How Do Recommender Systems Lead to Consumer Purchases? A Causal Mediation Analysis of a Field Experiment

The findings underscore the important role of consumers’ consideration sets in mediating the positive effects of recommender systems on consumer purchases. Practical strategies can be developed to facilitate the formation of the consideration sets. For example, to reduce consumers’ search costs and cognitive efforts, online retailers can display the recommended products in a descending order according to the predicted closeness of consumers’ preferences. Online retailers can further indicate the predicted closeness scores of consumers’ preferences for the recommended products. Given such a placement arrangement, consumers can quickly screen the recommended products and add the most relevant alternatives to their consideration sets, which should facilitate consumers’ shopping process and increase the shopping satisfaction. The findings also suggest that a larger consideration set due to the use of recommender systems could induce consumers to buy. Yet, it is difficult for consumers to manage many alternatives when the consideration set is very large. To facilitate consumers’ shopping process, online retailers need to consider strategies and tools that help consumers manage the alternatives in the consideration set in a better-organized manner and facilitate the comparison across the alternatives. How do recommender systems induce consumers to buy? Extant research neglects to examine the causal paths through which the use of recommender systems leads to consumer purchases. In this study, we conduct a randomized controlled field experiment on the website of an online book retailer and explore the causal paths by employing the recently developed causal mediation approach. Not surprisingly, the results show that the presence of personalized recommendations increases consumers’ propensity to buy by 12.4% and basket value by 1.7%. More importantly, we find that these positive economic effects are largely mediated through affecting the consumers’ consideration sets. Specifically, the presence of personalized recommendations increases both the size of consumers’ consideration sets ( breadth ) and how intensively they are involved with each alternative in consideration ( depth ). It is the two changes that go on to increase consumers’ propensity to buy and basket value. Furthermore, we find that the proportion of the total effects mediated through the breadth of consideration set is much larger and more significant than that mediated through the depth.