Google is giving ecommerce teams something they have rarely had for generative search: direct, product-level clues about what shoppers are asking AI systems and which information may be missing when their products appear.
Google Merchant Center's AI performance reporting has expanded with three practical views described in the interface as AI Search intent, AI Search terms and AI Attributes. The update was reported by Search Engine Roundtable on September 9 after Brodie Clark spotted the expanded report.
The significance goes beyond three new cards in Merchant Center. Google is effectively telling merchants that conversational shopping visibility can be analyzed and optimized through product data: understand the intent behind AI searches, identify the language shoppers use and fill structured attributes that those shoppers repeatedly ask about.
That does not make the recommendations a new set of confirmed ranking factors. Google does not say that inserting a suggested phrase guarantees an AI Mode citation or that completing one attribute directly raises an AI visibility score. But it does turn AI shopping optimization from a largely external guessing exercise into something merchants can partially observe through Google's own data.
Merchant Center is exposing the language of conversational shopping
Traditional ecommerce search optimization has revolved around relatively compact queries: product names, brands, categories and modifiers such as size, color or price. Generative interfaces allow shoppers to express much richer requirements in one request.
A user can ask for a lightweight waterproof shoe for a specific type of trip, a laptop suitable for a particular workflow or a piece of furniture that fits both a room dimension and an aesthetic constraint. The query is no longer just a keyword. It contains a bundle of attributes, use cases and decision criteria.
Google's official AI performance insights documentation describes the report as a way to understand this shift from keyword searches to complex conversational shopping paths. It shows how brands appear for shopping-intent queries in AI Mode and AI Overviews and provides data designed to identify optimization opportunities.
The new reporting makes that concept operational. Instead of telling a merchant only that AI traffic exists, Merchant Center can reveal the kinds of intent, terminology and specifications associated with that visibility.
AI Search intent shows why shoppers are looking for a product
The first view, AI Search intent, helps merchants understand how their products match customer AI searches. Google's current help documentation calls the corresponding opportunity view “Top search intents.”
This moves analysis beyond a list of words. Search intent captures the context behind the request: what the shopper is trying to accomplish, compare or decide.
Merchant Center also organizes AI shopping performance across three stages: discovery, evaluation and ready to buy. Discovery represents early exploration, evaluation covers comparisons and specifications, and ready-to-buy queries sit closer to a transaction.
Within those stages, merchants can see search types such as category exploration, product-specification research or review seeking. That gives product teams a way to identify not only whether they are visible, but where in the conversational shopping journey that visibility occurs.
Share of voice adds a competitive dimension
The AI performance report includes share of voice, allowing a merchant to compare its visibility with a Merchant Center-defined group of competitors. Google combines that with frequency data representing how popular particular search types, terms, intents or attributes are.
This creates a useful prioritization model. A high-frequency intent with low share of voice suggests a potentially valuable gap. A low-frequency intent where the brand already dominates may deserve less immediate attention.
Google also reports “products showing,” the number of a merchant's products appearing for particular top terms, attributes or intents. That can help distinguish a brand-wide visibility problem from a catalog-coverage problem.
There are limitations. Google chooses the competitor set and merchants cannot manually replace it. The documentation also notes that insufficient competitor data can produce a 100% share-of-voice value, so the number needs context rather than being treated as an absolute market-share measurement.
AI Search terms recommends language merchants can add to product copy
The second new view is AI Search terms. Search Engine Roundtable describes it as a section that recommends popular terms merchants can include in product descriptions to improve visibility for products appearing in AI searches.
Google's official documentation labels the underlying feature “Top terms.” It highlights terms within a product category that shoppers frequently prioritize in conversational queries, using examples such as “maximum cushioning” and “arch support.”
Those examples show why this is different from a classic keyword report. A shopper may not search only for “running shoes.” The deciding language can describe the physical benefit, problem or feature they care about. Merchant Center can surface those recurring concepts directly from conversational demand.
Google explicitly recommends incorporating relevant top terms into product titles and descriptions. That is unusually actionable guidance for AI Search optimization because it connects observed AI-shopping language to the product data merchants control.
This is not permission to stuff AI keywords into every description
The obvious risk is that merchants treat AI Search terms like a 2008 keyword-density tool. Google's recommendation is more constrained: add relevant terms where they accurately describe the product.
A suggested phrase such as “arch support” should not be added to every shoe simply because it is popular. If the product does not actually provide that feature, the resulting feed becomes less accurate and potentially misleading.
Google's broader Merchant Center product data specification continues to require accurate product descriptions that correspond with the landing page. Promotional text, irrelevant information and artificial keyword lists are not what the feed is designed for.
The optimization opportunity is semantic completeness, not repetition. If customers repeatedly describe a genuine product property in language absent from the feed and landing page, adding that language can make the product easier for both users and AI systems to understand.
AI Attributes identifies structured information the feed is missing
The third view, AI Attributes, may be the most strategically important because it pushes optimization beyond prose. Google's documentation calls the underlying feature “Popular attributes” and describes it as structured technical specifications shoppers frequently request, such as size, color or material, that may be missing from a merchant's product data.
This means Merchant Center can identify a demand signal and compare it with feed completeness. If shoppers frequently ask about a specification and a merchant has not supplied that attribute for eligible products, the report can expose the gap.
Google recommends populating those missing attributes, starting with the most popular opportunities. The logic is straightforward: a generative shopping system can make a better product match when the catalog contains explicit machine-readable evidence that the product satisfies the shopper's constraint.
For ecommerce teams, this reinforces a principle that is becoming central to AI discovery: product data quality is not merely a feed-management task. It is part of search visibility.
AI shopping optimization is becoming more structured, not less
Generative search can create the impression that ecommerce optimization will become entirely about natural-language content. The Merchant Center update points in the opposite direction.
Conversational interfaces may use natural language on the front end, but they still need structured facts underneath. A shopper can phrase “I need something waterproof in black under a particular size” conversationally, yet the system benefits enormously from having waterproofing, color and size represented as reliable attributes rather than trying to infer everything from marketing copy.
That makes structured feeds more important as queries become more complex. Every additional constraint in a conversational request creates another property the system may need to verify against a product.
The strongest product representation therefore combines descriptive language with structured attributes. Titles and descriptions explain the item in human terms; attributes provide explicit fields that systems can filter, compare and validate.
Google has also introduced conversational product attributes
The direction becomes clearer when the AI performance report is considered alongside another Merchant Center development. Google now documents conversational attributes designed to help AI systems and conversational agents understand product nuances.
These optional fields complement the standard product specification rather than replacing it. Google says they can help customers discover information about products across AI-driven surfaces such as AI Mode while also improving traditional search experiences.
One example is a question-and-answer attribute, which allows merchants to provide structured responses to common product questions. This effectively turns parts of the product feed into machine-readable sales knowledge rather than a simple inventory record.
Together with AI Search terms and AI Attributes, the direction is unmistakable: Merchant Center is evolving from a feed submission interface into a feedback loop between conversational demand and catalog information.
Google is telling merchants what AI shoppers care about
Historically, merchants had several ways to infer customer language. Search query reports revealed paid-search terms. Search Console exposed organic queries. On-site search data showed what visitors looked for after arriving. Reviews and customer support revealed recurring questions.
AI shopping creates a new source of demand data. The queries can be longer and richer, exposing combinations of requirements that short keyword searches hide.
A merchant might learn that shoppers repeatedly care about cushioning for a particular activity, compatibility with a device, suitability for a climate or material characteristics associated with a use case. Those patterns can influence more than Merchant Center descriptions.
They can inform product-page copy, FAQs, buying guides, merchandising, photography and even product development. The AI report is therefore potentially useful as consumer-language research, not only as a visibility dashboard.
The report focuses on organic AI shopping traffic
One important boundary is that Google's current help documentation says the report is limited to organic AI traffic, such as free listings. Paid Ads traffic is not included.
This makes the report particularly interesting for SEO and GEO teams because it provides visibility data outside the traditional advertising interface. It can reveal where product information is participating in AI-driven discovery without a paid placement being responsible for the exposure.
At the same time, merchants should not combine the report mechanically with paid-shopping metrics. An AI share-of-voice number and a Shopping Ads impression-share metric describe different surfaces and competitive environments.
The report is best treated as a new measurement layer alongside organic Search, paid shopping and onsite conversion data.
Current availability is still limited
Google announced AI performance insights in May 2026 and said the capability would roll out initially across the United States, Canada, Australia, India and New Zealand. Current help documentation says it is available for English-language queries for eligible Merchant Center accounts in those five markets.
That geographic and language limitation matters. Merchants operating in Europe or non-English markets should not assume the same report is already available or that observations from English conversational queries transfer directly to their audiences.
The report also works at the product-category level rather than providing one universal all-category view. Merchants need to filter and interpret opportunities in the context of a specific catalog segment.
Google says historical data is available and updated daily with a lag of several days, allowing teams to monitor how demand and visibility change rather than relying on a one-time snapshot.
Google's May announcement included Gemini, but the current help page is narrower
When Google first previewed AI performance insights in May, it described the feature as covering AI-powered shopping experiences including AI Mode, AI Overviews and the Gemini app. The current detailed Merchant Center help page specifically describes conversational shopping queries in AI Mode and AI Overviews.
That difference is worth preserving rather than assuming every current metric combines all three surfaces. Google can change scope as products roll out, and Merchant Center reporting should be interpreted according to the documentation available for the account and market being analyzed.
For GEO measurement, product scope matters as much as query scope. Visibility in AI Mode, AI Overviews and Gemini should not automatically be treated as one identical distribution channel unless Google explicitly reports them together.
The recommendations are not confirmed AI ranking factors
The most tempting interpretation of the new report would be that Google has finally published a checklist for ranking products in AI Search: add the recommended terms, complete the recommended attributes and visibility goes up.
The evidence does not support that certainty. Google describes these views as optimization opportunities and recommends implementing relevant terms and missing attributes. It does not publish a formula showing how each field affects AI retrieval or ranking.
There are also obvious confounding factors. A merchant that completes an attribute may simultaneously improve feed quality, traditional shopping eligibility and relevance for multiple query types. A better description may help systems understand the product without functioning as a direct weighted “AI ranking signal.”
The safest interpretation is that Google is exposing information it believes can put products in a better position to succeed on AI surfaces. That is valuable guidance without turning the interface into an algorithm specification.
Product feed optimization is becoming GEO
GEO discussions have often focused on editorial content: how to make articles, brands and websites more likely to appear in generated answers. Ecommerce has a different substrate. The primary source of truth is frequently the product feed.
Merchant Center now makes that connection explicit. AI visibility can depend on whether the system understands the product's features, specifications and suitability for the intent expressed in a conversational query.
For retailers, that means GEO work may happen in a feed-management system rather than a CMS. A missing material attribute or vague product description can be an AI-discovery problem even when the product page is technically indexed and ranks for its brand name.
The optimization unit becomes the product entity and its data completeness, not only the webpage.
The best workflow starts with demand, then validates the catalog
A sensible Merchant Center workflow would begin with high-frequency intents and terms where the brand has weak visibility. The merchant can then inspect which products appear, identify whether the relevant terminology accurately describes them and review the missing attributes Google highlights.
The next step should be validation against the actual product. Terms should be added only when true and useful. Attributes should be completed from reliable product specifications rather than inferred because they appear popular.
The corresponding landing page should also communicate the same facts. A feed claiming a material, feature or use case that the product page never substantiates creates inconsistency at precisely the moment AI shopping systems need trustworthy data.
Finally, the team can monitor whether products showing, share of voice and other AI performance indicators change over time. That turns the report into an iterative optimization loop rather than a one-time list of recommendations.
Merchant Center is becoming an AI-search observability layer for commerce
The broader importance of this update is measurement. AI search has been difficult for merchants to observe because conversational queries do not map neatly onto classic keyword reports and generated product recommendations can vary from one prompt to another.
Merchant Center is beginning to expose part of that hidden layer. Merchants can see what shoppers intend, which terms recur, which attributes matter, how many products show and how their visibility compares with competitors.
Google still controls the definitions, competitor groups and underlying aggregation, so the report is not a transparent view of the entire AI retrieval algorithm. But it is far more actionable than guessing from isolated AI Mode screenshots.
For ecommerce SEO, the direction is significant. Google is effectively saying that conversational shopping visibility should be optimized through better product understanding: use the language customers actually use, provide the specifications they care about and keep structured product data complete.
AI Search is turning product descriptions into retrieval infrastructure
The new Merchant Center insights connect three stages that used to be managed separately. Customer language appears in conversational searches. Google identifies recurring terms, intents and attributes. Merchants can feed that information back into titles, descriptions and structured product data.
That feedback loop is likely to become increasingly important as shopping journeys move from short keywords to complex requests. The more constraints a shopper can express in one sentence, the more complete a product's machine-readable representation needs to be.
The immediate opportunity is practical: review AI Search intent, inspect popular terms and fill genuine attribute gaps. The larger lesson is strategic. Product data is no longer just inventory infrastructure for Shopping feeds. It is becoming part of the retrieval infrastructure for AI-powered commerce.
Google Merchant Center is now showing merchants where that infrastructure is incomplete. The retailers that treat those insights as a product-data quality problem—not an invitation to stuff descriptions with AI keywords—will be in the strongest position to learn what conversational shopping actually rewards.