An agent-based diffusion model with consumer and brand agents
Introduction
The diffusion of innovations by consumers is a topic of ongoing interest to researchers and marketing managers. Diffusion of innovations in the social system reflects adoption decisions made by individual consumers. These decisions are made in a complex, adaptive system and result from the interactions among individual's personal characteristics, perceived characteristics of the innovation, and social influence [12].
There are two broad approaches to modeling diffusion: econometric and explanatory. Econometric approaches, such as the Bass [1] model, “describe and forecast the diffusion of an innovation in a social system” [21, p. 1]. Econometric approaches forecast growth within a product category by modeling the timing of first-purchases of the innovation by consumers [21]. Econometric modeling is most applicable when market growth rate and market size are of primary interest [28].
Explanatory models, such as the “consumer diffusion paradigm” proposed by Gatignon and Robertson [12], establish that the diffusion of a product in a defined market is equivalent to the aggregation of individual consumer adoption decisions. Adoption is primarily a function of three sets of consumer-related factors: personal characteristics, perceived innovation characteristics, and the consumer's exposure to social influence [12], [30]. The explanatory modeling approach is most applicable when developing marketing strategies such as segmentation, targeting, and product positioning [21]. This type of model does not support diffusion curve development.
Although established models have provided a strong basis for diffusion research, they do have limitations. ABM has the potential to complement these approaches since it addresses many of their limitations. However, since outputs differ among the models, ABM does not replace econometric or explanatory models [10], [36]. In established models, the ability to reflect real-world dynamics is limited since they do not incorporate the effect of interactions that occur among members of the social system (i.e., market segment) and brands during the diffusion process [10]. Their predictive power is especially poor in cases where the consumer population is very diverse or social influence has a significant effect on the diffusion process [16]. Further, since the unit of study in econometric models is at the aggregate rather than individual unit level, differences among brands, such as marketing mix characteristics, cannot be explicitly incorporated in the model. As a result, one cannot study the effect of marketing mix strategies on diffusion at the brand level. The ability to see this effect is of particular interest to the brand manager who is charged with understanding outcomes at this level in order to make strategic decisions.
The goal of this study is to address the limitations of current diffusion models by developing an agent-based diffusion model with consumer and brand agents. Utilizing this methodology allows characteristics of the consumers and brands in the market to be incorporated in the model. Agent-based modeling (ABM) also allows interactions and emergent behaviors in the simulated market to be studied. Modeling diffusion at the brand rather than product-category level offers insights of interest from both research and management standpoints. This approach to modeling reveals the diverse set of brand-level diffusion curves that result from differences among brands. These brand-level diffusion curves aggregate to form the product-category level curve.
Agent-based modeling is an emerging methodology well suited for studying complex, adaptive systems, such as those seen in diffusion studies. Brand-level diffusion models are especially complex since there are: multiple brands; many and diverse factors define each brand; “order-of-entry effects” [2]; and consumers and brands that may adapt to changing market conditions [11]. In a real-world system, for example, such adaptation might occur as a consumer is influenced by an opinion leader. ABM overcomes the limitations of econometric models in three ways. First, members of the market can be defined in terms of their attributes. Consumers, for example, can be defined in terms of factors such as their personal characteristics while brands can be defined in terms of their marketing mix characteristics and entry timing. Second, interactions among agents may be defined and simulated. Third, the agent adaptations that occur as the result of interactions can be modeled further adding to the simulation's ability to model real-world phenomenon [35].
Perhaps because of the widespread use of agent-based models in sociology, many of the current applications of ABM in diffusion research focus on the effect of the social network on adoption [10]. As a result of the network focus, factors at the consumer and product level that directly affect diffusion are not explicitly incorporated in these models. The ABM developed in this study clearly departs from others' emphasis on social network influence. This model incorporates a comprehensive set of consumer and brand attributes to develop brand-level diffusion curves then aggregates them to develop a product-category diffusion curve.
Little diffusion research incorporating the effect of complex adaptive systems is found in the innovation literature [35]. The current research contributes to the field by developing a complex, adaptive model that simulates key factors and interactions that occur in the market during the diffusion process. Further, since this approach incorporates consumer and brand characteristics, it may be used as a decision tool to compare the outcomes of different strategic choices during the marketing strategy development process.
The remainder of this paper will discuss diffusion theory, provide an overview of ABM, describe the agent-based diffusion model used in this study, review the simulation results, and end with conclusions.
Section snippets
Theory
The diffusion literature includes two major streams: econometric models and explanatory models. Econometric models vary in terms of the factors and algorithms they use to model purchases by customers in order to forecast product-category growth. These models typically reflect only first-time purchases [21]. The Bass [1] model, upon which much diffusion research is based, forecasts the probability of individuals adopting at a given point of time as the innovation diffuses based on whether they
Overview
ABM is a methodology for modeling complex systems. It has been applied in many disciplines including engineering, economics, sociology, and the natural sciences; its acceptance in the management sciences is emerging [10], [39]. Compared to other simulation methods, ABM offers several advantages especially in terms of modeling complex systems. For instance, in other models, the unit of study is the population. In ABM, the unit of study is the individual or agent. Definitions of individual system
Model overview
The primary objective in developing the ABM for this study is to demonstrate the feasibility and advantage of using such a model to understand the complex interactions that occur at the brand level during the diffusion process for a durable product. Durable product diffusion models are distinct since these products, compared to non-durable products, are purchased once or infrequently [1]. Similar to the approach used by Delre et al. [9], the agents in this ABM reside in a fixed network with a
Simulation overview
For this study, we use the digital camera market as an illustration of how consumer and brand interaction influences adoption decisions. This example illustrates how ABM supports the development of scenarios to understand the effect of changing brands' marketing mix variables on diffusion at the brand and product-category levels. First, a model representing the baseline market is developed. Then, model parameters are adjusted to investigate the effects of changes in marketing mix factors on
Simulation results
In Simulation 2, Brand Agents B, C, and D are altered to reflect enhanced features while leaving all other factors unchanged. By comparing the output details generated for the product-category level diffusion curves from the baseline or Simulation 1 (Fig. 4) and Simulation 2 (Fig. 5), it is seen that by Year 8, the result of introducing enhanced features into the market in Years 2 and 3 increases the number of adopters by 66% relative to the baseline. Thus, Simulation 2 demonstrates the
Limitations and future research
Several assumptions were made in the current simulation that represent limitations to be addressed in future research. First, the current simulation only considers interactions among consumers and between consumers and brands. Additional development incorporating interactions among brands would further advance this research by incorporating the effect of brand agents on one another in terms of strategy adaptation and entry/exit behavior. Such a simulation would also allow managers to better
Conclusion
ABM represents a new methodology with the potential to advance the study of diffusion of innovative products. It builds on extant econometric and exploratory diffusion models by advancing understanding of interactions among agents in the system and by incorporating adaptive responses by agents to system changes [10]. Further, interactions that occur during ABM simulation runs can reveal emergent results which may not be evident when studying system elements individually [36]. The ability to
Mary E. Schramm received her MBA from the University of Wisconsin-Oshkosh. She is currently a Ph.D. candidate in Marketing at Kent State University. Her research interests include product innovation, diffusion, new product development, and healthcare marketing. Her industry experience includes product marketing and new product development in industrial and healthcare markets.
References (39)
- et al.
Targeting and timing promotional activities: an agent-based model for the takeoff of new products
Journal of Business Research
(2007) The four P's in social simulation, a perspective on how marketing could benefit from the use of social simulation
Journal of Business Research
(2007)- et al.
A simultaneous model for innovative product category sales diffusion and competitive dynamics
International Journal of Research in Marketing
(1999) - et al.
An application of agent-based simulation to knowledge sharing
Decision Support Systems
(2009) - et al.
Agent-based simulation of consumer purchase decision-making and the decoy effect
Journal of Business Research
(2007) A new product growth model for consumer durables
Management Science
(1969)- et al.
Order of entry as a moderator of the effect of the marketing mix on market share
Marketing Science
(1996) - et al.
Competitive strategies for late entry into a market with a dominant brand
Management Science
(1990) - et al.
Dynamic models incorporating competition
- et al.
The innovation diffusion process in a heterogeneous population: a micromodeling approach
Management Science
(1990)
Diffusion dynamics in small-world networks with heterogeneous consumers
Computational & Mathematical Organization Theory
Uses of agent-based modeling in innovation/new product development research
Journal of Product Innovation Management
Competitive reactions to market entry: explaining interfirm differences
Journal of Marketing Research
A propositional inventory for new diffusion research
Journal of Consumer Research
Innovative decision processes
Kodak (A) — Case 9-703-503
Advertising and the diffusion of new products
Marketing Science
Cited by (63)
The group decision-making evolution of green consumption behavior under overreaction: Based on multi-agent simulation analysis
2023, Resources, Conservation and RecyclingInternational strategic alliances for collaborative product Innovation: An agent-based scenario analysis in biopharmaceutical industry
2023, Journal of Business ResearchDevelopment of an agent-based model to understand the diffusion of value management in construction projects as a sustainability tool
2022, Alexandria Engineering JournalCitation Excerpt :Diffusion models are one of the most commonly used tools to predict and analyze the spread of a product or innovation over time, and they can also show how diffusion affects agents’ decision to adopt that product [31,46]. Research on diffusion requires understanding how products, services, and ideas spread in a society over time, taking into account concepts like imitation, social norms, and coordinated spreading, among others, that are not considered in a top-down approach [61,65]. Eliashberg et al. [21] stated that “diffusion models that start at the micro level have a rich potential in terms of a better understanding of the diffusion process and as a tool for managerial action.”
A novel diffusion-based model for estimating cases, and fatalities in epidemics: The case of COVID-19
2021, Journal of Business ResearchDiffusion of green products in industry 4.0: Reverse logistics issues during design of inventory and production planning system
2020, International Journal of Production EconomicsIndustry 4.0 and circular economy: Operational excellence for sustainable reverse supply chain performance
2020, Resources, Conservation and Recycling
Mary E. Schramm received her MBA from the University of Wisconsin-Oshkosh. She is currently a Ph.D. candidate in Marketing at Kent State University. Her research interests include product innovation, diffusion, new product development, and healthcare marketing. Her industry experience includes product marketing and new product development in industrial and healthcare markets.
Kevin J. Trainor is an Assistant Professor of Marketing in the Richard J. Wehle School of Business at Canisius College. He holds a Ph.D. in Marketing from Kent State University. His research interests include new product development, product innovation, and the interface between marketing and R&D. His research is published or forthcoming in Journal of Business Research, Journal of Business and Industrial Marketing, and Journal of International Management Studies. His industry experience includes roles in software development and product management.
Murali Shanker is a Professor in the Department of Management & Information Systems, Kent State University. He holds a Ph.D. from the University of Minnesota. His research is published in INFORMS Journal on Computer, IIE Transactions, Journal of the Operational Research Society, Decision Support Systems, and Decision Sciences. His research interests lie in Distance Learning, Distributed Computing, Open Source, Simulation, and Neural-Network Modeling.
Michael Y. Hu has a Ph.D. from the University of Minnesota in management science/marketing. Currently he holds the Bridgestone Chair in International Business and is a Professor of Marketing at Kent State University. He has published over a hundred and thirty academic articles in the areas of applications of marketing, international business and artificial neural networks. His research has appeared in Journal of Marketing Research, Marketing Letters, International Journal of Research in Marketing, Decision Sciences, European Journal of Operational Research and many others. He won the University Distinguished Teaching Award in 1994 and the University Distinguished Scholar Award in 2006 at Kent State University.