Fanatics built Shop the Look with OpenAI to assemble complete game-day outfits around individual items a fan is viewing, and a trial covering about 30,000 MLB products produced a 10% lift in engagement S¹. That figure is a vendor's own number with no released code, no error bars, and no independent audit — and neither company has defined what "engagement" actually measures S¹.
My read: This is the kind of enterprise AI deployment I keep watching for, not because the technology is novel but because the headline metric is engagement, not revenue. Fanatics and OpenAI are telling us fans clicked more, not that they bought more. That gap between attention and conversion is where most AI commerce features stall. I'd want to see basket size and conversion data before calling this a win, and neither company has shared that.
The feature arrives ahead of the fall sports calendar P³. Fanatics, which operates across California, Colorado, Florida, India and the UK P⁵, ran the test on roughly 30,000 MLB products before launch S¹.
Shop the Look does something simple: when a fan views a jersey, the AI suggests the matching hat, hoodie or accessories to complete the outfit S¹.
The underlying OpenAI model is not named in the announcement.
OpenAI lists ChatGPT and Codex among the products involved P³, but the source does not specify which model drives the recommendations.
The 10% lift covers clicks, not checkout
The 10% engagement lift was measured across roughly 30,000 MLB products S¹. No prior period is given for comparison. No revenue, conversion or sales impact is attributed to the feature, and the only secondary source is a Google News RSS feed repeating the OpenAI headline verbatim S², which adds no independent corroboration.
This pattern, a vendor case study with a single positive metric and no methodology, is common in enterprise AI announcements.
For a merchandise manager at a mid-size sports retailer, Shop the Look means each product page now carries AI-generated complementary-item suggestions, turning a single jersey view into a multi-item pitch without manual merchandising work S¹. A retailer with 50,000 product listings across multiple leagues could test this pattern cheaply: group products by team, use a language model to suggest complementary items, and measure average order value against a control cohort over four weeks. The Fanatics result suggests the click-through is there. Whether it survives at checkout is the part nobody has disclosed.
The feature has not been confirmed beyond MLB S¹. No expansion timeline to other leagues or sports is mentioned in the announcement.
Fanatics has not announced a date for expanding Shop the Look beyond MLB, and neither company has committed to releasing conversion or revenue data.
Sources: S1 — How Fanatics and OpenAI streamline the fan experience · S2 — How Fanatics and OpenAI streamline the fan experience - OpenAI · P3 — How Fanatics and OpenAI streamline the fan experience | OpenAI · P4 — openai/gpt-oss · P5 — Fanatics · P6 — princeton-nlp/WebShop
Written from 6 sourced items, 5 of them primary.
