Conversion grew 1.5x
Major internet stores offer personalized recommendations to their customers. To boost personalization accuracy for a major marketplace, we built recommendation models that account for the weather. Conversion to purchasing recommended goods jumped by a factor of 1.5. For example, customers who were shown swimsuits on sunny days were more likely to make a purchase than those who saw recommendations that didn't consider the weather. Incremental GMV also grew by 0.56%.