The Power of Rankings: Quantifying the Effect of Rankings on Online Consumer Search and Purchase Decisions

Abstract

Online search intermediaries, such as Amazon or Expedia, use rankings (ordered lists) to present third party sellers' products to consumers. These rankings decrease consumer search costs and increases the probability of a match with a seller, ultimately increasing consumer welfare. Constructing relevant rankings requires understanding their causal effect on consumer choices. However, this is challenging since rankings are endogenous: highly ranked products are also the most relevant ones for consumers. In this paper, I use the first data set with experimental variation in the ranking from a field experiment at Expedia to identify the causal effect of rankings. Using this data set, I make three contributions. First, I show that rankings affect what consumers search, but conditional on search, do not affect purchases. I also exploit a feature of the data set (opaque offers), to show that rankings lower search costs, instead of affecting consumer expectations or utility. Second, I quantify the effect of rankings using a sequential search model and find an average position effect of $2.64, lower than previous estimates in the literature obtained without experimental variation. Finally, I show that a utility-based ranking build on this model's estimates leads to an almost twofold increase in consumer welfare, while also increasing transactions and revenue for the search intermediary.

An Economic Policy Perspective on Online Platforms

“…The "advisory theory" of search rankings is put to an empirical test by Ursu (2015). Using Expedia hotel booking data he confirms the existence of a conflict of interest between the platform operator and users.…”

Optimal Search and Awareness Expansion

“…Though similar ranking effects can be induced by either random search where the order by which products are searched is exogenously given, or by introducing product specific search costs into directed search (e.g. Ursu, 2018). As the next section highlights, however, both approaches lead to distinct search paths and market demand.…”

“…Given the prevalence of ranking effects in click-stream data, empirical studies have proposed an approach to alleviate this problem (e.g. Chen and Yao, 2017;Ursu, 2018). In this approach, search costs are modeled such that they depend on the position at which the alternatives are revealed to the consumer.…”

Mercados a través de internet

“…No obstante, aunque hace tiempo que se aportó evidencia empírica de que el orden en el que se presenta un listado de opciones condiciona la decisión final (rubinstein y Salant, 2006), la cuantificación de los efectos resulta difícil en los mercados que operan a través de internet, en la medida en que los usuarios prestan más atención a las ofertas que están situadas en los primeros lugares y, al mismo tiempo, las plataformas ordenan las ofertas situando los productos más relevantes los primeros. Es decir, se presenta un problema de endogeneidad que, a pesar de las distintas aportaciones realizadas en este ámbito, mantienen abierto el debate sobre la influencia de los rankings en las decisiones de los consumidores (Ursu, 2017).…”