Beschreibung
Many firms collect huge amounts of personalized transaction data and integrate them in their customer databases to analyse the profitability of their customers. Prior research shows that focusing marketing efforts on those customers with high estimated lifetime value can result in higher overall profitability and thus in a more efficient investment of scarce marketing resources. However, in spite of their forward looking properties, customer lifetime metrics tend to underestimate the dynamic nature and variety of evolving customer relationship patterns. We present a modelling framework for analyzing the profitability dynamics of noncontractual customer-firm relationships. Our approach is based on the building blocks of the customer lifetime value concept, but explicitly considers the evolution of customer profitability over time. We classify each customers period-wise profitability contributions to the firm into a discrete set of profitability tiers and model period-to-period transitions between these tiers as a Markov process with unknown transition matrix. Unsupervised heterogeneity is captured by assuming a finite mixture of Markov chains. The model parameters are estimated using empirical Bayesian methodology. We demonstrate that the presented Markov chain clustering model enables loyalty managers to detect segments of customers which differ significantly in terms of their profitability evolution over evolving periods. The approach is empirically illustrated using transaction data of a customer cohort acquired by an apparel retailer. We conclude with a discussion of managerial insights gained from the study and outline some extensions of the proposed approach as an agenda for future research.Zeitraum | 12 Juni 2008 → 14 Juni 2008 |
---|---|
Ereignistitel | INFORMS Marketing Science Conference |
Veranstaltungstyp | Keine Angaben |
Bekanntheitsgrad | International |