Customer Evolution in Sales Channel Migration.

We develop and estimate a model depicting the evolution of a newly-acquired customer’s channel choice decision process from a “trial” stage to a “post-trial” stage. The model consists of two logit choice models, a “trial” model and a “post-trial” model, linked by a geometric process that governs how quickly the customer moves from trial to post-trial. Each logit model is driven by channel preferences, state dependence, and marketing. We utilize data for a book retailer who sells through retail stores, the Internet, and catalogs, and estimate the model using Bayesian methods. Our results suggest (1) customers’ decision processes do evolve, (2) within our 4-year window, while the majority of customers stay with their original choice process, a sizeable segment switches decision processes, (3) as expected, “switchers” move from a decision process dominated by state dependence and marketing to one driven by channel preference with lower reliance on marketing, and (4) the “switcher” segment is multichannel, whereas the “stayer” segment is single channel. (5) identifying these segments can improve right-channel management. We discuss implications of our results for researchers and managers.