Fanatics' AI Shop the Look: what 10% engagement does and doesn't show

Fanatics said on 26 October 2026 that an OpenAI-powered version of its Shop the Look tool produced 10% higher engagement than its previous version, in a test across roughly 30,000 MLB products². AlixPartners' 2026 Consumer Sentiment Index, published on 20 October 2026, found that only 10% of U.S. consumers currently begin product discovery with AI⁶. The two figures measure different things, so neither justifies a large spend on its own. Both do, however, point to product data as a place to start, provided the work is tested before it is scaled.

What Fanatics actually reported

Fanatics launched an OpenAI-powered version of Shop the Look, a tool that recommends coordinated merchandise on product pages². When a fan views a team jersey, the tool suggests outfits using in-stock items that match the team and intended style². It is live on Fanatics.com and on the official NFL, NBA, MLB and NHL online stores^1,2.

The company's headline result is that, after testing across roughly 30,000 MLB products, the new experience achieved 10% higher engagement than the previous one². Coverage describes that figure differently. Analytics Insight attributes it to OpenAI³. Briefs reports it as views of suggested items rising 10%⁵. Retail Dive repeats the engagement wording¹.

The test's duration, engagement measure and control group were not disclosed⁴. Analytics Insight also notes that the companies have not disclosed a corresponding increase in revenue or conversion rates³.

The recommendations rely on specific inputs. OpenAI models draw on enriched product data, team and inventory information, Fanatics' merchandising rules keep selections "compatible and relevant", and the looks are refreshed daily as catalogues change².

What the AlixPartners figure measures

AlixPartners' 2026 Consumer Sentiment Index surveyed more than 9,000 U.S. consumers between 14 and 24 July 2026⁶. The report states that just 10% of consumers currently begin product discovery with AI, and that most are not ready to let AI agents make fashion purchases for them⁶.

Adoption varies by age and income. The report says one in 10 consumers begins discovery with AI, rising to 14% among Gen Z and millennials⁶. Among shoppers earning $150,000 or more, 68% consider innovative shopping technology important, against 53% of those earning less than $45,000⁶.

AlixPartners expects adoption to rise, particularly for commoditised goods such as replenishment basics⁷. Retail Dive gives the example of a query like "black leggings, TTS, under $30" being passed to AI agents more readily⁷.

What the sources say about product data

AlixPartners argues that product data matters more as AI adoption grows. In the report's words, "The machine doesn't read your beautiful imagery or your persuasive copy — it reads structured data: attributes, availability, price"⁷. The same passage warns that if product data is "thin or messy", an agent "either can't surface you or describes you badly"⁷.

Fanatics' recommendations depend on the same kinds of data: product data, team and inventory information². Analytics Insight reports that Fanatics sells more than 1.5 million distinct products across its online stores³. Briefs reports that the company adds upwards of a thousand new products every week⁵.

What this means for sellers

This section is our analysis. It is not sourced reporting.

Two numbers, two questions

The table compares the headline figures.

The Fanatics result is about what happens after a shopper has reached a product page. The AlixPartners result is about where shoppers begin. A lift on a recommendation module does not show that AI is bringing in new shoppers, and the AlixPartners figure says nothing about whether AI-led shoppers buy more.

Three scenarios

These scenarios are our assumptions for planning purposes, not forecasts.

• Scenario A: AI-started discovery stays near 10%. Metadata work mainly helps on-site and marketplace search¹. The return is likely to be modest.
• Scenario B: AlixPartners' expected uptick arrives in commoditised categories. Agents would read attributes, availability and price before recommending products⁷. Structured data would then matter most in replenishment-style categories. This is the scenario that would justify larger investment, and it is the least certain.
• Scenario C: The Fanatics result does not carry over. Fanatics combines enriched data, team and inventory information with merchandising rules and daily refreshes². A seller without equivalent data and refresh processes should not expect the same lift.

A practical test for the next quarter

We recommend the following approach.

1. Cap the effort. Give metadata work a fixed, small share of catalogue time, rather than an open-ended project.
2. Start with the fields the sources name. Price, availability and the product attributes that shoppers filter on are the starting point. AlixPartners names attributes, availability and price⁷.
3. Run a holdout test in one category. Keep a comparable set of products out of the metadata update and compare results over a fixed period.
4. Define one metric before you start. For example, click-through or add-to-cart on recommended items. Record it the same way for both groups.
5. Judge the result on sales, not engagement alone. Fanatics has not published revenue or conversion data³, so you should not assume an engagement gain translates into orders.

For a related development on AI shopping agents, see Walmart opens shopping to AI agents; Amazon blocks Meta's agent.

What to watch next

• Whether Fanatics or OpenAI publishes the test duration, engagement definition, control group and any revenue or conversion results.
• The full AlixPartners 2026 Consumer Sentiment Index report, including how the 10% figure is defined and broken down by category⁶.
• The next wave of AI-discovery adoption, especially among younger shoppers and in commoditised goods^6,7.
• Any platform guidance that turns product-data quality into a formal requirement for AI shopping. None is in the sources reviewed for this article.