This article addresses an integrated warehouse order picking problem. The company HappyChic is specialized in men’s ready-to-wear. A central warehouse is dedicated to supplying, every day, the shops of one brand. We focus on the picking area of this warehouse which relies on human picking system. For each picking wave (period of a working day), a set of customer orders has to be prepared. An order is a set of product references, with quantities, i.e., the numbers of items required. The problem consists in jointly deciding: (1) the division of orders into several boxes, respecting weight and size constraints; (2) the batching of boxes into trolleys, that implicitly defines the routing into the picking area. The objective function aims to minimize the total distance. To deal with the large size instances of HappyChic in short computation times, we design a heuristic method based on the split and dynamic programming paradigms. The results are very convincing: the total covered distance decreases by more than 20%. Moreover, we propose an adaptation of the algorithm to prepare homogeneous boxes with respect to classes of products. The logistic department of HappyChic is convinced by results obtained in this research work, and the warehouse management system is currently being updated in order to integrate the proposed algorithm.