Agent-Based Modeling: A New

Approach to Simulating the

Marketplace
August 30, 2012 / Lieberman Research Worldwide / 2 comments

Agent-based modeling is a relatively new simulation modeling

technique that has the advantage of being able to look at how
potential consumers interact with each other. It is more forward-
looking, can integrate with other modeling techniques, and can
address more complex business problems.

Editor’s Note: Mick McWilliams, LRW’s SVP of Research Methods, will be doing a series of posts over the
next several weeks on the topic of agent-based modeling. This is the first post in the series.

At LRW, we are continually asking ourselves “so what?” Usually, we’re actually asking, “So, based on the
results of the research, what should our client do?” How often have you asked yourself these questions when
trying to increase trial, market share, or revenue of a brand, product, or portfolio of products?
 How many products should you have in your portfolio?
 How should the products in your portfolio differ in their features and how should they be priced?
 Which of the brand’s strengths should be emphasized?

Every marketer asks themselves these questions and of course there are many methods that can be employed to
help answer these questions. One of the most powerful is simulation modeling. A well designed simulation
modeling tool helps compare potential outcomes from various product design and marketing options.

For example, a product designer can input different possible product design options into a modeling tool and
get an idea of how that specific product with those specific features will compete in the current market place.
Experimenting with different product design inputs enables the product planner to identify the most
competitively promising product designs.

A new tool in a marketer’s toolbox is called “Agent-based simulation modeling” – ABM for short – and it
offers the potential to significantly extend the capabilities of simulation modeling. ABM has multiple
advantages as compared to more typical simulation modeling approaches.

How is agent-based modeling better than current simulation modeling techniques?

ABM can deliver more realistically forward-looking decision support: Typical simulation modeling
approaches provide what is basically a single “best guess” of modeled outcomes. Agent-based modeling
provides more realistically forward looking planning support tools because it can predict the relative
probabilities of multiple potential planning scenario outcomes.
ABM can model the “complex interactivity” in the marketplace: For example, ABM can model the effects of
real-world consumer interactions, such as word-of-mouth and social media. And it can concurrently model the
effects of consumers’ interactions with brands (e.g. through exposure to different brands’ advertising).
Through this ability to model the “complex interactivity” that characterizes the real-world marketplace, ABM
is capable of indentifying potential “emergent” outcomes that cannot be foreseen using current popular
simulation modeling approaches, such as conjoint or choice modeling.
ABM can integrate disparate simulation models into a synergistic “uber-model”: The ABM approach is
flexible enough to lend itself to integrating other simulation modeling perspectives, such as conjoint modeling
for product optimization or messaging, that have typically only been studied in isolation. For example, you
might previously have used two separate simulation models to support decisions regarding product design
optimization vs. product advertising. However, ABM offers the potential to incorporate both product planning
and brand communications planning into an integrated decision support tool. Such an integrated simulator
would take both inputs and more realistically predict consumers’ probable brand choice decision outcomes.

In upcoming blog entries, we’ll go into further detail about how ABM works and how it can deliver next
generation simulation modeling improvements.
Agent-Based Modeling: Assessing
the Different Possible Outcomes
of Marketing Decisions
September 6, 2012 / Lieberman Research Worldwide / 2 comments

Understanding agent-based modeling requires thinking about

simulation modeling in a new way as it reveals the probabilities of
different possible outcomes in the marketplace and not just a single
“best guess” result.

Editor’s Note: Mick McWilliams, LRW’s SVP of Research Methods, will be doing a series of posts over the
next several weeks on the topic of agent-based modeling. You can read the first post in the series which is
anintroduction to agent-based modeling.
Though we often don’t explicitly recognize it, the fact is that product designers and marketers try to predict the
future on an ongoing basis. A product planner decides what features a product will have and
therebypredicts that the product will be more attractive to consumers than competitive products in the future. A
brand manager decides what aspects of his brand’s promise and image to emphasize and so predicts that the
resulting “brand story” will win against competitors’ brand stories in the future. Designing and marketing
successful products is all about deciding which of various competing decision alternatives will most likely
influence the future in your favor.
The venerable Jedi master, Yoda, once wisely noted, “Always in flux, the future…” And he was right. Anyone
who tells you they can confidently and accurately predict the exact future outcome of a given product design or
marketing decision is lying; either to themselves, to you, or both. The key challenge is that, in a very real
sense, there is no single, certain future out there. The future is more realistically thought of as a range of
different outcomes. Marketing planners would do well to recognize that any specific product design or
marketing decision – or set of decisions – actually has many possible outcomes, some of which are more likely
to happen than others. The product planning and marketing challenge would be best approached through
comparing different sets of decision alternatives by comparing the many possible outcomes.
Agent-based modeling is an improvement upon other simulation modeling techniques in that does not
implicitly ignore the reality of this situation. A well designed agent-based modeling simulator does not report a
single “best guess” as to the future outcome of a given set of product design and marketing decisions.
Rather, it reports the probability of each outcome in a set of possible outcomes.
An excellent example of determining the probability of different possible outcomes is forecasting the paths of
hurricanes. The six curving dotted lines extending from the hurricane’s present location combine to create a
curving “cone of possible outcomes” containing several paths which the hurricane might take, depending on
how multiple semi-predictable influences actually evolve over the days to come.

The forecast illustrated above is powerfully useful, in large part, because it does not create a false illusion of
certainty. The forecast is actually a combination of six models which, together, project that the greatest
likelihood is that the storm will mainly impact the coast of North Carolina. However, the forecast nonetheless
warns disaster planners in Florida, Georgia and South Carolina to be ready for the possibility that any one of
their states could suffer a direct hit.

Most methods of simulation modeling currently used in marketing research are like a single-model hurricane
forecast that produces just one curving dotted line. And, importantly, success in the marketplace will suffer to
the extent that that single-outcome forecast turns out to be wrong. Agent-based modeling combines multiple
simulation models to produce something like a marketing decision “cone of possible outcomes.” ABM thereby
enables the marketing planner to consider the wider range of possible results from one set of decision
alternatives versus another. In our next agent-based modeling post, we’ll delve more deeply into the powerful
techniques through which agent-based modeling identifies a range of possible decision outcomes and gauges
their relative probabilities.

Agent-Based Modeling: Modeling

the Complexity of the Marketplace
September 28, 2012 / Lieberman Research Worldwide / No comments

The early twentieth century author H.L. Mencken insightfully wrote, “For every complex problem there is an
answer that is clear, simple, and wrong.” We humans routinely make mistakes due to oversimplification. Most
of us instinctively prefer simple over complex, black and white over shades of gray, right or wrong over it
depends. We deal with our highly complex, global, super industrial reality using mental equipment that
evolved in the simpler world of hunters and gatherers. Our minds are not instinctively equipped to grapple with
the very nuanced and interactive complexities of our modern reality. So we look with biased eyes to see
simplicity in a world that actually has deep complexity.

Perhaps you’re thinking, “An interesting observation, but what does it have to do with using simulation
modeling for effective product design and marketing decisions?” The answer is that those decisions will
succeed or fail in a “complex adaptive system,” in this case the modern marketplace. As noted in our
last agent-based modeling blog post, the job of marketers is to try to predict the differing probable outcomes of
alternative product design and marketing decisions. To successfully predict which of our decisions have the
greatest chance for success, we need simulation modeling tools that can realistically model the complexity of
the complex social system that is the modern marketplace.

As a method for modeling a complex adaptive system like the modern marketplace, agent-based modeling has
four key advantages over other simulation modeling techniques that are more commonly used in market
research:

 Employing “bottom up” modeling methods

 Methods that are computational as well as mathematical
 Capable of modeling complex interactivity between marketplace agents
 Able to predict emergent outcomes in the marketplace

Bottom Up Modeling

Most marketing research simulation methods attempt to market the entire marketplace system at the aggregate
level. Agent-based modeling, conversely, models the system’s key individual agents individually, letting the
aggregate outcome emerge from the decisions and interactions of the agents. For example, each consumer
agent in an ABM simulator can be modeled on an actual survey respondent, using survey data that reflects
each consumer’s market behavior decisions as they are influenced by:

 Decision rules, that can differ from one consumer agent to another
 Perceptions of different brands
 Word of mouth influences
 Observations of other consumers’ behaviors
 Advertising influences

With agent-based modeling, our model of the marketplace does not have to be distilled to an over-simplified
representation of “average” behavior.

Computational Modeling

Most simulation techniques used in market research, such as choice modeling simulation, rely almost entirely
on mathematical algorithms. Agent-based modeling will often employ mathematical algorithms as
components, but also makes heavy use of computational methods. Emphasis on mathematical modeling in the
market research industry is not due to superiority of mathematical modeling over computational modeling.
Effective computational modeling requires a great deal of computational power, and it’s only in recent years
that the level of computational power needed to do advanced agent-based modeling has become widely
available.

Because agent-based modeling is computational as well as mathematical, it can model market processes with
greater flexibility, comprehensiveness, and detail. Each consumer agent in the simulation system:

1. Perceives what’s happening in the simulated environment. For example, marketing messages that are
programmed as model inputs to reflect the marketing decisions we want to test
2. Uses their own programmed decision rules, which can differ from one agent to another, to make the
individual decisions that collectively create the marketplace outcome.

Complex Interactivity

Multiple times above we’ve referred to agent-based modeling’s ability to model consumer influences such as
word of mouth and advertising information. In the real world marketplace, such influences impact consumers’
buying decisions through their interactions with each other. A key advantage of agent-based modeling is that
it can model this process of complex interactivity, due mainly to its heavy use of computational modeling. The
ability to model complex interactivity results in an ability to predict potential emergent outcomes. This is
important because, in the real world marketplace, all outcomes are emergent outcomes.
Emergent Outcomes

The agent-based modeling approach accepts the reality that the marketplace outcomes from marketers’
decisions emerge from the complex interactivity of the marketplace. Appropriately, in an agent-based
modeling market simulator, the agents’ interactions can produce, and thereby predict, complex emergent
outcomes. Emergent outcomes occur commonly in the real world marketplace, but they cannot be predicted
through models that are based purely on mathematical equations.

Agent-based modeling can more realistically simulate and forecast consumers’ marketplace decisions and
behaviors because it uses computational methods to model consumer behavior from the “bottom up,” and can
therefore simulate the effects of complex interactivity to predict emergent marketplace outcomes. Next week
look for our final agent-based modeling blog post, with a summary of how agent-based modeling can be used
to integrate more narrowly focused simulation modeling techniques to provide more holistic market simulation
tools.

Agent-Based Modeling: Improving Decision

Support
October 11, 2012 / Lieberman Research Worldwide / One comment

Imagine you’re planning both product design and marketing communications for a new product
launch. You’ve conducted a survey that consisted of two main components:

1. A choice modeling exercise to support product design and pricing optimization

2. Brand attribute ratings, for your brand and the main competitors, to support marketing communications
So, now you have two simulation tools. One, that is based on your choice modeling data that predicts “share of
preference” for different product designs in competitive scenarios. The other, derived from your brand attribute
data, models the influences of consumers’ brand perceptions on their brand choices. Both are very useful tools.
However, even more useful would be a single, integrated simulation tool that simultaneously modeled
for both product design optimization and marketing communications. Ideally, that integrated simulation tool
should realistically model the interaction between consumers’ product design preferences and their brand
perceptions. As you’ve no doubt already guessed, agent-based modeling can provide this type of integrated
modeling tool.

ABM’s bottom up approach and computational flexibility allows for the integration of separate simulation
modeling perspectives. Continuing with the above example, a choice modeling simulator models consumers’
perceptions of product characteristics as they affect product choices. A brand communications simulator
models the effects of consumers’ brand perceptions on their brand choices. However, in the real world,
consumers’ preferences regarding product characteristics are intertwined with their brand perceptions and
inform the purchase decision that all marketers seek to influence.

Agent-based modeling can effectively link two such separate simulation models mainly because it starts with
modeling each individual agent’s behaviors and interactions and then builds up to the “emergent” aggregate
level outcome. In such an integrated ABM simulator, consumers’ decision rules regarding product
characteristics could be modeled interactively with their choice-affecting brand perceptions (also including the
effects of perceptions-affecting processes like word of mouth and brand advertising).The result is a simulation
modeling tool that can predict the probable results of product design decisions and brand communications
decisions from a more realistic and holistic integrated perspective.

To review briefly, over the course of these blog posts we’ve described four particular strengths of agent-based
modeling relative to other simulation modeling methods:

 ABM can deliver more actionably forward-looking decision support by more realistically predicting the
relative probabilities of multiple potential planning decisions, rather than producing just one best guess
 ABM can model the complex interactivity that characterizes the modern marketplace, like word-of-mouth
and social media, and consumers’ interactions with brands through marketing communications
 ABM can identify possible emergent outcomes which cannot be predicted by methods that rely solely on
mathematical techniques.
 Agent-based modeling offers a unique ability to integrate separate simulation modeling techniques to
create a multifaceted and holistically integrated decision support and planning tool.

Since the 1990s, agent-based modeling has been used increasingly in the life sciences, ecological sciences, and
social sciences. These disciplines are leveraging ABM’s multiple strengths and advantages to significantly
improve their predictive modeling capabilities. It’s time for the marketing research field to do the same.

We would love to hear your perspective on agent-based modeling. What are the business decisions you make
that could be improved by using agent-based modeling?
Big Data in Market Research:
Using Statistical Modeling to
Predict the Future
November 9, 2012 / Lieberman Research Worldwide / 6 comments

Nate Silver’s probabilities of the distribution of electoral votes for the 2012 presidential election. From

http://fivethirtyeight.blogs.nytimes.com/

In the wake of an election widely reported as “too close to call,” a surprisingly accurate predictor has emerged:
statistical modeling. Nate Silver’s impressively spot-on predictions of the 2008 and 2012 elections on
his FiveThirtyEight blog are quickly passing into legend (and translating into better book sales: The Signal and
the Noise, is the #2 best-seller on Amazon). But what Silver does isn’t magic, nor is he alone. Other
statisticians, like Sam Wang, a neuroscience professor at Princeton, and Simon Jackman, a political science
professor at Stanford, have also accurately predicted the results of the 2012 election using statistical
modeling. Math geeks everywhere are celebrating their win. Nerds rule!

LRW’s cutting-edge Marketing Sciences team is often called upon by clients to create predictive models that
can help them make better business decisions, and they’ve learned a thing or two about the practice (and the
pitfalls) of working with data to predict the future. We’ll let them break out the signal from the noise for you:

1. When developing predictive models, think probabilistically.

Naturally, marketers would be much happier if they could predict with absolute certainty what would happen
when they launched a new product, changed a price, or rolled out a new ad campaign. But no statistician
worth their salt would tell you they can predict the future. The world is changing constantly, and there are way
too many variables in the model: there’s rarely (if ever) one right “answer.”
“Don’t take one (hopefully) best guess,” cautions Mick McWilliams, PhD, Senior Vice President of Marketing
Sciences. “It’s better to model multiple possible outcomes and understand their relative probabilities of
actually happening.”

For example, forecasters during Hurricane Sandy presented their results as probable paths within a “cone” of
various possibilities (an example that should ring a bell for readers of Mick’s series on Agent-Based
Modeling). Our team recently designed a model for a client attempting to tune up their promotional programs,
and ran over 100 different simulations to identify the promotional mix that maximized profitability. Of course,
probabilistic forecasting methods are more complex and time consuming than traditional market research, but
they can significantly increase your chances of success.

2. Approach big data with a purpose.

Silver and his counterparts leveraged a massive amount of public and private opinion data to forecast the
election, tapping into a slew of polls across the US as inputs for their models. We’ve noticed lots of
conversation, excitement, and even fear about “big data” among our clients and colleagues. The excitement
stems from the potential knowledge that might be gleaned; the fear from the notion that increasingly advanced
statistical models will make human analysts less valuable.

What you should really be afraid of are the spurious correlations and patterns in big data that can lead market
researchers (and their clients) down the wrong path. That’s what Silver calls “noise,” and it can be
dangerously misleading without the right interpreter.

“Mega databases and fancy analytics software on their own do not a valid prediction make,” says Jason
Brooks, Vice President, Marketing Sciences. “They do not alleviate the need for an informed analyst guided
by relevant knowledge and armed with intelligent hypotheses to objectively test. You can lead Big Data to
water- but you can’t make it drink.”

3. Stay objective (and manage your expectations).

We know from innovations recently launched by our Pragmatic Brain Science Institute™ (aka, PBSI™) that
brains are wired to look for patterns as a means of simplifying our complex world. In an internet age
oversaturated with available data, even research professionals often focus in on the information they like and
ignore the rest. Anyone can see patterns where none exist, filter them through their biases, and cherry pick
data to fit with the story they want to tell.

Market researchers are not immune to this tendency, and must strive for objectivity in design, analysis, and
(perhaps most importantly) the presentation of their findings. Granted, academic statisticians don’t have to
worry about the consequences of telling their boss that a new product is likely to flop or that the new ad
campaign is falling flat. But the professional researcher needs to stay equally objective; what makes the most
sense from the business knowledge you have isn’t always what’s most likely to happen.

“Good forecasters must strike a balance between their prior knowledge and what the data is telling them,” says
Steve Karlin, Senior Vice President, Marketing Sciences. “Good forecasting involves generating hypotheses
about direct relationships that ought to exist, then exploring all available data for these relationships.
Importantly, you must be willing to accept that your hypotheses are incorrect.”

4. Never stop improving your models.

Top statistical modelers use a Bayesian approach to model development, which incorporates not only data
quality but hypothesis accuracy into the mix. When new data is available, they use it to update their models
and predictions on the fly, never resting with a tried and true past method.

“Be willing to walk away from norms and models that are too old,” advises Hilary DeCamp, our Senior Vice
President and Global Director of Marketing Science and Methods. “Purge old data points from benchmarks
and models, and revise estimation procedures off only the more recent data. More history isn’t better history-
unless it’s still relevant.”

Big data is only getting bigger, and statistical modeling isn’t going away. It might well be the most effective
way to use big data to predict human behavior, something any business should consider endlessly valuable.
The trick to using it is to go in with a strong hypothesis, manage your expectations, and weed out what doesn’t
work with a constantly improving methodology.
 Complexity Science and ABM Approaches to Brand Value
Posted by Greg Silverman on Tue, Apr 26, 2011 @ 08:00 AM
As Apple’s stock passed $300 as few months ago, it triggered quite a reaction in me – and not just because I sold in
the $100’s. I wondered how the value of this brand could be growing so quickly and, more importantly, how I missed
it. After some consideration about the topic I realized that a new form of intangible asset valuation is possible – one

that combines complexity science and agent based modeling.

Traditional measures of brand value have been built from three static and discrete
components:

 The financial forecast of the firm that owns the brand

 The role that brand plays in a buying decision
 An estimate of brand equity that is derived from estimates of other performance measures.

The leftover piece is often called brand strength and is typically someone’s best guess at
how important the brand is to the product we sell. Brand strength is translated into a
discount rate that gives the net present value of the brand after considering the financial
forecast and the role of the brand. These three static measures do not update each other
dynamically or continuously using consumer feedback so big changes are often overlooked
- evidenced by my missed forecast in the Apple share price.

In complexity science views of marketing, brand value is an emergent and latent property of
events, constant updating, and interaction. Because of feedbacks and interdependence,
brand equity links directly to sales endogenously not exogenously through a discount rate
calculation. In short, it’s possible to model brand performance dynamically usingAgent
Based Modeling (ABM). Because ABM is a robust aggregating tool for data and
philosophy, we don’t even need to isolate role of the brand or its strength. The model
generates brand value in terms of sales as an outcome of all brand touchpoints interacting
with their consumers based on the known initial conditions of the brand. The valuation
results can be validated and calibrated to real world sales continuously not just once a year.
 Basically the new “real time” valuation is formed through
interdependent variables like perception, awareness, and preferences operating alongside
characteristics of agents (consumers) as they respond to touchpoints containing messages
about brands. These touchpoints can be anything from direct media messages, to
information from people in their social network, to experiencing the brand directly.
Information from touchpoints interacts with agent personalities and agents update their
views about brands to be higher or lower. Brand equity, the collective experience of how
much brand influences purchase, emerges. It, too, can wax and wane, and it does so as an
aggregate outcome of many agents updating their views and experiences over time.

This is what is happening to Apple through the wisdom of crowds. People are experiencing
a positive feedback loop or increasing returns to scale of the brand performance. The
reality – not the expectation or forecast – is being translated into sales linked to brand
equity. This in turn breeds more positive feedback that raises the expected value of the
brand over time, something that was not predictable until the maturation of the Complexity
Science and ABM fields.

It is little relief to my investment portfolio that I've realized this now. It does, however, make
me want to build the model that explains what is the next powerful brand to break the static
lens I used.

Based in Cambridge, Massachusetts, Greg Silverman is the CEO ofConcentric, a company

empowering their customers to create innovative and breakthrough strategies through the
use of agent-based modeling and complexity science. Connect with him on LinkedIn here.