Retail Assortment Planning in the Presence of Consumer Search
multinomial logit model
Consumers often know what kind of product they wish to purchase, but do not know which specific variant best fits their needs. As a result, a consumer may find an acceptable product in one retailer but nevertheless purchase nothing, opting to search other retailers for an even better product. We study several models of retail assortment planning, some of which explicitly account for consumer search and one that does not, which we call the “no-search” model. Even though the no-search model never includes an unprofitable variant in the assortment, in the presence of consumer search, it may indeed be optimal to include an unprofitable variant. Furthermore, we find that the no-search model can lead to an assortment with an expected total profit that is significantly less than optimal. In the extreme, the no-search model may recommend closing down a category (i.e., carry no variants) even if a profitable assortment exists (a 100% profit loss). We conclude that failing to incorporate consumer search into an assortment planning process may cause a retailer to underestimate the substantial value a broad assortment has in preventing consumer search. We discuss how the insights from our stylized models may apply to actual assortment planning processes.