Merchant Center Adds AI Search Intent, Search Terms and Attributes Reporting

Google has expanded the AI Performance Insights report in Merchant Center with three new reporting sections, giving retailers their most detailed view yet of how products are surfacing inside AI-driven shopping experiences.

The additions were spotted by Brodie Clark on 9 September 2026, who posted side-by-side comparisons of the updated report against its previous structure. Google's Merchant Center help documentation has been updated to match.

What's new

AI search intent. Shows how your products align with the intent behind customer searches in AI surfaces — moving the reporting away from raw keyword matching toward the underlying need the shopper is expressing.

AI search terms. Surfaces the terms people are actually using in AI search, and recommends adding popular ones to your product descriptions to improve visibility.

AI attributes. Identifies product attributes you're missing from your feed and suggests adding them to increase the chance of your products being surfaced in AI search results.

The AI Performance Insights report itself launched in July 2026 and expanded to more regions in early September. These three sections are the first substantial addition to what it reports on.

Why this is a bigger deal than it looks

For most of the last two years, the honest answer to "how do my products perform in AI search?" has been that nobody could tell you. AI Overviews and AI Mode surfaced products, shoppers clicked through, and the data landed in analytics as a blur of referral traffic with no keyword-level visibility. Retailers were optimising into a black box.

This is Google handing over three specific things that were previously unavailable:

1. Query-level visibility for AI surfaces — the AI search terms section is, in effect, a Search Console performance report for AI shopping. That's the piece the industry has been asking for.
2. A diagnostic, not just a metric — the AI attributes section doesn't just tell you how you're doing, it tells you what's missing from the feed and implies the fix. That is unusually actionable for a Google report.
3. An intent layer — grouping by intent rather than by string is a hint about how the underlying systems evaluate product relevance, and it's worth reading as guidance about what Google's models are matching on.

How to use it

Start with AI attributes, because it's the cheapest win. Missing attributes are a feed problem, and feed problems have deterministic fixes. If Google is explicitly telling you which attributes are absent from products that would otherwise surface, that's a prioritised backlog handed to you. Work through it before touching anything else.

Mine AI search terms for description rewrites — but don't keyword-stuff. The recommendation to add popular search terms to product descriptions is straightforward, and the risk is equally straightforward: descriptions that read as term lists will underperform with both shoppers and models. Fold the language in naturally, and prioritise terms that describe genuine product characteristics you can support.

Use AI search intent to check your assortment logic. If the intents surfacing against your products don't match how you've categorised and titled them, that's a merchandising signal as much as an SEO one. It may indicate you're being matched to searches you can't satisfy well, or missing ones you could.

Baseline now. The report is new and its shape is still moving. Export what's there today so you have a starting point to measure against, rather than discovering in three months that the earliest data you hold is already the changed version.

Caveats worth holding

This reporting is new, it has changed twice in two months, and regional availability has been rolling. Treat month-over-month comparisons cautiously until the report has been stable for a full reporting cycle — some of what looks like performance change will be reporting change.

It's also worth noting the recommendations are Google's, generated by Google's systems, about how to perform better in Google's surfaces. That's useful, and it's also not neutral advice. Sanity-check suggested description changes against what actually converts for you.

Sources

• Google Merchant Center Help: AI Performance Insights
• Report changes documented by Brodie Clark on X