We collaborate with a large e-commerce platform, and construct a new dataset of consideration sets by matching consumer search and browsing behavior with seller product and operations information. We also conduct a year-long experiment among new sellers on the platform. The treatment group receives access to a free online entrepreneur training program customized by an AI algorithm.  We first document a growth barrier (``GB'') in the competition for consumer attention: new sellers must achieve 10% higher conversion rates to obtain the same consumer exposure as incumbents.  The training program lowers this barrier by enhancing sellers' marketing and customer service efforts, hence boosting their revenue. We then estimate an empirical model to capture heterogeneity in consumer demand. This quantifies the effect of the GB: conditional on making a sale, new sellers have 23% higher quality than incumbents. By lowering this barrier, the training program benefits buyers. Although only 0.25% of all sellers on the platform are trained, removing the program would have reduced consumer surplus by 0.07%.