Fanatics uses OpenAI to assemble coordinated, in-stock game-day outfits

Shop the Look combines enriched product data, inventory and merchandising rules, with a reported engagement gain in a bounded MLB catalogue test.

Fanatics is deploying an OpenAI-powered version of Shop the Look to recommend coordinated sports merchandise on product pages. OpenAI's October 26 account says the experience is live across many Fanatics-network sites, including the official NFL, NBA, MLB and NHL online stores.

The recommendations combine enriched product information, team identity and available inventory. Fanatics' own merchandising rules help keep selections compatible, and the looks are refreshed daily as the catalogue changes. A shopper viewing a team jersey can therefore see complementary items rather than only a list of similar jerseys.

The case study describes using Codex to help engineers build tools for enriching data, finding compatible products and evaluating recommendations. An initial fifty-product test took less than a day, followed by iteration with Fanatics reviewers. The development tool's contribution is distinct from the models assembling customer-facing recommendations.

OpenAI reports ten percent higher engagement than the previous experience after testing across roughly thirty thousand MLB products. That is a vendor-reported measure from the described test, not evidence of a ten percent increase in sales or profit.

The interesting retail integration is the connection between a generated suggestion and a real, currently available item. Team compatibility, style and inventory constrain the output in ways that a free-form outfit description does not. Daily refreshes then keep the merchandising process connected to the catalogue as stock changes, making product data and business rules central to the service.

Sources

- [How Fanatics and OpenAI streamline the fan experience](https://openai.com/index/fanatics-streamline-fan-experience/) — OpenAI