Assessing the influence of marketing activities on customer behaviors: a dynamic clustering approach

This research presents an application of a mixed hidden Markov model to data from a multichannel retailer. The objective of this research is to develop a dynamic model of channel choice and purchasing behavior that accounts for consumer heterogeneity, changes in behavior over time, and the influence of marketing activities on managerially relevant consumer behaviors. The model allows marketers to reduce their direct mailing spending while controlling for potential negative effects on their sales. More specifically, we develop a model that captures the evolution of a consumer’s buying behavior over time across retail channels and compare our model to several other approaches. We find our model outperforms existing models including standard latent class models, including those belonging to the latent transition analysis framework. Using several criteria of model performance and fit, we find a hierarchical clustering structure in the data. Each cluster responds differentially to marketing activities. We find catalogs, on average, are an effective tool to keep consumers active whereas retail promotions are more likely to influence consumers to migrate to another channel.