D E V E LO P I N G

DISRUPTIVE
BUSINESS
ST R AT EG I E S
WITH
S I M U L AT I O N
W H I T E PA P E R
CONTENTS

Introduction ...............................................................................................................................................01

01 Traditional Methods for Analyzing Business Strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .0 2

02 Two Judgements About Uber’s Potential Market Size ......................................................04

03 How Simulation Can Help Create Business Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .0 8

04 Case Study: Major US Airline Decides Not to Charge Additional Fees ..........................15

Conclusion .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 0

Additional resources ...............................................................................................................................21

I N T RO D U C T I O N
Innovation in products, markets, and business models fuels
business growth, and disruptive strategies radically change the
business ecosystem. When managing innovation, firms must decide
whether to invest in new products for markets that do not yet
exist, with business models that have not yet been tried. Leaders
consistently miss big opportunities when faced with disruption
because traditional analysis methods are inadequate when faced
with innovation. Working with disruptive strategies is different than
predicting the future based on our view of the past. It is important
not to just predict the future, but to think about how to create the
future with our decisions.

In this paper, we will cover different approaches for modeling and

analyzing business strategies, determine how to develop an optimal
business strategy, and explain why simulation modeling is the best
approach to address these kinds of challenges.

The white paper is based on the presentation of Lyle Wallis, director

at PwC, at the AnyLogic Conference 2016.

01
TRADITIONAL
METHODS
FO R A N A LY Z I N G
BUSINESS
ST R AT EG I E S
 TRADITIONAL METHODS FOR ANALYZING BUSINESS STRATEGIES

TRADITIONAL METHODS
FO R A N A LY Z I N G B U S I N E S S ST R AT E G I E S
Really, how can we predict the future? To Q U A L I TAT I V E S C E N A R I O
DATA - D R I V E N T EC H N I Q U E S
give a substantive answer, we must create A N A LY S I S
a representative model. This is inherently a
These approaches are very subjective and general. These are, for instance, correlational technique,
creative process and we have to consider the
They usually provide a narrative and contextual statistical, machine learning, and all techniques
tools that can support this. Most importantly,
description of how the present will evolve into the that look for patterns in the historical data.
which tool is right for a particular project?
future. Qualitative scenario analysis can be seen as a They try and project those patterns forward
process of understanding, analyzing, and describing to say something useful about the near future.
Anyone can perform a simple analysis. However,
the behavior of complex systems consistently and Consequently, when planning fundamental
the more complex the analysis becomes, the
completely using non-numerical data. Qualitative change and market disruption, information about
more powerful tools are needed. If we
scenario analysis does not give organizations any the past can have limited relevance. Overall, it
look at the tool sets that have
insight about investments they should make, to is a highly useful technique because it helps us
historically been available, we find three:
what degree they need to make investments, or over understand where we are and what might happen
what timeframe. It may be contextually useful, but it if everything remains stable. Its main problem is,
does not take us far enough. when seeking to disrupt a market, its predictions
about the future are only of limited value.

SPREADSHEET MODELING

This is the most popular approach. Although commonly used, spreadsheet models are inadequate
for describing innovative business systems. All business systems have feedback, non-linearity, and
delay, and spreadsheet-based models cannot describe these features adequately. The quality of the
model that is used for decision making is crucial for the quality of thinking around important strategic
questions, and consequently, for the success of the business. Spreadsheet models cannot provide us
with high quality strategic insight. Let us illustrate this statement with an example.

03
T WO
JUDGEMENTS
A BO U T U B E R’ S
P OT E N T I A L
MARKET SIZE
 TWO JUDGEMENTS ABOUT UBER’S POTENTIAL MARKET SIZE

T WO J U D G E M E N T S A B O U T
U B E R’ S P OT E N T I A L M A R K E T S I Z E
During 2014, a professor of finance at the New York University’s Stern The model of the Uber’s opportunity looked like this:
School of Business, Aswath Damodaran, published a post in his blog about
the value of Uber1 . Using market data and financial analysis, Professor Market opportunity = TAM * % Share
Damodaran came to the conclusion that, with his best estimation, the value
of Uber was $5.895 billion. This number is far short of the $17 billion that was The opposite opinion on the same topic was given by professional investor
invested into Uber in 2014. Damodaran summed up that Uber is being highly Bill Gurley2. In his blog, he declared there was a critical error in both core
overpriced. It is worth mentioning here, that at the time of writing this paper assumptions. In choosing to use the historical size of the taxi and limousine
(September 2018), Uber’s latest valuation is $72 billion . 2
market, Damodaran assumed that the future would look very much like
the past. This meant that the arrival of a new service, like Uber, would
Damodaran’s estimate of value was tied to two assumptions: the total make no difference in the size of the car-for-hire transportation market – a
available market (TAM) and Uber’s market share within that TAM. As TAM was dangerous presumption. If you significantly improve the quality of service,
decided to use the historical size of the taxi and limousine market, which was create new features, and, in addition, extend the sphere of application, you
estimated at $100 billion. Financial analysis results depended entirely on can materially expand the market. With these things in mind, what were
these two assumptions. Uber’s strong points?

1
Aswath Damodaran, “A Disruptive Cab Ride to Riches: The Uber Payoff”, Blogspot, June 9, 2014, http://aswathdamodaran.blogspot.com/2014/06/a-disruptive-cab-ride-to-riches-uber.html
2
Bill Gurley, “How to Miss By a Mile: An Alternative Look at Uber’s Potential Market Size”, Above the Crowd, July 11, 2014,
http://abovethecrowd.com/2014/07/11/how-to-miss-by-a-mile-an-alternative-look-at-ubers-potential-market-size/

05
 TWO JUDGEMENTS ABOUT UBER’S POTENTIAL MARKET SIZE

Waiting time
Uber had reduced average waiting time to less than five minutes,
significantly helping improve the quality of service.

Coverage
Uber became popular even in geographic regions where consumers
rarely order taxis. So, the market size expanded.

Payment method
With Uber you don’t need to pay with cash. This makes it much easier
to use than an ordinary taxi.

Civility
Uber’s customers and taxi-drivers can rate each other. This leads to
politer interactions. If your ride was pleasant, you are more likely use
their service again.

Safety
With Uber, there is a record of every ride, every customer, and every
taxi-driver. This makes it safer and more trustworthy for customers,
especially women.

Lower price
Uber was becoming more and more popular, so it was able to lower
prices without any damage to taxi-drivers' income. Fall in prices led
to an increase in demand.

06
 TWO JUDGEMENTS ABOUT UBER’S POTENTIAL MARKET SIZE

Bill Gurley’s model demonstrates the dynamics of the taxi market when
introducing change, far more complex than the fixed market of Damodaran.
To analyze the system, it is necessary to understand its structure, inside
interdependences, and feedbacks. So, he captured this idea in a stock and
flow diagram (see right) that is commonly used in system dynamics.

The small positive feedback loop generates a larger market and a

competitive cost advantage. The large positive feedback loop is an engine
of rapid growth. Still, this is a simplified model and it leaves out important
growth limiting negative feedbacks, like competition for customers and
drivers. Beyond that, deeper analysis, and perhaps market research, is
needed to quantify these effects and the company's potential growth. As
a result of his own analysis, Bill Gurley came up with a market opportunity
that was much larger than the spreadsheet version (~$250B).

It’s difficult to say which is better because both provide worthwhile

viewpoints, and to any stakeholder it is better to analyze the situation
from different perspectives.

Nevertheless, this example confirms that formulas are good for expressing
static dependencies between variables, but they fail when describing
systems with dynamic behavior. The positive loops seen in Bill Gurley’s
model generate a change in the addressable market – change we could
never represent in a spreadsheet. Bill Gurley’s model precisely highlights
the weakness of the spreadsheet approach.

In recognition of the dynamic nature of systems that Damodaran and

Gurley tried to represent, it is necessary to explore a modeling technology
designed specifically for analyzing dynamic systems – simulation modeling.

07
H OW
S I M U L AT I O N
CAN HELP
C R E AT E
BUSINESS
ST R AT EG Y
 HOW SIMULATION CAN HELP CREATE BUSINESS STRATEGY

H OW S I M U L AT I O N C A N H E L P C R E AT E
B U S I N E S S ST R AT E G Y
Simulation modeling means creating a I N C R E A S E D ACC U R AC Y EASE OF USE CO O P E R AT I O N
digital prototype of a real-world system.
Simulation enables experimentation
on a valid digital representation of a Simulation models enable you to Once you have selected the The structure of a simulation model
system to analyze, optimize, and predict analyze systems and find solutions appropriate level of abstraction, naturally reflects the structure
processes inside. where traditional methods fail. the development of a simulation of the real system. As simulation
This is largely because a simulation model is a more straightforward models are developed using mostly
The main advantages of using
model can take into account more process than analytical modeling. visual languages, it is easy to
simulation modeling as a problem-
solving tool, compared to spreadsheet- complex interdependencies and the It typically requires less intellectual communicate the model internals
based modeling are: behavior of the system over time. effort, is scalable, and is modular. to other people.

MEASURABILITY V I S U A L I Z AT I O N PERSUASIVENESS H A N D L E U N C E RTA I N T Y

In a simulation model, you can The ability to play and animate Simulation models are a lot more Simulation considers the
measure any value and track any system behavior over time is one convincing than spreadsheets randomness and interdependencies
entity that is not below the level of the greatest advantages of (not to mention slides or reports which characterize real-life
of abstraction. Measurements and simulation. Animation is used not with numbers). If you bring and business.
statistical analysis can be added at only for demo purposes, but also run a simulation to support
any time. for verification and debugging. your proposal, you will have an
advantage over those who bring
just numbers.

09
 HOW SIMULATION CAN HELP CREATE BUSINESS STRATEGY

Consequently, we can say that the power to “look into the future” and understand the impact on
multiple key metrics, instead of making assumptions and using averages, is what simulation provides.

Analyzing innovation is a tricky and complicated process, and building the simulation model is only
the first step. In general, we can divide this process into four basic steps:

1 . M O D E L D E V E LO P M E N T 2 . P E R FO R M A N C E M E T R I C S
The most important step is to identify and formulate the problem correctly. The Data on system specifications, input variables, and the performance of the existing
phase of building the model is a necessarily creative process. Mapping the real world system is required. Statistics should measure an organization's behavior, activities,
to the world of models, choosing the abstraction level and the modeling language, is and performance. Many metrics are finance based, but they may also focus on other
the least formalized element in the whole process of using models to solve problems. things. Performance metrics are traditionally used by stakeholders to evaluate
A model should capture the dynamics of a business system (for example, customer strategic decisions. So, you must decide what performance measures will be
adoption, competitor response, asset utilization). Only components important for analyzed in your simulation model.
solving the problem should be represented in the model.

3 . V I S U A L I Z AT I O N S 4 . S C E N A R I O A N A LY S I S
Even if you are not going to make an exact digital twin of your organization in A model can be used to test thousands of strategy variations and scenarios that
3D, visualization should be done for verification of your model. Verification is the capture uncertainty. For this purpose, you can use Monte Carlo, Sensitivity Analysis,
process of ensuring that the model behaves as intended. To verify your model, and Parameter Variation experiments. They help you discover how randomness and
you can input real data to the model and compare the results with historical data. parameter change can affect your model’s behavior.
Good data representation will also help you choose between strategic alternatives.
Moreover, simulation software helps users visualize vehicles, staff, equipment,
buildings, and other items and processes in their models.

10
 HOW SIMULATION CAN HELP CREATE BUSINESS STRATEGY

When developing a strategy, each alternative must be evaluated and PwC built an AnyLogic simulation model and hosted it on
compared against each other using a whole range of conditions. Many
their own cloud infrastructure. AnyLogic Cloud simplifies
factors should be considered, including:
this process and provides a platform for analysts to
• Policy alternatives. Examples of different strategies include increasing
marketing spend, competitive pricing, and maximizing service levels. integrate models into their data flows.
• Environmental assumptions. Outside control factors that can impact
strategy effectiveness (for example, customer acceptance, competitor
response, economic trends).
• Randomness. Stochastic factors that can impact effectiveness include
customer choice, word of mouth messaging, and marketing reach.

In some of their projects PwC conducted strategy analysis that utilized

~500K simulation runs of a single model with different scenarios and
varying parameters.

Many simulation runs can pose a problem because it is hard to evaluate

thousands of them without access to a lot of computing power.

To make this practical, PwC built a cloud-based analysis workflow for

their projects that allows them to iteratively generate new strategies and
explore results. This provides a valuable example of how business analysts
can organize data flows when using simulation for strategy analysis. See
figure on the next page.

11
 HOW SIMULATION CAN HELP CREATE BUSINESS STRATEGY

12
 HOW SIMULATION CAN HELP CREATE BUSINESS STRATEGY

Because of deep uncertainties in the environment and random variation, each

strategy alternative can produce a range of outcomes. One way to summarize these
for easy comparison is a Box and Whisker plot.

13
 HOW SIMULATION CAN HELP CREATE BUSINESS STRATEGY

The distribution of outcomes for each strategy can be compared to Robust strategies are those that perform consistently under uncertainty.
determine the performance, robustness, and risk associated with each There may be a cost associated with predictable strategies. An example
decision. Some strategies will have the potential for a greater upside and of a more robust strategy could involve policies that seek to increase
less risk compared to the base case. Better strategies will also have higher asset utilization through active management. Good strategies have
average expected returns within its range of possible outcomes. disproportionate upside potential. Large upsides are driven by strong
positive feedback loops and nonlinearity.

Real data for Box & Whisker plots can

only be obtained using simulation
modeling, since it is the only analytics
method that allows the consideration
of randomness inside a model.

Let us illustrate these ideas with an

example from the PwC practice. This
case study is based on the presentation
of Mark Paich from PwC at the
AnyLogic Conference 2013.

14
C A S E ST U DY:
MAJOR US
A I R L I N E D EC I D E S
N OT TO C H A RG E
ADDITIONAL
FEES
 CASE STUDY: MAJOR US AIRLINE DECIDES NOT TO CHARGE ADDITIONAL FEES

CASE STUDY: MAJOR US AIRLINE DECIDES

NOT TO CHARGE ADDITIONAL FEES
The U.S. commercial airline industry is one of the most diverse, dynamic,
and perplexing in the world. It is fast-evolving, labor intensive, capital
intensive, hyper-competitive, and very vulnerable to the ebb and flow of
business cycles, as well as being among the most regulated of businesses.

Airline management is required to make long-term decisions regarding

fleet sizes, market fluctuations, and fuel prices, while seeking ways to
increase profit in a competitive environment.

P RO B L E M
A major U.S. airline was facing a situation where opportunities to extend
the existing strategy were limited, coupled with an increasing cost
structure due to competition, commodity prices, and acquisition integration
activities. The airline began to explore several options to generate new
profits through ancillary products or changes to existing policies, and
was under intense pressure from board members, Wall Street, and various
analysts to do so.

Although the revenue generation through charging additional fees was

apparent in the short term, prior to implementing a policy change, the
airline opted to evaluate the long-term perceived impact on brand equity,
market share, and customer loyalty.

16
 CASE STUDY: MAJOR US AIRLINE DECIDES NOT TO CHARGE ADDITIONAL FEES

SOLUTION
PwC was employed by the airline to model the predicted impact of the
client’s ticket market share and company brand sentiment after introducing
new products or policy changes.

PwC found traditional marketing mix models to be limited and unable

to analyze the airline’s challenges. First, because they are aggregate, all
customers are represented in a single regression equation which disregards
the fact that not all consumers behave the same. Secondly, these types of
models do not show interaction between consumers, when on the contrary,
customers share stories, attitudes, and memories, which is known as
emergent behavior. Emergence is used to describe the behavior a group
exhibits because individuals make different choices than they would if
they were not part of a group. A third limitation, when using typical market
models, is the lack of explicit representation of consumer decision making.
Analysts are then unable to see consumers gathering information, making
informed decisions, and forming consideration sets as they do in the real
world. Lastly, in traditional regression models, nonlinear relationships
are not accounted for, data is limited to time series data, and there is a
relatively short time horizon. In the end, this type of model is inappropriate
for most consumer behavior analytics.

17
 CASE STUDY: MAJOR US AIRLINE DECIDES NOT TO CHARGE ADDITIONAL FEES

These restrictions, and an increased likelihood of inaccurate results,

prompted PwC to explore other modeling options. They chose AnyLogic
simulation software due to its flexibility, scalability, and capability to
handle sophisticated, computationally intensive techniques that model the
behavior of agents (consumers) in the market.

Utilizing AnyLogic software, PwC built the Experience Navigator, an

agent-based consumer behavior model of multiple airline markets which
included client competition, the process of consumers making choices, and
a relatively complete representation of the ecosystem in each market. The
project used historical industry data, behavioral economics principles, and
measurable experiences to create a behavioral model to help understand
the impact on customer purchase behavior and the airline’s social contract.
They built behavior of each consumer at the market with agent statechart
(see right).

The information used during model building and calibration included:

• Time series of airline market shares (volume, prices)

• Cross section of individual travel behavior
• Market research from PwC Experience Radar (customer experience survey
completed by PwC)
• Market research from client discrete choice models
• Process and theory from consumer choice literature
• Qualitative knowledge of the airline industry

PwC and the airline were now able to understand how the interaction of
different factors (fare utility, past experience, loyalty, and word of mouth)
may produce behavior, influence market share, and modify markets overall.

18
 CASE STUDY: MAJOR US AIRLINE DECIDES NOT TO CHARGE ADDITIONAL FEES

O U TCO M E :
O U TCOME
PwC’s Experience Navigator is based on a model built with AnyLogic
PwC’s Experience Navigator is based on a model built AnyLogic simulation
simulation software and was used to:
software and is used to:
• Analyze the sales funnel at a particular segment and individual
• Analyze the Sales funnel at a particular segment and individual competitor level
competitor level.
• Visualize consumers changing beliefs over time
• Visualize consumers changing beliefs over time.
• Dive into different agents to understand their positive and negative
• Dive into different agents to understand their positive and negative
experiences
experiences.
• Forecast revenue impact of change in consumer experience
• Forecast the revenue impact of change in consumer experience.
• Set price and marketing influence levels of the Airline
• Set price and marketing influence levels of the airline.
Ultimately, the model results showed that losses in market share and revenue
over the long term would significantly offset any gains from charging additional Ultimately, the model results showed that losses in market share and
fees. In addition, the model proved that if the company would set the fees the revenue over the long term would significantly offset any gains from
same as the competition, their loss of market share would be considerably charging additional fees. In addition, the model proved that if the company
greater than the competition, because the choice behavior for this Airline was set the fees the same as the competition, their loss of market share would be
due to positive brand equity and positive perception. considerably greater than the competition because the choice behavior for
The model provided substantial evidence to convince stakeholders and Wall this airline was due to its positive brand equity and positive perception.
Street that the Airline should not implement the charging of additional fees but
should cultivate an alternate strategy to increase revenue. The model provided substantial evidence to convince stakeholders and Wall
Street that the airline should not implement the charging of additional fees
but should cultivate an alternate strategy to increase revenue.

19
CONCLUSION

Innovation disrupts an existing business ecosystem, creating a new Simulation software can be used in these situations to cover a range
economic reality that must be worked with. And, as we have proven, of issues by unveiling hidden interactions, testing dependencies, and
traditional methods, including popular spreadsheet-based approaches, are revealing sensitivities. Simulation considers the randomness which
inadequate to fully understand this process. characterizes the behavior of the real-life business environment and
provides a safe way to test and explore different “what-if” scenarios.
In most cases we cannot afford to search for the right solutions by Visualization builds trust in a model, making it a lot more convincing than
experimenting with real businesses. As such, modeling is about finding spreadsheets or slides. Professional users can set up complex data flows
a way, from the problem to its solution, using a risk-free world where we that utilize cloud computing to carry out in-depth strategy analysis, with
are allowed to make mistakes, undo things, go back in time, and start all thousands of simulation runs of a single model, with different scenarios
over again. This is exactly what people need when they are creating a and varying parameters.
business strategy.
Modeling results enable informed thinking and provide effective
management, with mitigation strategies for risks and negative scenarios.
Simulation is the perfect analysis tool for creating and evaluating
innovative and disruptive business strategies.

20
ADDITIONAL White Papers

Case studies
R E S O U RC E S
Videos

Books

Download AnyLogic Simulation Software

Seminars and Trainings

21
C O N TAC T S
T HE ANY LOG I C COM PAN Y

info@anylogic.com
Contact your local office >>