CMStatistics 2022: Start Registration
View Submission - CMStatistics
Title: A scalable dynamic bayesian mixture model for fine-grained marketing mix analysis of digital coupons Authors:  Gourab Mukherjee - University of Southern California (United States) [presenting]
Abstract: A novel dynamic mixture model is developed for analyzing the effects of varied marketing components in a digital promotion campaign that uses online coupons. A key feature of the proposed model is that it segments customers based on their purchase history and provides fine-grained estimates of the heterogeneous effects that marketing mix variables have on the different consumer segments. The proposed model captures not only long-term heterogeneous segments in the customer pools, but also tracks short-term changes in customer engagement through dynamic indices that tabulate stocks of unresponded recent coupons. We conduct Bayesian estimation of the model parameters by using a novel Gibbs algorithm, which is highly scalable due to the usage of Polya-Gamma distributions based data-augmentation strategy in handling Binomial likelihoods of customer responses to promotional coupons. Finally, through a path-algorithm we provide an integrated framework for providing fine-grained analysis of the marketing component effects at various levels of heterogeneity. We establish large-sample properties on the operational characteristics of the developed algorithm. We apply the proposed model to recent consumer response data from the apparel industry and obtain encouraging results.