[en] Recently, considerable attention has been given to the prediction of customer lifetime value (CLV). To do so, one needs to predict the expected number of transactions a customer makes in the future and the average benefits per transaction. The CLV is then calculated as a product between these two numbers. The Pareto/NBD model is the well-known method in predicting the number of transactions however; it is a difficult model to apply due to difficulties in estimating the model parameters. The recently developed beta-geometric/NBD (BG/NBD) model simplifies the task and is easy to implement. The independency between the number of transactions of a customer and the related profit per transaction is an important assumption in the BG/NBD model. In this paper, we propose an alternative approach and show that the dependence between the number of transactions and their profitability increases the accuracy of the CLV prediction.