The Choice Overload Effect in Online Recommender Systems: Theoretical Framework and Field Experiment

Abstract

Firms are increasingly relying on recommender systems to help guide consumer choice. An important but under-studied question is how many products to offer in a recommendation set. In this work, via a field experiment involving 1.6 million consumers on an online retail platform, we causally demonstrate that consumers' likelihood of making a purchase first increases and then decreases as the number of products in a recommendation set grows. Importantly, as much as 64% of the decrease in purchase rate (i.e., the choice overload effect) can be attributed to a decrease in consumers’ likelihood of starting a search (i.e., clicking on any recommended product). We illustrate via a two-stage behavioral choice model that these results are consistent with anticipated regret (as opposed to information overload) as the main mechanism of the choice overload effect. We further discuss alternative mechanisms and analyze heterogeneous treatment effects via both reduced-form regressions and a causal forest approach. Altogether, this work presents real-world experimental evidence for the choice overload effect in recommender systems, highlights the importance of consumer search behavior in driving this effect, and provide insights into when limiting the number of options in a recommender system may be particularly beneficial to online retailers.

Keywords: choice overload, search cost, anticipated regret, field experiment, recommender systems, retailing