Google Merchant Center now has an AI performance insights report, and it is the first Google product to expose query-level data for AI shopping. It shows how your products surface across AI Overviews, AI Mode, and the Gemini app, broken into four modules: your share of voice, a Discovery-Evaluation-Purchase funnel, the shopping terms buyers use, and the product attributes they ask for. As of mid-2026 it is a limited US pilot. It covers organic "free listing" visibility only, not ads.
That "query-level" part is the news. Google's Search Console Generative AI performance report, launched 3 June 2026, is impressions-only: no queries, no share of voice. Merchant Center's report is the first native Google surface to tell shopping brands which terms they show up for inside AI answers, and how they stack up against similar merchants.
What is in the AI performance insights report?
Per Google's 2026 announcement and its Merchant Center Help documentation, the report has four modules. Each answers a different question about how your products appear in AI shopping.
| Module | What it shows | The question it answers |
|---|---|---|
| Share of voice | Your AI impressions divided by total impressions, benchmarked against similar merchants | "How visible am I versus my category?" |
| Shopping funnel performance | Visibility split across Discovery, Evaluation, and Purchase stages | "Where in the journey do I show up, and where do I drop out?" |
| Product term insights | Popular conversational shopping terms, with your share of voice for each | "Which shopping questions surface me, and which don't?" |
| Product attribute insights | The specifications buyers ask for (size, colour, material), with feed-completeness signals | "Which attributes should my feed cover?" |
The report scope is narrow and worth stating up front: Google describes it as covering conversational shopping and brand queries with organic AI visibility, the equivalent of free listings, excluding paid ads. It is product-and-shopping data, not a report on your whole site.
How does the Discovery-Evaluation-Purchase funnel work?
Google splits your AI shopping impressions into three stages, so you can see where in the buying journey a product enters the conversation and where it falls away.
| Stage | What buyers are doing | Example query types Google groups here |
|---|---|---|
| Discovery | Broad, early exploration | Category, features, or style ("waterproof hiking jackets") |
| Evaluation | Comparing and checking specs | Product specifications, user reviews, comparisons |
| Purchase | Ready-to-buy signals | Pricing, availability, local stock |
The funnel is diagnostic. A product with healthy Discovery share but near-zero Evaluation share is being found but not shortlisted, usually because the specs, reviews, or comparison facts an AI needs at the Evaluation stage are missing or not extractable. That maps directly to the attributes buyers ask AI Mode about most, which Google published separately: price, availability, colour, size, and material.
Why does a native share-of-voice number matter?
Because until now, share of voice in AI answers was something you inferred from third-party trackers, never a figure Google handed you. A first-party, benchmarked number, "your impressions as a percentage of total, versus similar merchants", is a different kind of signal: it is corroborated by the platform doing the surfacing.
Merchant Center's AI report is the first time Google tells a shopping brand which terms surface it inside AI answers and how it ranks against its category, not just that it appeared.
Two cautions keep the read honest. First, it is shopping-scoped: it measures product visibility across Google's AI surfaces, not citations of your editorial or brand content, and not any non-Google engine. Second, like the Search Console rollout, practitioners with early access report it is light on direct actionability so far, useful for spotting where you are absent, less so for telling you the exact fix. Treat it as a diagnostic layer, not a to-do list.
How do I access it, and who has it?
The path, per Google's documentation: log into Google Merchant Center, open the Analytics tab, select Products, then select the AI performance tab.
- Availability (as of mid-2026): a limited US pilot with a subset of Merchant Center accounts. Google has said it will expand to Australia, Canada, India, and New Zealand in the coming months.
- Why it may be empty for you: most accounts are not in the pilot yet, and the data is shopping-and-product only, so a site with no product feed will not populate it.
- What to do now if you have it: benchmark your share of voice by funnel stage, note the Product Term Insights where your share is lowest, and check whether the Product Attribute Insights flag gaps in your product catalogue.
For the broader context on how these shopping answers are assembled, see how Google AI Mode works and Google's own AI Mode usage data. If you sell into India, note it is on the announced expansion list, which pairs with the AI shopping in India picture.
How should this fit your measurement stack?
As one first-party signal among several, not the whole picture. Merchant Center tells you about Google's AI shopping surfaces; it says nothing about ChatGPT, Perplexity, or Claude, and nothing about whether you were cited as a source or merely recommended. Fold it into the five-metric reporting stack alongside Search Console's AI impressions, Bing's citation data, and cross-engine tracking so no single vendor defines your view.
That cross-engine, answer-level scoreboard, presence, citations, and sentiment tracked daily across every AI surface rather than one platform's shopping tab, is exactly what Buffy Intel is built to give you. Questions: [email protected].