Fanatics adds coordinated shopping suggestions built with OpenAI Codex

Fanatics wants shoppers to feel more confident buying team gear, and its new Shop the Look feature suggests matching items alongside a product. The early result is narrower than that ambition: Fanatics says views of recommended merchandise rose 10% after rollout. Built using OpenAI Codex, the feature matches in-stock products by team and style.

A catalog that keeps getting harder to browse

The feature addresses a problem created by Fanatics’ own assortment. The company offers more than 1.5 million unique products across its sports merchandise business, from jerseys and hats to socks and team-branded streetwear. Finding one item means navigating a catalog that stretches well beyond the most familiar game-day gear.

Justin Tsai, chief product officer for Fanatics Commerce, told CNBC that shopping sessions routinely averaged more than 30 minutes. Fanatics operates hundreds of sites and releases upwards of a thousand new products each week, he said. That steady flow adds more choices to an already lengthy search for something to buy.

Conventional outfit curation cannot always keep up. Photo shoots and styled combinations take time to arrange, Tsai said, making them a poor fit for some fast-moving releases, including championship shirts and hats. The challenge is not just selecting attractive accessories. Suggestions also have to respect the team loyalties that brought a customer to the site.

Nobody who is a Los Angeles Dodgers fan wants to be matched with a San Francisco Giants item … team and player alignment is extremely important in our world,

Justin Tsai, chief product officer for Fanatics Commerce, speaking to CNBC

Matching the jersey, not the rival

Shop the Look places coordinated recommendations on product pages. A shopper considering a jersey might see a compatible hat, socks or other accessories. The selections account for the product’s team and intended style, and the suggested merchandise must be in stock. The starting point is an item the shopper is already considering.

Fanatics used Codex to build the tool, which parses product and catalog data to find compatible merchandise and reevaluate recommendations. The disclosed workflow centers on catalog matching, rather than a conversation with a shopping chatbot.

Rosemary DeArago, OpenAI’s head of industry GTM, described retail personalization as an old problem. Signals such as purchase history, location and a logged-in account have given retailers ways to identify customers, she said, without always producing the right suggestions. Her argument for AI here is more precise matching of products within the existing shopping experience.

More browsing, with purchases still unmeasured

The reported 10% increase is a company-supplied measure of views of recommended items. It is not a disclosed increase in completed purchases or revenue. That distinction matters for a feature presented as a way to help shoppers choose: opening another product page shows engagement, but does not by itself demonstrate a better buying decision.

Tsai said seeing merchandise alongside other team gear gives customers more conviction about an item they are considering. DeArago also left room for curated looks and trying products on in person. Neither executive presented automated matching as a replacement for every part of shopping; the stated opportunity is to narrow a large selection and show compatible choices.