Fanatics ran an OpenAI-powered "Shop the Look" feature across roughly 30,000 Major League Baseball products, and the test produced a 10% increase in engagement. The trial, which the company has now disclosed, marks one of the more concrete deployments of generative AI in sports merchandising.
The tool lets shoppers interact with product imagery in a way that surfaces related items, though Fanatics hasn't detailed the exact mechanics. What's clear is the scale: 30,000 MLB items is a meaningful slice of its catalog, not a token pilot.
What 'Shop the Look' actually did
Fanatics hasn't published a technical breakdown, but the concept is straightforward. A customer engages with a product visual — likely a jersey, cap, or other licensed gear — and the OpenAI-backed system identifies and presents matching or complementary items. The 10% engagement lift suggests shoppers spent more time interacting with the tool than they would have with a standard product page.
Engagement is a squishy metric. It can mean clicks, time on page, or add-to-cart actions. Fanatics didn't specify which definition it used, and that matters. A 10% bump in clicks is nice; a 10% bump in purchases is a different story. The company also didn't say whether the lift was measured against a control group or simply compared to historical baselines.
Why 30,000 products is the real story
Plenty of retailers have toyed with AI-powered recommendations. Most keep the scope small — a few hundred SKUs, a single category, a limited user pool. Fanatics went wider. Thirty thousand MLB products covers the bulk of its baseball assortment, from replica jerseys to novelty items.
That scale introduces complexity. The model has to recognize subtle differences between, say, a home jersey and an away jersey, or a throwback cap and a current one. It also has to avoid recommending items that are out of stock or not licensed for certain markets. Fanatics hasn't said how it handled those edge cases.
The company also hasn't disclosed whether the test ran across all MLB teams or a subset, or how long it lasted. Those details would help gauge whether the 10% figure is robust or a short-term novelty effect.
The OpenAI connection
Fanatics is using OpenAI's technology, though the specific model or API isn't named. The partnership is notable because OpenAI has been pushing into retail use cases, and sports merchandise is a high-volume, visually driven category that suits image recognition and natural-language search.
Fanatics declined to make an executive available for comment, and OpenAI didn't respond to a request for details. That leaves key questions unanswered: Is this a paid enterprise agreement? Will the tool expand to NFL, NBA, or NHL products? And does Fanatics plan to roll it out to its broader e-commerce platform, which handles licensed gear for multiple leagues?
What happens next
Fanatics hasn't set a public timeline for expanding Shop the Look beyond the MLB test. The company also hasn't said whether it will share more granular performance data — conversion rates, average order value, return rates — that would show whether the engagement lift translated into revenue.
For now, the 30,000-product trial stands as a proof point: generative AI can work at catalog scale in sports retail. Whether it becomes a permanent fixture or stays a limited experiment depends on numbers Fanatics hasn't released. The next signal will come if the tool appears on other league stores or if Fanatics updates investors on AI-driven sales. Until then, the 10% engagement figure is the only hard data point on the table.
