The success of a direct marketing campaign is driven by the ability of companies to estimate customers’ future contribution to their profitability. Especially when considering that in retailing companies are wasting resources when targeting customers who will make purchases even in case they would not receive a mailing. We present an advanced profit evaluation, which rates customers for the net impact of a campaign on their buying behavior. Moreover, in contrast to current practices and theory, we model each part of the profit function to improve the accuracy of expected customer value. We employ logistic regression and multiple linear regression to estimate future purchase probabilities and customer expenditures. Variables of different types are considered and a variable selection technique is used to avoid overfitting. To validate our findings, we implemented the method into the mailing system of a European retailer. Our results are of major importance for direct marketing managers, since they make the company’s total profit increase by 5 per cent. This result can be attributed to both a reduction of the optimal mailing depth by 65 per cent, which shows that current procedures lead to systematic ‘overmailing’, and a modified ranking of the customers in the segmentation list.