AI Training for Online Entrepreneurs: An Experiment with Two Million New Sellers on an E-commerce Platform

We conduct a year-long experiment among new sellers on a large e-commerce platform. The treatment group receives access to a free and customized entrepreneur training program, in which an AI algorithm assigns training materials to sellers based on their real-time operating statistics. With a 24% take-up rate, new sellers that are eligible for training see 1.7% higher revenue on average (6.6% ATT), driven largely by higher traffic and enhanced automation. To dissect the economic mechanisms and calculate the welfare impact, we develop a model of platform ranking and consumer choice. We then use search records to construct a panel dataset of consideration sets. Using exhaustive fixed effects, we document an entry barrier primarily driven by search frictions and not by information asymmetry. However, training does not affect this barrier—it works purely because automation adds value. Lastly, model estimation shows that while treated new sellers only account for 0.25% of products in consideration sets, removing the program would reduce consumer surplus by 0.07%.