Valuing customer is essential to any firm and enables marketers to identify key customers. In non-contractual settings, such as retailing, probabilistic customer attrition models are often used to predict customer life time value (CLV). These models are capable of reflecting the first principles of customer behavior: the process of buying a product and the process of churning. While predictions of CLV on an aggregate customer level are fairly good, current models suffer from relatively poor accuracy for predictions of an individual CLV. A possible explanation is that the most popular models do not consider the most important contextual determinants. For example, they do not model a firms direct marketing activities.
In this project we develop an extension to the most popular probabilistic model to explicitly consider contextual determinants . Thus, we are not only capable of increase the accuracy of customer lifetime value predictions, but we can also identify the impact of these contextual determinants. For example, we enable marketers to measure and understand the impact of direct marketing campaigns and thus, to improve the campaigns efficiency.