Google has made Merchant Center's AI Performance Insights available to merchants in the United States, Canada, Australia, India and New Zealand, giving brands a way to measure how their products appear in AI Mode and AI Overviews. More than adding a new report, the change begins to answer a question that accompanied the expansion of AI-powered search: how to measure commercial visibility when the experience is no longer organized primarily as a list of results?
Google's initial answer combines familiar digital marketing metrics with new indicators adapted to conversational queries. The main one is share of voice, calculated from the brand's or its products' impressions relative to the impressions obtained by a set of competitors defined by Merchant Center itself. The report also measures term and intent frequency, number of products shown and performance across the discovery, evaluation and purchase intent stages.
Position loses part of its meaning when the answer is built by AI
In traditional SEO, much of performance analysis was structured around four metrics: impressions, clicks, CTR and average position. Search Console itself defines position as the relative placement of a link in Google's results.
This model remains relevant, including because Google has not abandoned these metrics. Since August, Search Console has globally offered a specific report for generative AI features, including AI Overviews and AI Mode, in which sites can track impressions, pages shown, countries, devices and evolution over time.
For e-commerce, however, Merchant Center starts to answer a different question. Instead of simply reporting whether a page occupied a certain position, the report shows how much of the available exposure within a category or commercial intent is being captured by a given brand compared with its competitors.
It is a logic closer to the impression share already used in advertising. In Google Ads, this metric compares the impressions received with the estimated total impressions for which the ad was eligible. In AI Performance Insights, the concept is transferred to organic commercial recommendations produced in AI experiences.
This does not eliminate ranking, clicks or impressions. It creates an additional layer in which participation in the answer can be more informative than an isolated position.
The new commercial SEO begins to measure intent and product coverage
The report also signals a second change: queries are no longer analyzed only as keywords.
Merchant Center classifies interactions across discovery, evaluation and “ready to buy” and allows observing share of voice at each stage. It also identifies recurring terms, attributes sought by consumers and intents present in conversational queries.
In practice, a brand may discover that it has good presence when consumers are already close to purchase, but appears little during initial searches about a product's characteristics. Or it may identify a category with a high frequency of queries in which few items from its catalog are being shown.
The metric products showing adds another dimension: it is not enough to know whether the brand appeared; it becomes possible to observe how many products from the catalog are managing to participate in responses related to certain terms, attributes or intents.
This brings AI optimization closer to management of the catalog itself. Titles and pages remain relevant, but the quality and depth of the data sent to Merchant Center gain additional weight.
Feed goes from technical inventory to source of AI context
This change becomes more evident with the so-called conversational attributes. Google allows merchants to add information to the feed such as questions and answers about products, related documents, relationships between items, variants and popularity indicators. The documentation says that these fields help AI systems and conversational agents understand specific nuances of the products.
In a test with Lululemon, conversational attributes provided by the brand were incorporated into 50% of relevant product recommendations in AI Mode, according to Google. The result is specific to this test and does not establish an expected performance rate for other merchants, but it demonstrates that structured information provided directly by the merchant is already being used in the composition of recommendations.
The consequence is that attribute completeness may start to work as an operational indicator of visibility. If consumers ask for “maximum cushioning,” a certain material or a specific characteristic and the catalog does not adequately describe these elements, the brand can lose space before a click even exists.
In this scenario, optimization no longer means only choosing words for a page and also comes to mean providing sufficiently structured data so that a system can understand when a given product is an appropriate answer.
From visibility to transaction, Google prepares a second layer of metrics
The move happens at the same time as Google expands the Universal Commerce Protocol, used to connect commerce systems to AI experiences.
UCP already allows checkout on surfaces such as AI Mode and Gemini. Now, the Merchant Center integration hub is adding cart transfer to the retailer's site and enhanced checkout flow tests, while additional analytics tools are expected to arrive later.
This evolution is important because share of voice measures only part of the problem: being recommended does not necessarily mean selling.
If Google later connects exposure in AI responses to actions such as add to cart, transfer, checkout and purchase, the set of metrics could form its own conversational commerce funnel: participation in recommendations, product coverage, query stage and, finally, conversion.
Business Agent is also being brought to YouTube ads in the United States, allowing consumers to ask questions about products directly within the ad experience. This reinforces the trend of turning research, comparison and purchase into less separate stages than in the classic search funnel.
Share of voice still has important limits
The new indicators do not yet replace traditional metrics.
AI Performance Insights are currently limited to English-language queries and five markets. The report considers only organic AI traffic and does not include paid ads. In addition, the competitor group used in the share of voice calculation is defined by Merchant Center and cannot be changed by the retailer. Data is updated daily but may arrive with a few days' delay.
There are also situations that require caution in interpretation. An account without sufficient competitor data may, for example, appear with a share of voice of 100%, without that meaning real dominance of the category. Changes in the competitor set can also alter the percentages over time.
Therefore, the most relevant signal is not that positions, impressions or clicks are being replaced immediately. It is that Google has begun to institutionalize a second performance language for commercial searches with AI.
For retailers, the new question is no longer only “what position does my product appear in?” and now includes “in how many relevant conversations does my brand participate, at what stage of the decision, with how many products and against which competitors?”
The evolution of UCP analytics will be the next point to watch. If Google connects these visibility indicators to the cart and purchase stages, share of voice may cease to be just a new discovery metric and start to directly integrate the measurement of revenue generated by AI commerce experiences.
