Fanatics Puts OpenAI Recommendations on Jersey Pages, Lifting Engagement 10%
"Shop the Look" picks items to pair with a jersey based on team and inventory data. In a test across roughly 30,000 MLB products, OpenAI says engagement ran 10% higher than with the previous version of the feature.
Sports merchandise retailer Fanatics has launched “Shop the Look,” a shopping feature built on OpenAI models. OpenAI published the case study on October 26. Timed to the fall sports calendar, the feature surfaces items worth wearing with a jersey directly on that jersey’s product page, so fans browsing jerseys can scan related products without leaving the screen.
What to buy with the jersey
The recommendations draw on product descriptions, team information and inventory data. Combinations are assembled according to merchandising rules set by Fanatics, and only items that are actually in stock are eligible to appear. In other words, the system weighs both whether a product suits the jersey a fan is looking at and whether it can be purchased right now.
The recommended combinations are regenerated daily. When the product lineup or inventory changes, the companion items are reselected, meaning a combination built once does not stay on the page indefinitely. Fans putting together an outfit around a jersey get their options on the product page itself.
The feature is available across a number of sites, including Fanatics.com, as well as the official online stores for the NFL, NBA, MLB and NHL. Shoppers browsing jerseys on those stores encounter the AI recommendations as they look for pieces to wear with them.
How much better than the old recommendations
OpenAI says that in a test covering roughly 30,000 MLB products, engagement with the new feature ran 10% higher than with the previous one. Fanatics had already offered a feature under the same “Shop the Look” name.
Testing the initial recommendations also moved quickly, according to OpenAI: validating recommendations for the first 50 products took less than 24 hours.
How the feature was built
The development team used Codex to build the tooling behind the recommendations themselves, including tools to enrich product data, find items that go well together, and evaluate recommendation results or generate combinations. Fanatics also uses ChatGPT Work and Codex across several internal business units. OpenAI products, in short, sit on both sides of the equation: the recommendations customers see and the internal work that prepares them.
What this release changes is the step between choosing a jersey and finding what to wear with it. Team- and inventory-aware combinations are now visible on the same product page. The open question is whether the lift in engagement also translates into actual purchases.