated messaging to users who opt
into their messages.566 “Whether it be discount codes, new products or brand
news, users could stay up to date with businesses they invest in and brands
would be presented with the opportunity to reach people on a platform that
sees 6 out of 10 users accessing the app on a daily basis,” argues Turner. 566
WhatsApp’s history will work in its favor. Having never allowed any form of
advertising previously, WhatsApp currently feels like a very intimate and private
environment for its users.566 Turner believes that, “content delivered to users
would, therefore, benefit from having an ‘organic’ feel; providing useful and
totally personalised content.”566
A good example of the potential can be seen in the Nike On Demand WhatsApp
service.566 It is a one-to-one messenger-based service that was “created to
connect athletes with Nike experts on a regular basis to keep them motivated
and on-track with their fitness goals.”566 The campaign delivers “personalised
content in the form of images, conversation, playlists, etc. as well as providing
expert advice from pacers and trainers all through the WhatsApp platform, akin
to a real peer’s motivational reminder,” explains Turner.566
“Whether WhatsApp intends to follow the crowd by implementing a chatbot
strategy or go against the grain to offer users something truly useful and
personalised will soon become clear,” adds Turner.566 “What is already very
clear, however, is the opportunity WhatsApp business accounts presents,
regardless of what strategy they choose, to reach inside the walled gardens of
messaging apps,” Turner concludes.566

Messaging
In the future, it is likely that all marketing will become interactive and the
consumer will become a participant rather than a “target audience”. As Shar
VanBoskirk states in his article US Interactive Marketing Forecast, 2007 To
2012567, “Instead of planning for a set ‘search budget’ or an ‘online video
campaign’, marketers will instead organize around ‘persona planning’ — that is,
they will plan around generating a desired response from a customer type. In

417
 ANDREW W. PEARSON

response to changing customer behavior, channel optimization will take place

on the fly, shifting between channels dynamically.”
Cruise line operators can also use social media to manage their brand, enhance
brand loyalty, as well as engage both their current customers and their potential
customers. The social media world is also the perfect place to harvest customer
feedback, provide real-time customer service, build fanbases, and drive traffic to
a cruise line’s website.
Blogs and micro-blogging sites are also important mobile and social media
channels and cruise lines should monitor Twitter feeds for both their satisfied
and unsatisfied customers.
Cruise lines should also feel compelled to reward their customers through
Facebook, Twitter, WeChat, and Weibo, or any number of blogging and micro-
blogging services. The beauty of using these channels is the ability of the
customer to share these stories with their friends and contacts. It wouldn’t be
that hard to do, either, as a cruise line operator can ask customers for their social
media accounts when they sign up to a loyalty program.
Cruise lines need to empower their customers to post on Facebook or WeChat
or Twitter or comment about their experience and, hopefully, turn them into
apostles. In Jones and Sasser’s zone of affection, satisfaction levels are high and
“customers may have such high attitudinal loyalty that they don’t look for
alternative service.”80 It is within this group that “Apostles” — members who
praise the firm in public — reside and this is the group that is responsible for
improved future business performance.80 A simple search of the Twitter feed on
the multiple services I mentioned in the previous chapters will probably reveal a
list of customers who could be courted for marketing purposes.
Facebook should be a part of every cruise line’s social and mobile media marketing
plan, but simply putting up a Facebook page won’t cut it these days; creativity and
uniqueness are needed to get noticed in today’s highly competitive social media
world. Gamification is also a good way to stand out from the crowd. Facebook bots
can also add a customer service channel that can answer common customer
questions quickly, efficiently and inexpensively.
Instead of filling newspapers and magazines with advertisements, a cruise line
operator should create databases of opted-in customers who have the
propensity to purchase goods in the future. By tracking OTT, SMS opt-in use, a
cruise line can get immediate and highly quantifiable data on who has signed up
for their loyalty cards, who has used a coupon, who has opted-in to their mobile
campaigns, and who might be planning to attend certain advertised events.
In her article Retailers Doing It Right in Social Media568, Cherise Luter advises that
whether done on Wordpress or Tumblr, blogging is an important part of a
retailer’s social media strategy. Luter adds568:

418
 THE A.I. CRUISE LINE

“Of all the places a company can build a lifestyle around their
brand, a blog is the best suited. On Tumblr, Labrotatoria's
Musing On... and The Classroom's Stay Classy Houston are
great examples. They both share inspirational style images,
product info and behind the scenes posts. CakewalkStyle Shop
has three blogs, yes three. Lifes a Cakewalk is its main blog,
Influencers focuses on bloggers and taste makers, and Style
Guide shares its latest store items and trends.”
What works for a retailer will, in many cases, also work for a cruise line. The best
part about being on Twitter or Weibo or any of the other instant messaging
services is the ability to interact with a customer in real time. 568 This, I believe,
can really be a game changer. A direct, two-way dialogue can be created, which
helps with engagement and, probably, sales. As Luter explains, this can help
“Resolve customer service issues, get a pat on the back, or valuable feedback all
from the comfort of home or the store backroom.”568
“The retailer Lilly Rain is really great about interacting with its customers online.
They retweet and reply to the messages and even re-post blogger websites that
show them love. They also have an active Google+ page, which is great for
Google searches,” Luter explains.568 Luter also advises that “Twitter should not
be ignored. Use Twitter to tweet out ‘instant sales.’ Get the word out about
leftover products, new products in limited supply, or last-minute sales. Believe it
or not — your customers ARE using Twitter. And if they see something great at
your store — they WILL retweet.”568
Table 24 reveals Pega System’s Next Best Action Advisor showing the typical use
case for marketing to cruise line patrons.
Use case Description
Pega Marketing provides the agent with the tools to ensure
Retention that they understand the value of the customer and the best
options to retain customers who are looking to churn.
Next Best Offer prioritized ranking is supported by all the
Cross-Sell/Upsell necessary arguments and collateral to convert the
opportunity.
This engagement type takes a more consultative approach to
Negotiation-Based selling by putting the customer in charge of the conversation
Selling while ensuring that the negotiation stays within the budgets
calculated for this customer.
Communicating to customers proactively to entice or
Product/Service stimulate product or service usage. Calculating the right
Usage Stimulation incentive for each customer will ensure a higher conversion
rate.

419
 ANDREW W. PEARSON

Use case Description

Renewal Targeting customers who are approaching key product events
Reminders/Retention to ensure they remain loyal customers.
Driving traffic to landing pages or microsites to capture
Newsletter, Article,
customer details and fuel lead capture and conversion
Content Marketing
processes.
Significant Daily Campaigns that seek out customers with important
Event/Anniversary anniversary dates with specific, personalized messages, e.g.,
Communication Birthday Campaign.
Either direct to customers or via prospect lists to drive growth
Acquisition
of the customer base.
Initial communications focusing on things like Welcome
On-Boarding packs, Sign-up for Auto-Pay, Payment reminders, and product
and service awareness.
Product/Service Product and service promotions to generate awareness or
Promotion education to support or coincide with product launches.
Targeting customers who partially engaged on with
Retargeting
appropriate messaging to reinforce the features and benefits.
Capturing customer details and other related data items and
using these to trigger other marketing messages to create the
Viral Marketing
potential for very high visibility and visibility of the company
messages.
Incentive-Driven, Using what we know to tailor individual communications for
Game-Oriented each customer to ensure they are relevant, timely, and
Marketing appropriate.
Keeping organizations in the mind of the customer at relevant
Seasonal Marketing
times of the year.
By packaging up multiple offers within a parent offer,
Bundling bundling enables organizations to communicate a tailored
package of offers to each customer.
Location-based Target customers based on their location using Geofence
marketing functionality.
Event Respond to real-time marketing opportunities by pushing
initiated/Trigger relevant offers to customers.
marketing
Significant Daily Campaigns that seek out customers with important
Event/Anniversary anniversary dates with specific, personalized messages, e.g.,
Communication Birthday Campaign.

Table 24: Pega Systems Marketing Use Case Examples72

420
 THE A.I. CRUISE LINE

Next-Best-Action Advisor72

Data & Analytics

In this final section of the customer journey, cruise lines should “acquire social
identity tied to customer records.” Neuberger argues that it is very important to
monitor the market conversation to understand what the marketplace is (or
isn’t) saying about a cruise line (their brand, products, services, etc.).564 Retailers
– and cruise lines – “need to understand the tone and impact of the conversation
and begin to identify areas of opportunity for helping shape that conversation
and gather valuable market intelligence,” says Neuberger.564
As previously mentioned, ROI is not that tricky of a thing to measure with social
media, quite the contrary. Today, the endless search for Facebook fans should
be replaced by short-term campaign ROI as the main measure for individual
campaigns. Cruise lines should look at correlation analysis between activities,
engagement and sales, which might be unsettling for some traditional
marketers, but the reward should be worth the time and the effort.45
“The explicit use of active and control groups, and experimentation of using
different treatments will help marketers understand the impact of specific SM
activities.”45 More direct marketing type disciplines will be required, in a world
where there is real-time feedback on attitude and behavior and a plethora of
data.”45 This has become a much more demanding world in terms of capturing
and utilizing all of this data, but making the effort to turn this data into actionable
intelligence will be noticed by fickle consumers, I have no doubt.
Cruise lines should look to assign a percentage value to social media so that a
true attribution measurement can be created. Values should be ascribed to
social media for being the site of new customer contact or for numbers of
positive reviews by current customers. These are all important metrics to know
because a highly followed influencer might not be spending that much at your
store, but their followers might be.
For Neuberger, “Social media metrics include sentiment, activity, share-of-voice,
and thematic content of online conversations. Trends and key influencers
(“mavens”) and the most active sites/blogs are identified and tracked. By
understanding the impact, retailers will have a way of identifying measurable
progress, quantifying the return on social media investment, and enabling
benchmarking against future efforts.”564 In this case, what goes for retailers also
goes for cruise lines.
Another good example of Big Data use in retail is profiled in Bernard Marr’s Big
Data in Practice.57 In it, Marr describes the story of Pendleton & Son, a local
butcher based in north-west London.57 Established in 1996, the butcher shop had
enjoyed a steady customer base and good reputation for years, but in 2014, a
supermarket chain store moved onto the same street, and it affected overall

421
 ANDREW W. PEARSON

footfall and hit revenues hard.57

As Marr explains, “While founder Tom Pendleton was certain his shop offered
superior quality and choice compared to the supermarket, the trouble was
conveying this message to the public and getting customers through the door.
Trying to compete on price wasn’t working and, with falling income, son Aaron
Pendleton turned to data to help keep the business afloat.”57
A Big Data consultant suggested “installing simple, inexpensive sensors inside
the store window to monitor footfall and measure the impact of window displays
and promotions.”57 Using this sensor data as well as internal data such as
transaction and stock data, the butcher shop was able to “measure how many
people walked past the shop, how many stopped to look at the window display
and sandwich board sign and how many people then came into the store as a
result.”57 Armed with this knowledge, the butcher shop was able to refine its
displays and messaging based on what most interested customers.57
The insight on one’s current business was almost eclipsed by sensor data that
pointed to an unexpected and potential new revenue stream.57 “As two popular
pubs were located at the end of the street, the hours of 9 p.m. to midnight
proved particularly busy in terms of passers-by — almost as many as the busy
lunchtime period. So the Pendletons decided to trial opening at night and serving
premium hot dogs and burgers to hungry folk making their way home from the
pub.”57
Analytics was even used to decide menu items; “In order to decide on what
products to offer at night, Aaron analyzed trend data from Google Trends to see
what food items were particularly popular. This led to the creation of their pulled
pork burger with chorizo.”57
Going forward, the Pendletons were hoping to expand their use of data in order
to increase their knowledge of customers even further. 57 Weather data is now
included to predict demand even more accurately and a customer loyalty app
was in the works.57 Pendleton & Sons are a perfect example of how a company
can step into analytics and they are, literally, being richly rewarded for their leap
of faith.57
The data revealed some interesting behavioral information on their customers
as well.57 “In short, the Pendletons found that local customers favoured
inspiration and ideas over cheap deals, which were available every day in the
supermarket. They were able to use this insight to improve their messaging and
get more people through the door — and those who entered the shop were far
more likely to make a purchase as a result.”57
In addition, the late-night openings proved enormously popular and the
company decided to make this a permanent feature on Friday and Saturday
nights. Not only did this provide much-needed additional revenue, it also

422
 THE A.I. CRUISE LINE

introduced the company and their products to a whole new set of customers.
Besides data that is collected by a cruise line, firms can also purchase data from
third parties. According to Bourreau et al, data can be obtained from data
brokers, which have been defined by the US FTC as, “companies whose primary
business is collecting personal information about consumers from a variety of
sources and aggregating, analyzing, and sharing that information, of information
derived from it, for purposes such as marketing products, verifying an
individual’s identity, or detecting fraud.”178 Companies like Acxiom, Corelogic,
Datalogix, eBureau, ID Analytics, Intellius, PeekYou, Rapleaf, and Recorded
Future fit under the FTC categorization and could sell useful data to cruise lines
for their customer acquisition initiatives.178

The Future is Now

According to Kyle Wiggers in VentureBeat’s article Mews Systems raises $33
million for hotel and hostel property management tools 569, “Mews’ cloud-based
and mobile-first tool suite aims to simplify tasks like registration while playing
nicely with legacy systems. To this end, Operator — Mews’ reception desk
product — allows guests to register with tablets and takes most forms of
payment.” “Customers provide credit or debit card details at check-in, at which
point their accounts are either charged or preauthorized in order to cut down on
chargebacks,” says Wiggers.569
“Mews claims its six-step Distributor booking engine generates some of the
industry’s highest conversion rates and that 40% of guests choose to check in
online so they can skip queues at reception,” contends Wiggers.569
“From within Mews’ bespoke multi-property backend, managers get an
overview of departments from reservations to maintenance, and they’re able to
generate reports and schedule report export jobs,” says Wiggers.569
“The Marketplace extends Mews’ capabilities further by letting property owners
plug in popular apps, tools, and services, from accounting software like
Bookboost and Basware to guest technology platforms such as I Am Max and
Acentic,” adds Wigger.569 “The breadth of support extends to reputation
management solutions like GuestRevu and to software that addresses pain
points in facility management (e.g., 4Suites, Leviy, Room Checking), point of sales
(Lightspeed), distribution (SiteMinder), upselling (GuestJoy, HotelFlex), event
management (LetShare), business intelligence (Hotellistat), and more,”
concludes Wiggers.569
“With services like Hilton’s digital check-in, guests can choose their specific room
before arrival, just like selecting a seat on an airplane,” says Hollander.520
“Now that hotels have mastered Facebook and Twitter, there are new platforms

423
 ANDREW W. PEARSON

to focus on. Hoteliers can reach additional audiences with strategic use of TikTok
videos and Instagram stories, for example,” says Hollander.520 I would add sites
like Pinterest, WhatsApp, Vero, LinkedIn, Quora, Periscope, as well as keep an
eye on the Chinese social apps, which can explode to millions of users almost
overnight.
“More and more, guests are completing the entire booking process on their
smartphones, so hoteliers must ensure their websites are mobile-friendly,” adds
Hollander.520
“WiFi isn’t a one-speed-fits-all amenity anymore; if your internet provider hasn’t
made any upgrades in a few years, your WiFi might be too slow to stream movies
or download large files. 50 megabits per second (mbps) is the threshold for
adequate WiFi these days, and some providers, like Google Fiber, offer up to
1000 mbps,” says Hollander.520
As someone who travels between the worst Asia has to offer in terms of internet
speeds – the Philippines – and the best Asia has to offer – South Korea – it’s very
clear how important fast WiFi is to a businessperson. For someone who uses his
laptop as his office, having fast WiFi is the difference between a highly
productive day and a day of frustration and wasted time.
“First we swiped, then we inserted the chip, and now we can pay by tapping a
credit card or mobile wallet, so hotels must upgrade their payment
technology to accept payments via near field communication (NFC),” says
Hollander.520
In China right now, QR codes are being used to make payments, but they might
soon be replaced by facial recognition. In her article Forget the QR code. Facial
recognition could be the next big thing for payments in China 570, Yen Nee Lee
writes that, “technology giants such as Tencent are now studying the use of facial
and fingerprint recognition for such transactions.”570
“In China, payment methods using QR codes have replaced cash and cards in just
five years. It’s possible that in the next few years, new and better products could
emerge to replace QR codes,” said Greg Geng, vice president of Tencent’s
WeChat Business Group.570
In August 2019, WeChat Pay “introduced its ’Frog Pro’ system that allows
customers to make payments by simply scanning their faces — without the use
of their mobile phones,” says Yen.570 “The technology is now being tested in
several retail chains in China and came after Alipay rolled out its own facial
recognition payment system, the ‘Dragonfly,’ last year,” she adds.570
Today, “Hotels collect a slew of data about guests, but hoteliers rarely use that
data to personalize the guest experience. With new customer relationship
management tools (CRM) like Revinate and Cendyn, hotels can pull data points
out of the cloud and into the guestroom to create a more tailored experience,”

424
 THE A.I. CRUISE LINE

says Hollander.520
AR & VR can also help market a hotel or an IR. “Travelers want to know
everything about a hotel before booking it, so what could be better than a virtual
reality tour of a guestroom? In addition to VR, augmented reality will allow
guests to experience a hotel by simply strapping on a headset,” explains
Hollander.520
“In a world so connected by business and economic ties, it only makes sense that
globalization would have implications in the hotel industry,” claims Hollander.520
“As globalization drives incomes in countries around the world, more people can
afford to travel, which means that hotels face opportunities and challenges that
come with accommodating new travelers from different places,” contends
Hollander.520 “Along with this rising middle class, increased income inequality
further distances the highest earners from the rest,” states Hollander.520 “Luxury
travelers continue to have an appetite for over-the-top experiences, so high-end
hotels must continuously come up with creative and innovative offerings to wow
their guests,” warns Hollander.520
The A.I. Cruise Line is a cruise line that takes into account how a patron who walks
through the door of an integrated resort, or onto a cruise line bus, or is even
driving down I-15 toward Vegas will affect every facet of a cruise line company.
It follows a customer before arrival, through his or her entire stay, then keeps
tabs on them once they leave.
As Kahle states, eventually we're going to set a time frame on the sales funnel
that never expires.361 From the moment of first contact, when the cruise line’s
systems capture an IP Address, through capturing the social ID, to understanding
the social activity, all the way through to the patron card sign up process so that
the cruise line can understand gaming and commerce behavior. The only thing
remaining is to capture post-transaction information if and when it comes in.
Once all of this data is captured, geofencing, location-based services, and
location-based advertising offer some unique ways to reach consumers when
they are in the all-important “decision mode.” A cruise line company that is able
to offer highly specific advertisements to customers who might just need a little
extra nudge to make that purchase should find an investment in geofencing
applications very profitable.
In The A.I. Cruise Line, a patron can sign into the cruise line WeChat account or
on a standard cruise line-branded app and he or she will be able to pull up his
player card points balance as well as all kinds of information about what is
happening on-property. The patron will be able to receive and use coupons at
onsite restaurants and bars, as well as to sign up for gaming tournaments.
Several cruise lines currently have branded games and/or apps that let patrons
play for free or for money and trade in game points for real rooms or free food.

425
 ANDREW W. PEARSON

By giving customers instant access to the information they need when they need
it most, a cruise line can enhance their patron’s on-property experience.
Whether the cruise line’s patrons are on-property to play baccarat, poker,
blackjack, slots, bingo, or if they want to gamble in the cruise line, these instant
messaging services can provide a patron with information that can not only
enhance their experience but, potentially, shape it.
As previously mentioned, perhaps one of the best uses of location-based services
is in the MICE space. The massive size of some exhibition halls can make finding
a particular booth or floor section a daunting proposition. Indoor mobile
communication technology with location awareness technology can help
conference-goers navigate a vast conference floor.321
Before arriving at a conference, a mobile user would be able to register his
personal preferences and, once he enters the exhibition hall, a route map would
be sent to his mobile phone. Vendor appointments could even be set up so that
they are located near each other so that the conference-goer wouldn’t have to
run around frantically trying to make meetings that are spread out all over the
convention floor.321
In their paper Marketing Analytics for Data-Rich Environments191, Wedel and
Kannan, argue that there are many promising areas of research that businesses
should keep on their radar. The technological developments of tomorrow might
just be some of the ones discussed in Table 29. At the beginning of the book, I
quoted Ramsbotham et al., who stated that, “Genuine success with AI — over
time — depends on generating revenue, reimagining organizational alignment,
and investing in the organization’s ability to actually use AI across the
enterprise.”7

DATA DETAILS
• Behavioral targeting with cross-device data; mobile,
location-based, and social analytics.
• Fusing data generated within the firm with data
Structured data generated outside the firm; integrating “small stats”.
Combining machine learning approaches with
econometric and theory-based methods for big data
applications; computational solutions to marketing
models for big data.

426
 THE A.I. CRUISE LINE

DATA DETAILS
• Development of diagnostic, predictive, and prescriptive
approaches for analysis of large-scale unstructured data.
• Approaches to analyze unstructured social, geo-spatial,
mobile data and combining them with structured data in
Unstructured data
big data contexts.
• Using, evaluating, and extending deep learning methods
and cognitive computing to analyze unstructured
marketing data.
• Aligning analysis of disaggregate data with that of
aggregate data and including unstructured data in the
analysis of the marketing mix.
• New techniques and methods to accurately measure the
impact of marketing instruments and their carryover and
Marketing-mix
spillover across media and devices using integrated path-
modeling
to-purchase data.
• Dynamic, multi–time period and cross-category
optimization of the marketing mix.
• Approaches to incorporate different planning cycles for
different marketing instruments in media-mix models.
• Automated closed-loop marketing solutions for digital
environments; fully automated marketing solutions.
• Personalization and customization techniques using
Personalization cognitive systems, general artificial intelligence, and
automated attention analysis; personalization of
content.Mobile, location-based personalization of the
marketing mix.
• Methods to produce and handle data minimization and
data anonymization in assessing marketing-mix
effectiveness and personalization.
Security and privacy
• Distributed data solutions to enhance data security and
privacy while maximizing personalized marketing
opportunities.

Table 25: Area of Focus Promising and Important Issues for Research
Source: Wedel and Kannan, Marketing Analytics for Data-Rich Environments

In Greek mythology, once Pandora opened the box and let out all of the evils of
the world, the one thing remaining inside was hope. Now, hope is
unquestionably what the cruise line industry is selling, the hope of getting rich,
and the hope of leaving all your troubles behind. Of course, this only happens
for a small minority, but most of the others at least get entertained and that’s
not usually free.
Today, more and more cruise line companies are referring to themselves as

427
 ANDREW W. PEARSON

“entertainment companies” rather than gaming companies, not only because of

the stigma attached to gambling and wagering, but also because of the new
reality of the business; i.e., it is a multi-faceted, multi-offering business, and
gambling is only one piece of a giant pie. Admittedly, a substantial piece of the
pie, especially in places like Macau, but it’s not the entire property’s raison
d'être. More and more, gambling is becoming entertainment and cruise line
companies must deal with customers interested in a multitude of entertainment
opportunity, not just the ones who want to gamble.
I wrote this book hoping that it would help clients and potential clients discover
the treasure that is hidden within their data, so that they shouldn’t feared the
incredible technology that is out there; technology that can make every cruise
line customer’s instore or online visit a personalized adventure that will make
them want to return again and again to their favorite cruise line and cruise line.
This, hopefully, brings a smile to every cruise line executive’s face, as well as a
healthy return on investment to the cruise line’s coffers.

503 Gourévitch, Antoine, Faeste, Lars, Baltasis, Elias, and Marx, Julien. May 23, 2017.
Data-Driven Transformation: Accelerate at Scale Now. Boston Consulting Group.
https://www.bcg.com/en-in/publications/2017/digital-transformation-transformation-
data-driven-transformation.aspx (Accessed 8 September 2017).
504 https://aws.amazon.com/lex/ (Accessed 11 March 2019).
505 https://aws.amazon.com/transcribe/ (Accessed 11 March 2019).
506 https://aws.amazon.com/comprehend/ (Accessed 11 March 2019).
507 Yakuel, Pini. Optimove. Turning Up Your Brand’s Voice to Reach the Most Advanced
Customers. October 25, 2018. https://www.optimove.com/blog/using-digital-assistant-
to-help-market-your-brand (Accessed 14 April 2019).
508 Maruti Tech. 14 Powerful Chatbot Platforms. https://www.marutitech.com/14-
powerful-chatbot-platforms/ (Accessed 8 April 2019).
509 https://chatfuel.com (Accessed 8 April 2019).
510 https://botsify.com (Accessed 8 April 2019).
511 https://flowxo.com/ (Accessed 8 April 2019).
512 Baron, Justine. SAP.com. Recast.AI will be renamed SAP Conversational AI early 2019!
November 15, 2018. https://cai.tools.sap/blog/recast-ai-renaming/ (Accessed 14 April
2019).
513 https://www.octaneai.com/ (Accessed 14 April 2019).
514 Rowe, Kevin. (2017). Search Engine Land. How to use AI for link building and improve
your search rankings. September 26, 2017. https://searchengineland.com/use-ai-link-
building-improve-search-rankings-283150 (Accessed 14 April 2019).
515 Facial recognition: Who’s Tracking You In Public. (December 30, 2015) Consumer
Reports. Online: http://www.consumerreports.org/privacy/facial-recognition-who-is-
tracking-you-in-public1/ (Accessed 25 November 2017).
516 Riddell, Chris. Qantas have seen their Future. It’s Facial Recognition. 1 August 2017.
Chrisriddell.com. http://chrisriddell.com/qantas-future-facial-recognition/ (Accessed 9
October 2018).

428
 THE A.I. CRUISE LINE

517 Butterfield, Brett. (2018). Adobe. See It, Search It, Shop It: How AI is Powering Visual
Search. 12 December 2018. https://theblog.adobe.com/see-it-search-it-shop-it-how-ai-
is-powering-visual-search/ (Accessed 20 January 2019).
518 Johnson, Lauren. 2017. Pinterest Is Offering Brands Its Visual Search Technology To
Score Large Ad Deals. Adweek. October 2, 2017.
http://www.adweek.com/digital/pinterest-is-offering-brands-its-visual-search-
technology-to-score-large-ad-deals/ (Accessed 18 November 2017).
519 Matt Cimaglia, Matt. (2019). Entrepreneur. The Future of Video Advertising is
Artificial Intelligence. December 12, 2018.
https://www.entrepreneur.com/article/323756 (Accessed 4 January 2019).
520 Deshpande, Indrajeet. (2019). MarTech Advisor. Top 5 Programmatic Advertising
Platforms for 2020 and Beyond. September 4, 2019.
https://www.martechadvisor.com/articles/ads/top-5-programmatic-advertising-
platforms-for-2020-and-beyond (Accessed 18 December 2019).
521 Fuller, Brian. 2018. Arm Community. AI’s role in next-generation voice recognition.
IoT Blog. https://community.arm.com/iot/b/blog/posts/artificial-intelligence-is-
changing-voice-recognition-technology (Accessed 12 January 2019).
522 Kinsella, Bret. 2019 Predictions from 35 voice industry leaders. January 1, 2019.
Voicebot.ai. https://voicebot.ai/2019/01/01/2019-predictions-from-35-voice-industry-
leaders/ (Accessed 21 January 2019).
523 Olson, Christi. (2018). Marchtech Today. Voice search isn’t the next big disrupter,
conversational AI Is. October 11, 2018. https://martechtoday.com/voice-search-isnt-the-
next-big-disrupter-conversational-ai-is-226537 (Accessed 3 March 2019).
524 https://aws.amazon.com/polly/ (Accessed 11 March 2019).
525 SeoClarity. The Next Generation of Search: Voice.
https://go.seoClarity.net/hubfs/docs/research/seoClarity_whitepaper_next-generation-
search-voice.pdf (Accessed 16 April 2019).
526 https://www.dialogtech.com/blog/search-marketing/voice-search-statistics
527 Batra, Varun. Marchtechseries. How AI is Driving a New Era of TV Advertising. March
29, 2018. https://martechseries.com/mts-insights/guest-authors/how-ai-is-driving-a-
new-era-of-tv-advertising/ (Accessed 12 April 2019).
528 Ku, Chih-Hao, Chang, Yung-Chun, Wang, Yichung, Chen, Chien-Hung, Hsiao, Shih-Hui.
(2019). Artificial Intelligence and Visual Analytics: A Deep-Learning Approach to Analyze
Hotel Reviews & Responses. January 2019. DOI: 10.24251/HICSS.2019.634 Conference:
52nd Hawaii International Conference on System Sciences (HICSS 2019).
https://www.researchgate.net/publication/327673310_Artificial_Intelligence_and_Visu
al_Analytics_A_Deep-Learning_Approach_to_Analyze_Hotel_Reviews_Responses
(Accessed 15 November 2019).
529 Yu, Y., Li, X., and Jai, T.-M. (Catherine). The Impact of Green Experience on Customer
Satisfaction: Evidence from Tripadvisor. International Journal of Contemporary
Hospitality Management 29, 5 (2017), 1340–1361.
530 TripAdvisor Q1 2018 results, http://ir.tripadvisor.com/static-files/90504393-5f9d-
4ac8-a5d7-b67d8b4adb60 (Accessed 5 December 2019).
531 The Final Word: TripAdvisor Management Responses,
https://www.tripadvisor.com/TripAdvisorInsights/w637 (Accessed 5 December 2019).
532 Pantelidis, I.S. Electronic Meal Experience: A Content Analysis of Online Restaurant
Comments. Cornell Hospitality Quarterly 51, 4 (2010), 483–491.

429
 ANDREW W. PEARSON

533 Schuckert, M., Liu, X., and Law, R. Hospitality and Tourism Online Reviews: Recent
Trends and Future Directions. Journal of Travel & Tourism Marketing 32, 5 (2015), 608–
621.
534 Öğüt, H. and Taş, B.K.O. The Influence of Internet Customer Reviews on the Online
Sales and Prices in Hotel Industry. The Service Industries Journal 32, 2 (2012), 197–214.
535 Gavilan, D., Avello, M., and Martinez-Navarro, G. The Influence of Online Ratings and
Reviews on Hotel Booking Consideration. Tourism Management 66, (2018), 53–61.
536 Sun-Young Park and Jonathan P. Allen. Responding to Online Reviews: Problem
Solving and Engagement in Hotels. Cornell Hospitality Quarterly 54, 1 (2013), 64–73.
537 Xie, K.L., Zhang, Z., Zhang, Z., Singh, A., and Lee, S.K. Effects of Managerial Response
on Consumer eWOM and Hotel Performance: Evidence from Tripadvisor. International
Journal of Contemporary Hospitality Management 28, 9 (2016), 2013–2034.
538 Mauri, A.G. and Minazzi, R. Web Reviews Influence on Expectations and Purchasing
Intentions of Hotel Potential Customers. International Journal of Hospitality Management
34, (2013), 99–107.
539 Zhang, J.J. and Mao, Z. Image of All Hotel Scales on Travel Blogs: Its Impact on
Customer Loyalty. Journal of Hospitality Marketing & Management 21, 2 (2012), 113–131.
540 King, R.A., Racherla, P., and Bush, V.D. What We Know and Don’t Know About Online
Word-of-Mouth: A Review and Synthesis of the Literature. Journal of Interactive
Marketing 28, 3 (2014), 167–183.
541 Fang, B., Ye, Q., Kucukusta, D., and Law, R. Analysis of the Perceived Value of Online
Tourism Reviews: Influence of Readability and Reviewer Characteristics. Tourism
Management 52, (2016), 498–506.
542 Sparks, B.A., So, K.K.F., and Bradley, G.L. Responding to Negative Online Reviews: The
Effects of Hotel Responses on Customer Inferences of Trust and Concern. Tourism
Management 53, (2016), 74–85.
543 Noone, B.M. and McGuire, K.A. Pricing in a Social World: The Influence of Non-Price
Information on Hotel Choice. Journal of Revenue and Pricing Management 12, 5 (2013),
385–401.
544 Liu, D., Li, Y., and Thomas, M.A. A Roadmap for Natural Language Processing Research
in Information Systems. Proceedings of the 50th Hawaii International Conference on
System Sciences, (2017).
545 Smith, A.K., Bolton, R.N., and Wagner, J. A Model of Customer Satisfaction with
Service Encounters Involving Failure and Recovery. Journal of Marketing Research 36, 3
(1999), 356–372.
546 Martin-Fuentes, E. Are Guests of the Same Opinion as the Hotel Star-Rate
Classification System? Journal of Hospitality and Tourism Management 29, (2016), 126–
134.
547 Gu, B. and Ye, Q. First Step in Social Media: Measuring the Influence of Online
Management Responses on Customer Satisfaction. Production and Operations
Management 23, 4 (2014), 570–582.
548 Law, R., Fong, D.K.C., Chan, I.C.C., and Fong, L.H.N. Systematic Review of Hospitality
CRM Research. International Journal of Contemporary Hospitality Management 30, 3
(2018), 1686–1704.
549 Leung, D., Law, R., Hoof, H. van, and Buhalis, D. Social Media in Tourism and
Hospitality: A Literature Review. Journal of Travel & Tourism Marketing 30, 1–2 (2013),
3–22.

430
 THE A.I. CRUISE LINE

550 Liu, X., Schuckert, M., and Law, R. Can Response Management Benefit Hotels?
Evidence from Hong Kong Hotels. Journal of Travel & Tourism Marketing 32, 8 (2015),
1069–1080.
551 Kardes, F.R., Posavac, S.S., and Cronley, M.L. Consumer Inference: A Review of
Processes, Bases, and Judgment Contexts. Journal of Consumer Psychology 14, 3 (2004),
230–256.
552 Andreassen, T.W. Antecedents to Satisfaction with Service Recovery. European
Journal of Marketing 34, 1/2 (2000), 156–175.
553 Xie, K.L., Zhang, Z., and Zhang, Z. The Business Value of Online Consumer Reviews and
Management Response to Hotel Performance. International Journal of Hospitality
Management 43, Supplement C (2014), 1–12.
554 Kim, W.G., Lim, H., and Brymer, R.A. The Effectiveness of Managing Social Media on
Hotel Performance. International Journal of Hospitality Management 44, Supplement C
(2015), 165–171.
555 Zhang, X. and Yu, Q. Hotel Reviews Sentiment Analysis Based on Word Vector
Clustering. 2017 2nd IEEE International Conference on Computational Intelligence and
Applications (ICCIA), (2017), 260–264.
556 Chong, E., Ngai, E.W.T., Lee, F., Li, B., and Chong, A.Y.L. Predicting Online Product
Sales Via Online Reviews, Sentiments, and Promotion Strategies: A Big Data Architecture
and Neural Network Approach. International Journal of Operations & Production
Management 36, 4 (2016), 358–383.
557 Glauber Eduardo, de O.S. Worldwide Hedonic Prices of Subjective Characteristics of
Hostels. Tourism Management 52, (2016), 451–454.
558 Ye, Q., Law, R., Gu, B., and Chen, W. The Influence of User-Generated Content on
Traveler Behavior: An Empirical Investigation on the Effects of E-Word-of-Mouth to Hotel
Online Bookings. Computers in Human Behavior 27, 2 (2011), 634–639.
559 Berger, J., Sorensen, A.T., and Rasmussen, S.J. Positive Effects of Negative Publicity:
When Negative Reviews Increase Sales. Marketing Science 29, 5 (2010), 815–827.
560 Vermeulen, I.E. and Seegers, D. Tried and Tested: The Impact of Online Hotel Reviews
on Consumer Consideration. Tourism Management 30, 1 (2009), 123–127.
561 Mathison, Rob. Hootsuite. (2017). 14 of the Best Social Media Monitoring Tools for
Business. August 14, 2017. https://blog.hootsuite.com/social-media-monitoring-tools/
(Accessed 23 November 2017).
562 Perlmutter, Rachel. (2016) Why You Need Social Proof on Your Website.
Entrepreneur.com. July 6, 2016. https://www.entrepreneur.com/article/296644
(Accessed 23 November 2017).
563 Biron, Bethany. (2017). How Facebook is wooing luxury brands. Digiday.
https://digiday.com/media/bringing-retail-speed-feed-facebooks-quest-court-luxury-
brands/ (Accessed 24 November 2017).
564 Neuberger, W. (n.d.). Engage Customers and gain advocates through social media and
social networking. January 24, 2013.
ftp://public.dhe.ibm.com/software/solutions/soa/newsletter/2010/newsletter-mar10-
article_social_media.pdf (Accessed 13 November 2017).
565 Dua, Tanya. (2017). June 26, 2017. Business Insider. Facebook wants to become the

new mobile storefront. http://www.businessinsider.com/facebook-wants-to-become-

the-new-mobile-storefront-2017-6 (Accessed 27 December 2017).
566 Turner, Holly. 2017. Whatsapp For Business—What Does it Mean? M&C Satchi
Mobile. http://www.mcsaatchimobile.com/whatsapp-business-mean/ (Accessed 23
November 2017). 431
 ANDREW W. PEARSON

567 VanBoskirk, S. (2007). US Interactive Marketing Forecast, 2007 To 2012 for Interactive
Marketing Professionals. Forrester Research.
568 Luter, C. (2014, March 10). Retailers Doing It Right On Social Media. Retrieved from
HoustonPressBlogs: http://blogs.houstonpress.com/artattack/2014/03/cruise
lines_doing_it_right_on_so.php?page=2 (Accessed 13 November 2017).
569 Wiggers, Kyle. (2019). VentureBeat. Mews Systems raises $33 million for hotel and
hostel property management tools. August 29, 2019.
https://venturebeat.com/2019/08/29/mews-systems-33-million/ (Accessed 10
December 2019).
570 Yen Nee Lee. CNBC. (2019). Forget the QR code. Facial recognition could be the next
big thing for payments in China. 19 November 2019.
https://www.cnbc.com/2019/11/19/tencents-wechat-china-may-soon-use-facial-
recognition-for-payments.html (Accessed 10 December 2019).

432
THE A.I. CRUISE LINE

433
ANDREW W. PEARSON

434
 THE A.I. CRUISE LINE

ABOUT THE AUTHOR

ANDREW PEARSON was born in Pakistan, grew up in Singapore, and was
educated in both England and America. With a degree in psychology from UCLA,
Pearson has had a varied career in IT, marketing, mobile technology, social
media, and entertainment.
In 2011, Pearson relocated from Los Angeles to Hong Kong to open Qualex Asia
Limited, bringing its parent company's software consulting experience into the
ASEAN region.
In 2016, Pearson founded his own consulting company, Intelligencia Limited, and
he is currently the managing director, overseeing worldwide operations, with
clients in the ASEAN region, the Middle East, North America, Australia, and
Eastern Europe.
Intelligencia is a leading analytics, AI, BI, CX, digital marketing and social media
company, implementing complex customer experience and personalization
solutions for clients like The Venetian Macau, Galaxy Macau, Genting HK, Tatts
Lottery, Tabcorp, Resorts World Retailer NY, Mexico’s Logrand Group, Junglee
Games, and Macau Slot.
In 2010, Pearson published The Mobile Revolution, and, in 2013, Pearson was
invited to write a chapter in Global Mobile: Applications and Innovations for the
Worldwide Mobile Ecosystem, a book on mobile technology that was co-
authored by several of the mobile industry's leading figures.
The first in the Predictive series — The Predictive Cruise line — was published in
2017, while The Predictive Sports Book, The Predictive Cruise line, and The
Predictive Airliner followed in 2018. The A.I. Marketer and The A.I. Cruise Line
started a completely new series.
Pearson is also a noted columnist, authoring articles on such topics as analytics,
AI, smart technology, Chinese social media, esports, and cloud technology.
Pearson has written for such publications as ComputerWorld HK, The Mobile
Marketer, and The Journal of Digital and Social Media Marketing, where he is
also a contributing editor. Pearson is the president of the Advanced Analytics
Association of Macau and one of the founders of Grow uP eSports, a Macau
association that promotes esports throughout the world, including the
GirlGamer Festival.
An avid traveler, Pearson is a sought-after speaker on such disparate topics as AI
and machine learning, data privacy, analytics, gaming, social media, and esports.
In 2019, he spoke at G2E Asia, at Forbes Middle East’s Digital Trends and Big Data
Big Trends events in Dubai, as well as Travel Tech in Dubai. 2020 will see him

435
 ANDREW W. PEARSON

back in Dubai for AI in Gaming, in Jeddah for Travel Tech, and in Latvia for the
Retailer Operations Summit
If Pearson's not pounding the pavements of Hollywood, he's probably wandering
the labyrinthine streets of Hong Kong's Lang Kwai Fong, or tearing up useless
betting slips at Happy Valley (perhaps the most perfectly named racecourse in
the world (for some)), or haggling in a Hyderabad street market, or dining at a
hawker center in Singapore, or marveling at the historical importance of
Moscow’s Red Square, or grabbing a Denny’s Grand Slam breakfast as the sun
sets over the Birj Khalifa in Dubai, or doubling down at the gaming tables in
Macau. Basically, Pearson's out there trying to find the next great story that the
world doesn't yet know that it desperately wants to see…
Social Media
LinkedIn: andrew-pearson-96513a3
Amazon: www.amazon.com/Andrew-Pearson/e/B005M5ACG0
Google Scholar:
scholar.google.com/citations?user=jmkqGowAAAAJ&hl=en&authuser=1
Twitter: intelligentsia.AI
Academia: PearsonAndrew
Blog: medium.com/@intelligentsiaf

436
 THE A.I. CRUISE LINE

INDEX

Activity tracing, 189 Astra beer, 219

Adobe, 47, 54, 69, 235, 238, 270, 271, Attribution analysis, 55, 86
272 Augmented reality, 15, 48, 354, 355,
Auto Crop, 237 356, 357, 373
Auto Tag, 390 wearable, 355
Deep Cutout, 237 Augmented search, 385
Adobe Advertising Cloud, 178, 180, Backlinks, 228
240 Baidu, 227
Adobe Enterprise Content, 238, 239 Behavioral information, 81, 128
Adobe Experience Cloud, 53, 236, 389 Behavioral patterns, 79
Adobe Sensei, 15, 54, 235, 236, 237, Big Data, 1
238, 390 Blogs, 34, 269
Auto tagging, 237 Bluetooth, 192
Adobe Sensei Team, 400, 401 advertising, 411
Advertising Brand management
banner ads, 186 Apostles, 63, 83, 418
Affdex, 221, 222, 223, 224, 226 BT Group, 195
Affectiva, 220, 222, 223, 224, 225, 226 Caesars, 37, 43, 270, 271, 272
Affective Computing, 220 Cambridge Analytica, 50, 214, 215, 241
AI, 10, 48 Capacity Management, 35, 315
AI dilemma, 305 Casino Engagement and Loyalty
AI-generated content, 75 Platform, 87
AI-powered search, 96 Chatbots, 77, 78, 285, 379, 380, 381,
Amazon, 49, 96, 228, 391 382, 383, 384, 392
Amazon Comprehend, 380 Chauhan, Alok S., 140
Amazon Transcribe, 380 Cialdini, Robert, 207, 208, 209
Analytics Cicso, 292
descriptive analytics, 18, 19 Clickstream, 23, 103, 112, 113, 127,
Descriptive analytics, 126 166, 167, 172, 189, 192
diagnostic analytics, 18, 19, 127 Clickstream analysis, 25, 55, 166
edge analytics, 19, 360, 374 Click-thrus, 186
predictive analytics, 14, 18, 19, 20, Cloud computing, 187
37, 78, 80, 85, 105, 112, 125, Cluetrain Manifesto, The, 292, 349
127, 128, 145 Collaborative projects, 34
Predictive analytics, 126, 127, 128, Common Short Codes, 192
283 Content communities, 34
prescriptive analytics, 18, 20, 129,
CRM, 18, 22, 23, 25, 29, 30, 42, 45, 55,
360
58, 59, 60, 61, 63, 64, 70, 74, 79, 84,
Prescriptive analytics, 126, 129
89, 100, 128, 269, 352
Propensity models, 77
Text analytics, 73, 113 CrowdOptic, 354
Apple, 192 CSC, 28, 192
Artificial Intelligence, 117 Cultural exchanges, 190
Ashton, Kevin, 21, 41 Customer acquisition, 55

437
 ANDREW W. PEARSON

Customer Acquisition Model, 146 Fitbit, 226

Customer analytics, 27, 70, 71, 72, 73, Forrester Research, 27, 28
103, 104, 106, 107, 108, 109, 110, Foursquare, 87, 242, 243, 246, 281, 352
111, 112, 113, 188, 302, 403 Gartner, 41, 355, 358, 373, 374
Customer behavior, 22, 23, 27, 109, Geofencing, 242, 243, 265
188, 403 Giaglis, George, 243
Customer Centric Relationship Global Trust in Advertising, 84
Management, 63 Google, 49, 96, 177, 188, 189, 192,
Customer churn, 24, 26, 55, 62, 63, 71, 227, 228, 229, 245, 387, 419
103, 110, 112, 163, 171, 281 Google Answer Box, 397, 398
Customer Churn Model, 163 Google Assistant, 49
Customer Conversion Model, 149 Google Cloud Natural Language, 285
Customer dissatisfaction, 83 Google Home, 12
Customer journey, 74 Google search, 228
Customer loyalty, 22, 24, 61, 65, 70, Greenberg, Paul, 59
82, 83, 85, 109, 112, 172 Gulbransen, Scott, 292
Customer satisfaction, 83 Hadoop, 24, 42, 73, 86, 112, 113, 119,
Customer segmentation, 103, 145 293, 306, 353
Customer’s Psychological Profile, 212 Hexum, 33
Das Magazin, 212, 215 High rollers, 80
Data catalog, 310, 312, 314 Hitachi, 81, 386
Data cataloging, 309, 310, 311 IBM, 1, 40, 69, 86, 104, 108, 109, 112,
Data governance, 306, 308, 309, 310, 129, 172, 225, 265, 284, 293, 294,
311, 313, 314 295, 304, 306, 307, 363, 369
Data lake, 19, 23 IBM Watson, 285
Data profiling, 310 IBM Watson Analytics, 385
Databricks, 305 Image search, 389, 391
Decision tree InfiniteInsights, 86
construction of, 131 Influence
Deighton, John A., 187, 189, 190 The Psychology of Persuasion, 207
Dell, 292 Instagram, 55, 87, 206, 240, 246, 389,
Deloitte, 355, 356, 373 411
Digital Interactive Marketing: The Five Internet, 189
Paradigms, 189 Internet of Things, 21, 22, 358
Disney, 219 Intromercials, 187
Dushinski, Kim, 191 IoT, 22, 26, 48, 125, 225, 306, 311, 358,
eBay, 93 359, 360, 361, 362, 363, 374
Eckerson, Wayne, 81, 128 Jiepang, 33, 243
Emarsys, 29, 42 Jones, Thomas, 62, 83
Facebook, 1, 18, 23, 24, 30, 32, 33, 50, Kaliouby, Rana el, 220, 221, 222, 223,
55, 60, 62, 73, 78, 84, 86, 96, 112, 224, 225, 227
113, 177, 178, 188, 191, 211, 212, k-nearest neighbors, 135
213, 214, 215, 216, 218, 220, 226, Knowing What to Sell, When, and to
228, 240, 241, 242, 245, 269, 270, Whom, 79
271, 283, 292, 293, 301, 381, 382, Kornfeld, Leora, 187, 189, 190
383, 384, 387, 415, 418 Kosinski, Michal, 212, 213, 214, 215
Facial Action Coding System, 222 Ku6, 33
Facial recognition, 217, 218, 219, 263, Kumar, R., 79
386, 387 Lift and gains charts, 81, 128

438
 THE A.I. CRUISE LINE

Lighthouse Signal Systems, 244 messages, 192

Link building process, 385 marketing campaigns, 192
LinkedIn, 30, 60, 73, 112, 113 mobile advertising, 192
Location analytics, 169, 170 mobile publicity, 192
Location-based services, 411, 425, 426 mobile search, 192
advertising, 411, 425 mobile web, 192
Lovelock, Christopher, 58, 59 opting-in, 191
Loyalty, 29, 45, 61, 62, 63, 64, 73, 79, persona planning, 417
82, 83, 86, 89, 106, 107, 109, 112, privacy, 411
113, 125, 167, 171, 217, 242, 244, proximity marketing, 192
270, 293, 302, 417, 418, 422 push media model, 191
Machine learning, 1, 15, 48, 53, 58, 76, social networking, 192
text messaging, 192
77, 78, 79, 94, 96, 97, 117, 141, 173,
types of advertising campaigns,
215, 220, 222, 229, 231, 235, 237,
191
238, 239, 240, 272, 309, 313, 314,
voice, 191
380, 381, 382, 385, 392, 403 Mobile Marketing, 186, 191, 228, 260
Macy’s, 391 Digital Interactive Marketing: The
MapReduce, 86 Five Paradigms, 189
Market segmentation, 64, 65, 67, 144 Mobile Marketing Handbook, The, 191
Marketing campaigns, 27, 81, 104, 128, Mobile positioning technology
188, 403 kinds of, 243
automating campaigns, 80, 413 Mobile search, 228
Marketing promotions, 22, 23 Mobile value chain, 186
Marr, Bernard, 20, 421 MyPersonality, 213
Marriott MySpace, 112, 188
Marriott Rewards, 89 Napster, 190
Micro-blog, 34 Natural Language Processing, 12, 92,
Microsoft, 15, 49, 69 95
Microsoft Text Analytics, 285 Neural networks
Millward Brown, 222, 223 types of training, 141
MMS, 28, 245 Neural Networks in Data Mining, 140
Mobile Nike FuelBand, 226
digital advertising, 186 Ocean Method, 212, 213
Mobile advertising Omni-commerce, 28
broad-based brand advertising Optimizing Offers, 163
campaign, 187 Oracle, 15, 69
campaigns, 187
OTT, 26, 28, 192, 245, 415, 418
Interactive, direct response
Pandora, 76
campaign, 187
Patron Worth Model, 150
targeted search advertising, 187
Mobile Advertising, the book, 186, 187, Personalization, 1, 45, 47, 50, 51, 52,
188, 353 53, 54, 57, 77, 79, 177, 193, 198,
Mobile analytics, 104 207, 215, 231, 232, 233, 238, 242,
Mobile commerce 257, 380, 393, 435
personalization, 27, 45 Pinterest, 220, 270, 289, 415
Mobile coupons, 418 Pokémon Go, 357
Mobile marketing Post-roll video, 187
advantages, 191 Predictive analytics
location-specific marketing k-Means cluster, 132

439
 ANDREW W. PEARSON

k-nearest neighbors, 134 Qieke, 33

Predictive analytics QQ, 33
Decision Trees, 130 QR codes, 26, 28, 187
Predictive analytics RACE framework, 77
Logistic Regression, 136 RACI Model, 312, 313
Predictive analytics Real-time marketing, 25, 349
A/B Testing, 138 Reichheld, Frederick, 82
Predictive analytics Reinartz, V.K., 79
Time Series Model, 139 Reinforcement learning, 118
Predictive analytics RetailNext, 170
Neural networks, 140 RFM Models, 147
Predictive analytics Roku, 178, 257
Discriminant Analysis, 143 Salesforce.com, 69
Predictive analytics Samsung, 226, 391
Survival or Duration Analysis, 144 SAP, 15, 69, 86, 383
Predictive asset maintenance, 118, 361 SAS, 2, 40, 86, 115, 116, 119, 129, 172,
Predictive modeling, 80, 82, 129 176, 352
predictive models, 81, 128 Sasser, W. Earl, 62, 82, 83, 418
segmentation methods, 81, 128 SCRM, 45, 59, 79, 84, 89
Predictive models Search engine, 14, 214, 227, 228, 229,
six steps of creating them, 81, 129 404
Procter & Gamble, 186
Sentiment analysis, 280, 281, 304
Programmatic advertising, 9, 76, 177,
SEO, 232, 385
178, 180, 400
Sharma, Chetan, 186
Programmatic Media buying, 76
Short Message Service, 192
Propensity to Respond Model, 148
Singh, Dr. Yashpal, 140
Property exchanges, 190
Siri, 192
Proximity marketing, 192
SMS, 28, 30, 192, 242, 245, 418
Psychographics, 50
Snapchat, 218, 220, 387
Psychology of personalization
Social exchanges, 190
acquiescence effect, 199
Social media, 192
availability heuristic, 199, 204
The six types of, 33
buffer effect, 199, 204
Social media analytics, 272, 273, 293
conformity and social influence,
Social media monitoring, 60
199, 202
consistency principle, 199, 200 Social media monitoring Tools, 286
decoy effect, 199, 203 Social media uses
endowment effect, 199, 200 Social Media Monitoring, 291
foot-in-the-door technique, 199 Social Media uses
framing effect, 199, 201 Discover Important Brand Trends,
informational social influence, 199 186
loss aversion, 199, 202 Social network analysis, 272
mere exposure effect, 199, 203 Social networking, 192
propinquity effect, 199, 204 Social networking sites, 34, 192
reciprocity, 199, 208 Social proof, 207, 413
scarcity principle, 200 wisdom of the Crowds, 210
Psychometrics, 212 Spark, 119
Push technology, 30 SPSS, 86
Python, 19, 86, 119 Starbucks, 33, 242, 273
Qantas, 387, 388 Stodder, David, 104, 108, 109, 283

440
 THE A.I. CRUISE LINE

Stream Processing, 24, 42, 353 Virtual Reality, 354

Streaming Analytics, 349, 353 Virtual social worlds, 34
SugarCRM, 69 Vlogs, 269
Supervised learning, 118 Voice Activated Internet, 396
Sutton, Scott, 125, 138, 148, 149, 150, Voice recognition, 239, 394, 395
151, 163, 164, 172 Voice Search, 228, 397
Tableau, 129 Wähner, Kai, 24, 25, 42, 353, 354
Talend, 308, 309, 310, 311, 312, 313, Wearables, 362, 363
314, 315 Webopedia, 355
Taobao, 92, 95, 96 Website Morphing, 192, 193
Rebate-for-Feedback, 95 WeChat, 18, 23, 24, 30, 32, 33, 47, 55,
TDWI Research, 71, 280 62, 73, 87, 112, 167, 218, 220, 240,
Text analytics, 281 242, 245, 246, 353, 387, 415, 418,
Thought tracing, 189 425
Toyota, 281, 304 Weibo, 18, 23, 24, 32, 33, 60, 62, 73,
Tracinski, Rob, 38, 43 112, 273, 293, 418, 419
Trust index, 310 WhatsApp, 30, 55
Twitter, 30, 32, 60, 62, 73, 84, 98, 112, Wheel of Loyalty, 61, 404
113, 211, 244, 269, 283, 292, 415, Wikipedia, 188, 349
418, 419, 436 Wirtz, Jochen, 58, 59, 61, 404
Twitter Revolution, 31 Woods, Dan, 13, 177, 349, 404
Uber, 257 Xbox One, 226
Unsupervised learning, 118 Yahoo!, 186
Ushi, 33 Youku, 23, 30, 33
VanBoskirk, Shar, 417 YouTube, 23, 24, 30, 188, 191, 292,
Venkatesan, Rajkumar, 79 293
Verizon, 225 YY.com, 32
Video advertising, 392 Zone of affection, 62, 83, 418
Video pre-rolls, 187 Zone of defection, 62
Virtual game worlds, 34 Zone of indifference, 83

441