Google Merchant Center AI Performance Insights Now Generally Available

Google Moves AI Shopping Reporting Into General Availability

Google first previewed AI Performance Insights on May 27, 2026, describing a report designed to show how products and brands were discovered through conversational shopping experiences. At that stage, Google said the feature would roll out to the United States, Canada, Australia, India, and New Zealand over the following months.

By early September, the report had expanded with additional sections covering search intent, terms, and attributes. On September 16, Google formally announced that AI Performance Insights was generally available to businesses across the five supported countries.

The report is available in Merchant Center under Analytics > Products > AI performance. Google’s current documentation defines its scope as conversational queries with shopping intent on AI Mode and AI Overviews.

The traffic filter is currently limited to organic AI traffic such as free listings. Paid Ads traffic is excluded.

This makes the Merchant Center report a different measurement layer from the broader Search Console AI Performance reporting. Search Console measures website visibility across supported generative AI Search features, while Merchant Center focuses specifically on brands, products, shopping stages, and product-data opportunities.

The Report Measures More Than AI Visibility

The main change for retailers is not simply knowing that a product appeared in an AI experience. Google now provides several ways to understand the context around that visibility.

Share of Voice Comes With an Important Competitor Limitation

Share of voice is likely to attract the most attention because it gives retailers a comparative AI visibility metric. But the number needs context.

Google calculates a merchant’s share using its AI impressions relative to the merchant and competitor impressions for related queries. The competitor set is defined through Merchant Center data. Retailers cannot manually change the competitors used in this report.

That has a practical consequence. Google says a Merchant Center account with insufficient competitor data may show a share of voice of 100%. In that situation, 100% does not mean the brand has captured the entire AI shopping market. It can mean there was no usable competitor set for the comparison.

The report also distinguishes between a zero and a dash. A share of voice value of zero can mean there were insufficient impressions, while a dash indicates no impression data.

I would therefore check the competitor context before treating the headline percentage as a performance result. A precise number is still only as useful as the comparison behind it.

Shopping Stages Give the Report More Operational Value

Google classifies conversational shopping queries into three stages:

Merchant Center provides share of voice across these stages and also identifies search types within them. That is more useful than treating all AI shopping visibility as one number because the question being asked changes across a shopping journey.

What the report does not establish is why visibility is stronger at one stage than another. A lower share of voice in the ready-to-buy stage, for example, can identify an area worth investigating. It does not prove that pricing, structured data, reviews, page quality, or any single factor caused the difference.

Frequency Is Consumer Demand, Not Crawl Frequency

This distinction is especially important because the metric name is easy to misread.

In AI Performance Insights, Frequency represents how popular particular search types, terms, intents, or attributes have been. Google describes it as a signal of overall consumer demand that can help merchants decide what to prioritise.

It has nothing to do with how often Google crawls a product page.

Products showing is a separate metric. It counts how many of the merchant’s products appear for top terms, popular attributes, and search intents.

The useful comparison is therefore demand against product coverage. A high-frequency attribute with very few relevant products showing may justify a closer product-data review. A low-frequency term with extensive product coverage may deserve less attention.

Neither combination proves that a feed change will improve rankings or sales. It simply helps decide where to investigate first.

Top Terms and Attributes Create Product Data Opportunities, Not Keyword Rules

Google explicitly recommends using the report to improve product data. For Top terms, retailers can identify relevant language that shoppers are using and consider whether those terms are accurately represented in product titles or descriptions.

Popular attributes can reveal a different type of gap. If shoppers frequently look for specifications such as size, colour, or material and those fields are genuinely applicable but missing from the Merchant Center data, completing them gives Google a more accurate product record.

The important condition is relevance.

An attribute should not be added because it is popular if it does not describe the product. Nor does a high-frequency term justify inserting the phrase mechanically into every title. The report is useful for identifying missing information; it is not an instruction to rewrite a catalogue around whatever phrase appears most often.

For a large ecommerce site, I would start with one product category and the highest-confidence gaps rather than changing an entire feed at once. That keeps the effect of the update easier to review and reduces the risk of introducing inaccurate product data.

AI Performance Insights Does Not Measure Sales Impact

The new reporting improves AI shopping visibility measurement, but several boundaries remain.

• Share of voice is not conversion share. A higher percentage means stronger visibility relative to the defined competitor set, not more purchases.
• Frequency is not keyword volume. Google describes it as relative popularity across the report’s search types, terms, intents, and attributes.
• Products showing does not reveal a ranking formula. It records product coverage for the relevant insight, not why Google selected each product.
• Paid Ads traffic is excluded. The current report is limited to organic AI traffic such as free listings.
• Historical data has a lag. Google says the report updates daily but can run several days behind.

That means the report should be combined with Merchant Center performance data, website analytics, and commercial results before a retailer concludes that an AI visibility change helped or hurt the business.

The Current Report Scope Is Narrower Than Google’s May Preview

There is also a documentation detail worth watching.

When Google previewed AI Performance Insights on May 27, it said the insights were designed to show product discovery across AI Mode, AI Overviews in Search, and the Gemini app.

Google’s current AI Performance Insights Help page is more specific: it describes the live report as measuring conversational shopping queries on AI Mode and AI Overviews.

Because the current product documentation is the stronger source for the live report, retailers should not assume that every Gemini app product appearance is included in the present report unless Google explicitly confirms that scope again.

This is exactly the sort of detail worth checking when Google expands AI commerce reporting. Product previews can describe a broader direction than the final reporting interface available on a particular date.

Conversational Attributes Are a Separate Merchant Center Tool

AI Performance Insights also should not be confused with Merchant Center’s conversational attributes.

The report is an analytics layer. It shows patterns in AI shopping visibility and highlights terms, intents, and product-data gaps.

Conversational attributes are product-data inputs. Google currently supports optional fields including questions and answers, document links, related products, item group titles, variant options, and popularity rank. These fields are designed to give Google’s AI systems additional product context.

Using one does not automatically validate the other. A retailer can use the report to identify a data gap, then decide whether the correct response is to improve a standard Merchant Center attribute, update the product page, or use an appropriate conversational attribute.

What Ecommerce Teams Should Do With the New Report

The report is most useful when it changes the order of analysis rather than creating another dashboard to monitor.

1. Select a product category. Google does not provide one AI Performance Insights report covering every category at once.
2. Record the existing share of voice and shopping-stage pattern. Establish a baseline before changing product data.
3. Review high-frequency terms, intents, and attributes. Separate genuine product-data gaps from language that is irrelevant to the catalogue.
4. Compare Frequency with Products showing. Prioritise areas where demand appears meaningful but relevant product coverage is weak.
5. Make controlled product-data changes. Correct missing or incomplete information rather than rewriting every listing for AI search.
6. Wait for the reporting lag. Google says historical data is updated daily with a delay of a few days.
7. Validate business outcomes elsewhere. Use Merchant Center performance reports, web analytics, and sales data to determine whether visibility changes are associated with useful traffic or transactions.

The temptation will be to treat AI share of voice as a new ranking KPI. I would resist that. Its stronger use is diagnostic: it shows where the merchant is visible, where competitor visibility differs, and which parts of the product data deserve the next review.

What to Watch After General Availability

The first question is geographic and language expansion. As of September 18, 2026, Google documents AI Performance Insights for English-language queries in Australia, Canada, India, New Zealand, and the United States.

The second is reporting scope. Google says it is working on additional AI Performance Insights features, but it has not published a firm schedule for new countries, languages, or metrics.

The September 16 announcement also placed the report alongside wider agentic commerce changes, including updates to Universal Commerce Protocol integration and conversational product data. Those systems are related to Google’s broader AI shopping direction, but they should not be folded into the AI Performance Insights metrics themselves.

For retailers, the immediate change is narrower and more useful: Merchant Center now provides an official way to compare AI shopping visibility, shopper intent, product terms, and product-data gaps. The sensible next step is to use those signals to prioritise better product information, then check whether the commercial results follow.