Big data analytics in returns management – Are complex techniques necessary to forecast consumer returns properly?

The more people shop online, the more consumer returns e-tailers face. In order to plan the returns management process capacity adequately, it is necessary to forecast the expected amount of returned parcels. Big data analytics provides a vast number of methods to perform such tasks. However, it should be noted that particularly small- and medium-sized e-tailers lack the capabilities and resources to employ such complex techniques. Against this background, this paper analyses the performance of several data analysis methods that differ in application complexitiy using real data from an apparel e-tailer. On the one hand, we find that –as expected– complex methods outperform simple ones. On the other hand, and from a practitioner’s perspective probably even more interesting, we also conclude that a binary logistical regression as the simplest analyzed method may already provide satisfactory results. The findings indicate that the use of big data analytics is of great value to effectively and efficiently manage consumer returns – even if not the most sophisticated state-of-the-art method is used.