Information Design in an Online Platform: Evidence from E-commerce

Abstract

This paper analyzes the impact of platform information design on seller product choice, sales performance, and market outcomes. The platform in question offers a data analytics service to sellers, enabling them to subscribe to information about thirty other competitors. We examine a natural experiment where the platform doubled the number of subscribed competitors, provided to a group of sellers through a market intelligence tool. Our findings indicate that with more information, sellers make better product selections, introducing 25% more products, and experience increased sales (39% more). To understand the broader effects, we propose a structural model that combines a seller learning model with product market competition. Our simulations show that more information from the tool benefits consumers and the platform, increasing consumer welfare by 0.32% and total transaction volumes by 3.92%. Interestingly, we observe that the relationship between the amount of information provided and platform transaction volumes or consumer welfare is not strictly monotone. As sellers become more informed about market trends, they tend to discontinue unprofitable products, which may reduce product variety and have a negative impact on the platform and consumers. Finally, our simulations suggest that the platform can achieve maximum transaction volumes by tripling the amount of information provided through the tool. Our findings have significant managerial implications regarding the design of platform information structures and the growth of sellers.

Keywords: Information Design, Data Value, Platform Design, Two-sided Market, Learning Models, Structural Modeling