Customer lifetime value (CLV) is a key-metric within CRM. Although, a large number of marketing scientists and practitioners argue in favor of this metric, there are only a few studies that consider the predictive modeling of CLV. In this study we focus on the prediction of CLV in multi-service industries. In these industries customer behavior is rather complex, because customers can purchase more than one service, and these purchases are often not independent from each other. We compare the predictive performance of different models, which vary in complexity and realism. Our results show that for our application simple models assuming constant profits over time have the best predictive performance at the individual customer level. At the customer base level more complicated models have the best performance. At the aggregate level, forecasting errors are rather small, which emphasizes the usability of CLV predictions for customer base valuation purposes. This might especially be interesting for accountants and financial analysts.

customer lifetime value, customer relationship management, customer segmentation, database marketing, forecasting, interactive marketing, value
Statistical Decision Theory; Operations Research (jel C44), Forecasting and Other Model Applications (jel C53), Financial Forecasting (jel G17), Business Administration and Business Economics; Marketing; Accounting (jel M), Marketing (jel M31)
hdl.handle.net/1765/325
ERIM Report Series Research in Management
Erasmus Research Institute of Management

Donkers, A.C.D, Verhoef, P.C, & de Jong, M.G. (2003). Predicting Customer Lifetime Value in Multi-Service Industries (No. ERS-2003-038-MKT). ERIM Report Series Research in Management. Retrieved from http://hdl.handle.net/1765/325