Skip to main content
Marketing and Consumer Behavior

Modeling the ‘Why’ Behind Consumer Behavior

By Eliana Chow

A new paper out of NC State takes a closer look at agent-based modeling, which uses a computer simulation to understand how individual consumers (or “agents”) act within an environment determined by the researcher.

“With this paper, we aim to provide researchers and industry practitioners with a foundational guide to agent-based modeling as they work toward a better understanding of new product development and diffusion,” says co-author Bill Rand, executive director of the Business Analytics Initiative and associate professor of marketing at Poole College of Management.

Rand and his co-author, Bielefeld University’s Christian Stummer, identify three key strengths of agent-based modeling. First, the model can be highly individualized to represent a wide range of consumer behavior when it comes to things like advertising, social influence and personal willingness to adopt innovation. By differentiating between early and late adopters of new products, for example, researchers can measure how long it will take for a product to catch on and how to accelerate or delay that process.

Second, agent-based modeling can simulate interactions between different types of consumers to understand how these encounters influence someone’s buying decisions. For instance, they can determine how are sales are affected when a brand-loyal consumer interacts with someone who has never heard of the brand.

Third, agent-based models make it easier for researchers to explore the effects of different advertising strategies on consumer behavior. Is it good for sales when a company throws a launch party for their new product, advertises the event on Instagram and gives out free branded merchandise? Researchers can program this scenario into their model and simulate how each consumer responds.

On the flipside, the open-ended construction of agent-based modeling has drawn criticism. Since it’s a newer technology, there aren’t as many established standards, restrictions or precedents in the literature for how far researchers can go with their finetuning. Rand underplays this line of criticism. “It’s true that you can make a model do anything, but the same can be said about regression or game theory modeling,” he says. “The key is being able to prove that the principles of your model are true or beneficial.” 

It’s true that you can make a model do anything, but the same can be said about regression or game theory modeling. The key is being able to prove that the principles of your model are true or beneficial. 

People tend to be more comfortable with the concrete values derived from statistics-based models, Rand acknowledges. “A second major criticism of agent-based models is that they simply simulate the potential ways sales could unfold, but there are always more factors to consider,” he says. “Real-life consumers are more complex than a computer can perfectly model. Researchers must then backup their simulation with tangible data to demonstrate that the model’s assumptions are as close as possible to how interactions play out in the real world.”

A third criticism of the model is computational cost. Rand and his co-author acknowledge that these models can take a long time to finetune. Evaluating millions of simulated consumers over that period can require significant resources, both in terms of upfront cost and computational power. Yet the benefits seem to outweigh the cost. “When we use cheaper equation-based models like game theory, there’s no reference to an individual,” Rand reflects. “Agent-based modeling helps us focus on why consumers are making their unique decisions, which is invaluable for furthering market research.”

As agent-based modeling continues to establish itself in new product market diffusion research, Rand is eager to explore new frontiers. “A lot of the models we’ve built so far make predictions based on how many people a consumer speaks to or how many advertisements they encounter,” he says. “But there’s a world of opportunity to model more nuanced cognitive behaviors. For example, not every word-of-mouth communication is the same. Having hundreds of friends on Facebook or Twitter tell me about a product has a different impact than hearing about it from someone I meet at a block party. I’m looking forward to seeing more models start to reflect the everyday realities of complex human beings.”

The paper, “Agent-Based Modeling of New Product Market Diffusion: An Overview of Strengths and Criticisms,” is published in the Annals of Operations Research. The paper was co-authored by William Rand of North Carolina State University and Christian Stummer of Bielefeld University.

Note to Editors: The study abstract follows.

“Agent-Based Modeling of New Product Market Diffusion: An Overview of Strengths and Criticisms”

Authors: William Rand, North Carolina State University and Christian Stummer, Bielefeld University

Published: February 10, 2021, Annals of Operations Research

DOI: 10.1007/s10479-021-03944-1

Abstract: Market diffusion of new products is driven by the actions and reactions of consumers, distributors, competitors, and other stakeholders, all of whom can be heterogeneous in their individual characteristics, attitudes, needs, and objectives. These actors may also interact with others in various ways (e.g., through word of mouth or social influence). Thus, a typical consumer market constitutes a complex system whose behavior is difficult to foresee because stochastic impulses may give rise to complex emergent patterns of system reactions over time. Agent-based modeling, a relatively novel approach to understanding complex systems, is well equipped to deal with this complexity and, therefore, may serve as a valuable tool for both researchers studying particular market effects and practitioners seeking decision support for determining features of products under development or the appropriate combination of measures to accelerate product diffusion in a market. This paper provides an overview of the strengths and criticisms of such tools. It aims to encourage researchers in the field of innovation management, as well as practitioners, to consider agent-based modeling and simulation as a method for gaining deeper insights into market behavior and making better-informed decisions.