Most choice models in marketing implicitly assume that the fundamental unit of analysis is the brand. In reality, however, many more of the decisions made by consumers, manufacturers, and retailers occur at the level of the stock-keeping unit (SKU). The authors address a variety of issues involved in defining and using SKUs in a choice model, as well as the unique benefits that arise from doing so. They discuss how a set of discrete attributes (e.g., brand name, package size, type) can be used to characterize a large set of SKUs in a parsimonious manner. They postulate that consumers do not form preferences for each individual SKU, per se, but instead evaluate the underlying attributes that describe each item. The model is shown to be substantially superior to a more traditional framework that does not emphasize the complete use of SKU attribute information. Their analysis also highlights several other benefits associated with the proposed modeling approach, such as the ability to forecast sales for imitative line extensions that enter the market in a future period. Other implications and extensions also are discussed.
