Customer Search and Product Returns

Abstract

Online retailers are challenged by frequent product returns. High return rates significantly decrease companies’ profit which makes the issue of managing product returns very important from the practical standpoint. Typically, practitioners study returns in connection with purchase decisions or as a part of customer behavior/type. In this paper, we show that the events which precede the purchase decision are related to the return decision. Generally, this information is readily available to online retailers and thus provides a low-cost opportunity to better understand and predict the product returns.

Based on the data provided by a large apparel retailer, we demonstrate that the way customers search for a product is indicative of product returns. We find correlational evidence that using search filters, spending more time, and purchasing the last item searched are negatively associated with the probability of return. We propose a joint model of search, purchase and return which is based on an analytic model of search, purchase, and returns. Our model is consistent with the findings in the data and provides insight into how search and returns are related. Finally, using a machine learning framework, we demonstrate that adding search data improves the prediction accuracy of individual-level return rate above and beyond prior models.