Predicting changes in individual customer behavior is an important element for success in any direct marketing activity. In this article we develop a hierarchical Bayes model of customer interpurchase times based on the generalized gamma distribution. The model allows for both cross-sectional and temporal heterogeneity, with the latter introduced through the component mixture model dependent on lagged covariates. The model is applied to personal investment data to predict when and if a specific customer will likely increase time between purchases. This prediction can be used managerially as a signal for the firm to use some type of intervention to keep that customer.
The analysis of customer value in direct marketing typically combines customer timing and quantity data into a single statistic that is used to compute lifetime values, rank-order customers for differential action, and identify prospects for cross-selling. However, current models assume that purchase timing and quantity decisions are independently realized (i.e., uncorrelated) over time given individual-level parameters. In this article, the authors show that customer value calculations can be severely biased in these models when timing and quantity are dependently related. The authors propose alternative models that lead to substantial gains in profitability in two direct-marketing data sets. The results indicate that the commonly held assumption of independence leads to an overvaluation of customer value.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.