This paper studies the factors underlying the heterogeneity in inventory behavior and performance across retail stores. We use a dynamic model of multi-product retailers and local competition to estimate store productivity and consumers' perceived quality of the shopping experience, and we analyze their relationship with inventory behavior and product variety. Using novel and detailed data on Swedish stores and their products, we find that stores learn from demand to improve future productivity. Store productivity is the main primitive that increases inventory turnover and product variety, and this increase is larger for stores with already high inventory turnover. Stores in small markets with intense competition from rivals have higher inventory turnover. Consumers in large markets and markets with large investments in technology benefit from a broader product variety. Counterfactual experiments show that the increase in inventory turnover due to innovations in productivity is three times greater when uncertainty in demand is reduced by 30 percent. Our analysis highlights important trade-offs between productivity and demand that allow retailers to reach high levels of inventory turnover and offer a broad product variety to consumers.