A Joint Segmentation Model of Consumers and Products Applied to the Sales of Music Albums

We develop a new model that simultaneously captures two principal sources of heterogeneity: variation across segments of consumers and variation across clusters of products. We focus on the recorded music industry as a managerial context, and construct a flexible duration-time model that can be used to capture the dynamic purchase patterns across a broad portfolio of different albums. Our basic premise is that a set of generic consumer segments exist and remain fixed across all albums, and each album (or each cluster of similar albums) can be viewed as drawing different proportions from each of these underlying segments. We allow explanatory variables to have a differential impact on both components of the model, i.e., accelerating purchase rates within segment, and/or changing the proportions drawn by each cluster. As a useful application and validity test for our model, we demonstrate how we can leverage the pooled information available from a set of existing albums to make early and accurate predictions about the sales performance of a new album.