Product Recommendation and Consumer Search
48 Pages Posted: 14 Jun 2019
Date Written: June 4, 2019
We study product recommendations in an online environment where a firm provides strategic product recommendations to consumers. We develop an analytical framework to integrate recommendations into the consumer product search process. The firm sells two imperfectly substitutable products with different profit margins and makes a personalized product recommendation to each consumer based on its knowledge of their preferences. We define bias as the firm’s deliberate decision to recommend the product with higher expected misfit cost to a consumer. Consumers can accept the product recommendation, search for the non-recommended product, or leave the website. We identify five consumer segments based on heterogeneous consumers’ responses to the firm’s recommendations. We show that the level of bias in product recommendations, the firm’s profit, and consumer surplus depend on the interaction between the firm’s information about consumer preferences and consumer search costs. An increase in information about consumers can lead to an increase or a decrease in the level of bias in product recommendations, depending on the level of search costs. Moreover, an increase in information about consumers leads to a corresponding increase in the firm’s profit, but does not necessarily result in a reduction in consumer surplus. On the other hand, an increase in search costs can lead to non-monotonic changes in the firm’s recommendation strategy, causing an increase or decrease in recommendation bias when the firm’s information about consumers is more accurate. Furthermore, the firm’s profit can behave non-monotonically with respect to search costs: the firm benefits from an increase in search costs when these costs are small and uncertainty about consumers is low, but it can be adversely affected when search costs are moderate. Interestingly, consumer surplus may increase when search costs increase.
Keywords: recommendation systems, product recommendations, bias, analytical modeling, consumer search
Suggested Citation: Suggested Citation