Identifying consumer consideration set at the purchase time from aggregate purchase data in online retailing

Online retailers provide a substantial amount of product information to their customers. The information includes not only product features and customer reviews, but also information on alternative products that may better fit a consumer's needs. The systematic provision of information on alternative products could have a significant impact on consumers' purchase decision process at online retailers. In this study, we analyze one aspect of the impact — the degree to which consumers consider multiple products at the purchase time. We leverage a popular feature provided by online retailers — “What Do Customers Ultimately Buy after Viewing This Item?” We show that information contained in this feature can be used to identify consumers' product consideration set and choice at the purchase time when combined with product sales data. The identification is exact in analyzing competition between two products. For competition involving three products, the identification is exact under the assumption that consumer choice follows a discrete choice model. For competition involving more than three products, the information provides a lower bound of the percentage of consumers that consider only one product at the purchase time. We apply the model to 38,400 unique products from Amazon's Electronics category. The results show that more than 78% of consumers purchase a product without considering any other products on Amazon at the purchase time.

► We identify consideration sets at purchase time from aggregate purchase data.
► Identification requires both product sales data and conditional viewership data.
► Identification is exact in a two-product market.
► Lower bounds of customers who view no alternatives are identifiable in all markets.
► Over 78% Amazon customers do not consider alternatives at purchase time.