Consumer Inferences from Product Rankings: The Role of Beliefs in Search Behavior

Abstract

Consumer behavior is not only influenced by realized firm strategies (e.g., price) but also by beliefs about such strategies (e.g., future discounts). Evaluating them typically involves short-term A/B tests, during which consumers' beliefs may differ from those they would hold if the policy were implemented long-term. Failing to account for belief responses may misattribute behavioral changes to model primitives, leading to mispredicting future behavior. We demonstrate this in the context of online product rankings, where consumers typically search among products displayed toward the top of the page first. Through two incentivized lab experiments, we show that these rank effects are due not only to cost differences but also to beliefs, a mechanism largely ignored in existing empirical work. Moreover, beliefs adjust gradually in response to ranking algorithms. Imposing rational expectations leads to incorrect estimates of search costs, implying that it is more costly to search among top-of-page products. We show that consumers are less susceptible to platform steering when beliefs are modeled, and that welfare losses are overstated when beliefs are misattributed to costs. 

Keywords: Consumer Search, Consumer Beliefs, Experiments, Position Effects, Product Rankings, Platform Steering

JEL Classification: D83, C93, L81, D84