Linking customer profitability predictions to direct marketing efficiency
[Prof. Dr. Corinne Faure, Goethe University Frankfurt]
Using data mining techniques, we develop a probabilistic model to estimate future earnings from existing customers in contractual relationships. The model is an extension of existing direct marketing models and relies on the estimation of individual cross-selling probabilities based on profiling.
After developing the model theoretically, we first empirically test its predictive power on a 3-million random sample of private customers from a large German retail bank. We then empirically assess the ability of the model to predict the efficiency of direct marketing contacts. Our results indicate that this relatively simple model performs well in both respects and can therefore be used to optimize the allocation of direct marketing budgets across customers.