The purpose of this study is to develop a methodology to guide managers in determining the optimal promotion campaigns to be directed towards different market segments in order to maximize the value of customers. For the purposes of this study, a two-step methodology is used, based on stochastic dynamic programming and the classification and regression tree. This methodology groups the customers according to their value. Within this framework, an experiment is conducted in which each of the different promotion campaigns is assigned to different randomly selected groups. The impact of each type of promotion on each type of market segment is analysed in order to find the optimal promotion campaigns appropriate for each. In contrast to previous research, this study takes into account a firm that provides more than one specific type of product or service. In addition, it analyses the impact of widely used types of promotion campaigns compared with the narrow scope of those investigated in previous studies. Therefore, this research presents important insights into managing relations with the customers in a more interactive and profitable way.