We are happy to announce that Prof. Dr. Thomas Reutterer from WU Wien – Vienna University of Ecomics and Business will be our visiting us as a guest speaker in the Business Economics Research Seminar. The topic of his talk is “Leveraging Purchase Timing Regularity for Predicting Customer Behavior”
In non-contractual business settings, stochastic customer behavior models based on recency-frequency (RF) data are popular tools to assist marketing analysts for making predictions about a customer’s activity status and future purchase propensities. Knowledge about these quantities is crucial for accurately estimating residual Customer Lifetime Value (CLV) of a company’s customer base and for managing the customer portfolio accordingly. In this research, we build on a recently revived discussion on the potential benefits of extending the RF-framework for customer-base analysis by adding another individual-level metric summarizing (ir)regularities in inter-transaction timing. We develop a new class of stochastic consumer behavior models which generalize the repeat-buying component of well-established baseline models while retaining the same level of data requirements and algorithmic efficiency. Using an extensive simulation study we can show that ignoring regularity leads to an average overestimation bias of 10% false positives. In empirical settings, predictive errors can be reduced by up to 14%. Compared to more complex models, this gain comes at no significant additional cost, neither in terms of data requirements or computation time. Against the background of your findings and recently available open source software implementations of our newly developed models, we expect a wider range of marketing analysts and data scientists to benefit from improving the accuracy of their predictions in the presence of regularity.