A Dynamic CRM Model Based on POMDP

This paper addresses the issue of modeling and understanding the dynamics of customer relationships. Our approach to modeling relationship dynamics is structurally different from the models in the existing literature. We construct a Partially Observable Markov Decision Process (POMDP) to relate the latent relationship states to the observed buying behavior. The POMDP enables the marketer to estimate which relationship state the customer stay, what kind of effect the marketing actions can make to change the relationship state, by maximizing the Customer Lifetime Value, the models can get the optimized policies for customer relationship management. Furthermore, we takes the model with three states (inactive, active, loyalty) and three actions(noaction, retention, reactive) as an example to illustrate how the model proposed above works.