Designing Demand Information Disclosure in the Presence of Capacity Constraints: A Large-Scale Randomized Field Experiment on a Matching Platform

Abstract

A common issue in online dating platforms is that many users focus their attention on a subset of popular peers, which leads to congestion and thus lower matching rates. Our study examines the results of a randomized natural field experiment, wherein our corporate partner, a large online dating platform, explored information sharing as a means to resolving the congestion problem, providing a random set of users with information on peers’ recent demand. We examine whether and for whom this demand information disclosure facilitated the redistribution of attention away from popular individuals, reducing demand concentration. In doing so, we first consider that two competing effects may emerge with demand disclosure: on the one hand, demand information may signal (a lack of) available capacity, which may attenuate demand concentration (capacity effect); on the other hand, demand information may signal quality, which in turn may exacerbate demand concentration (popularity effect). Although prior work on online platforms for the sale of goods and services has reported evidence in support of both effects, our work is one of the first to explore either relationship in online dating. Online dating platforms differ from other marketplaces in important ways, differences that call into question the likelihood that either effect will arise. Moreover, online dating markets incorporate unique platform dynamics that are likely to drive heterogeneity in the relative strength of the competing effects. Most notably, males and females (in heterosexual dating markets) face starkly different levels of competition; males typically outnumber females by a ratio of more than 2:1, which raises the prospect of very different responses to demand information across gender lines, and in turn, presents a targeting opportunity. Our results show that demand information disclosure increases the average user’s attention toward low-demand partners, decreases their pursuit of high-demand partners, and thereby drives a higher number of successful matches. Heterogeneity analyses further demonstrate that these average results are driven by the larger representation of males in the population, as females are largely unresponsive to the treatments. Our study provides theoretical implications for the literature on online dating and platform design.