Search Duration

In studying consumer search behavior, researchers typically focus on which products consumers add to their consideration sets (the extensive margin of search). In this article, we attempt to additionally study how much consumers search individual products (the intensive margin of search), by analyzing the time they spend searching (search duration). We develop a sequential search model where consumers who are uncertain (and have prior beliefs) about their match value for a product search to reveal (noisy) signals about it that they then use to update their beliefs in a Bayesian fashion. Search duration, in this context, is an outcome of the decision by a consumer to seek information on the same product multiple times; with a unit of time corresponding to one signal, the more the number of signals sought greater is the search duration. We also show how the model can be used to study revisits - a feature not easily accommodated in Weitzman’s (1979) sequential search model. We build on the framework by Chick and Frazier [Chick, S., P. Frazier. 2012. Sequential Sampling with Economics of Selection Procedures. Management Sci. 58(3), 1-16] for describing the optimal search rules for the full set of decisions consumers make (which products to search, for how long, and whether to purchase), and develop the model’s empirical counterpart. We estimate the proposed model using data on consumers searching for restaurants online. We document that search duration is considerable, even when consumers search few restaurants, and that restaurants that are searched longer are more likely to be purchased. Using our model, we quantify consumer preferences, search costs, and prior uncertainty parameters. We find that consumers start their search with high initial uncertainty, which rationalizes their choice to search few options intensively in the presence of search costs. In counterfactual simulations, we analyze the effect of prior uncertainty and search environment on consumer choices