A&F Co. uses AI to predict how much of each product shoppers will want at each store location. Instead of relying only on past sales, the AI also factors in weather forecasts, social media engagement, and shoppers' browsing behaviour on the company's websites to make more accurate predictions.

A&F Co.'s CDTO described this practice at NRF 2024, explaining that the AI demand forecasting system combines transaction history with external signals including weather data, social media trends, and clickstream data (records of how users browse the website) to forecast demand at the zip-code level. More accurate forecasts mean fewer markdowns on unsold stock and fewer instances of popular items being out of stock. The models run on A&F Co.'s Azure Databricks machine learning platform and apply to all 700-plus stores, the majority of which operate under the Hollister brand.