In a significant move to demystify the "black box" of generative AI search, Google has launched a pilot program within its Merchant Center that provides retailers with unprecedented—albeit limited—visibility into how their products appear in AI Mode and AI Overviews.
The new "AI performance insights" report, which began rolling out to a select group of U.S. merchants last week, marks the first time Google has offered query-level data for its generative AI surfaces. While industry experts caution that the data is not a panacea for performance tracking, it represents a pivotal shift in Google’s transparency strategy regarding its evolving search experience.
The Core Facts: What the Pilot Reveals
For years, the "holy grail" for SEOs and e-commerce managers has been granular data on how users interact with AI-generated search results. The new report, located under the Analytics > Products > AI performance tab in the Merchant Center, aims to bridge this gap.
Independent SEO consultant Brodie Clark, who gained access to the feature via a client sub-account, was among the first to document the tool. According to his findings, the report does not provide a raw, exhaustive list of individual user search queries. Instead, it utilizes a sophisticated grouping mechanism. Google categorizes user questions based on intent and vocabulary, allowing retailers to see the "shape" of demand within their product categories rather than the specific, verbatim strings typed by consumers.
This means that while a merchant might not see exactly who asked a specific question, they can identify the themes—such as specific product attributes or "jobs-to-be-done" queries—that are driving traffic to AI-generated shopping results.
A Chronology of AI Reporting
The launch of this pilot follows a flurry of activity from Google, which has been under mounting pressure from regulators and publishers to clarify how AI-generated search impacts traffic and brand visibility.
- May 2026: Google officially announced the development of AI performance reporting at the Google Marketing Live event, signaling to the industry that specialized dashboards were on the horizon.
- Late June 2026: Google initiated a test of dedicated generative AI performance reports within Google Search Console (GSC), initially restricted to a small subset of publishers in the United Kingdom.
- July 2026: In a directive to CMOs, Google clarified that third-party SEO tools lack access to internal proprietary metrics, effectively steering the industry toward using GSC and Merchant Center as the "source of truth."
- August 2026: The current pilot for Merchant Center AI performance insights officially went live for a select cohort of U.S.-based accounts.
This timeline suggests a strategic, staggered rollout, likely designed to test the load on Google’s infrastructure and the utility of the data before a wider global release.
Breaking Down the Metrics
The Merchant Center report is designed to be actionable rather than purely descriptive. By filtering data through specific lenses, Google is attempting to help merchants optimize their product feeds to better align with AI-driven discovery.
Key Performance Dimensions
- Query Type: This metric allows merchants to see if users are searching by category, researching specific technical specifications, or seeking social proof via reviews.
- Phase of the Shopping Journey: Questions are bucketed based on the consumer’s intent, helping brands identify whether they are capturing top-of-funnel discovery or bottom-of-funnel purchase intent.
- Product Terms: This is perhaps the most valuable feature. Google identifies the descriptive language users employ—such as "maximum cushioning" for footwear or "energy-efficient" for appliances. By integrating these terms into product feed attributes, merchants can effectively "feed" the AI the exact data it needs to surface their products.
- Share of Voice: This metric compares a brand’s AI impressions against a competitor set defined by Google. While useful for benchmarking, experts note that the metric is sensitive to the accuracy of the competitor list, which merchants currently cannot manually edit.
The Regulatory and Competitive Context
The timing of this rollout is not coincidental. In the United Kingdom, the Competition and Markets Authority (CMA) has imposed strict conduct requirements on Google, demanding that the search giant provide clear, segregated data on how its generative AI features impact publisher traffic.
While the CMA’s mandate specifically mentions the need for click-through rates (CTR) and impression data, the current Merchant Center pilot only addresses the "impression" side of the equation. Clicks remain a notable omission, leaving publishers and merchants in a state of limbo regarding the actual traffic-driving power of AI Overviews.
Furthermore, the disparity between the Search Console report (which covers non-product content) and the Merchant Center report (which covers product feeds) highlights a growing divide in search performance measurement. Editorial sites and affiliate publishers, who lack access to product feed tools, remain largely in the dark, forced to rely on the limited, query-free impression data provided in Search Console.
Implications for Search Professionals
For SEOs and digital marketers, this pilot necessitates a shift in workflow. The primary value proposition here is the demand signal. Previously, identifying the vocabulary used in AI-generated answers was largely guesswork or reliant on third-party sentiment analysis. Now, retailers have a direct, Google-verified list of attributes to optimize their feeds.
However, the "actionability" of the data remains a point of contention. As Brodie Clark noted, while it is a significant step forward to have any form of query data included, the current iteration lacks the depth required for advanced performance modeling.
Critical Challenges
- The "Zero-Click" Problem: Because clicks are not included, merchants have no way to verify if their presence in AI Overviews is actually translating into site visits.
- Metric Ambiguity: The "Share of Voice" metric can be misleading. A zero-percent share might indicate a lack of visibility, or it might simply mean the algorithm has not yet bucketed the site with relevant competitors. Conversely, a 100% share could indicate an isolated data set rather than market dominance.
- Limited Scope: Paid advertising data is entirely excluded from these reports. This creates a "blind spot" for marketing teams running integrated campaigns, as they cannot see the full picture of their brand’s presence in AI-enhanced search.
Official Guidance and Future Outlook
Google’s documentation emphasizes that the primary goal of these insights is attribute completeness. If a merchant notices that a significant volume of queries in their category references a feature they haven’t optimized for (e.g., "waterproof"), the path to improvement is clear: update the feed to include that attribute.
Looking ahead, Google has confirmed plans to expand the pilot to Australia, Canada, India, and New Zealand. This geographic expansion will be a litmus test for the scalability of the reporting. Whether these metrics eventually migrate to Search Console—or if they remain siloed within the Merchant Center—remains the subject of intense industry speculation.
For now, search professionals should treat the AI performance dashboard as a "directional" tool. It is an excellent guide for product feed hygiene and content strategy, but it is not yet a comprehensive analytical suite. Until Google integrates click-level attribution and provides more granular control over the competitor set, the "AI revolution" in search will continue to be a landscape where visibility is easier to measure than success.
Conclusion: A Step Toward Transparency
The introduction of the Merchant Center AI pilot is a calculated concession to a market that has grown increasingly frustrated with the opacity of generative AI. By providing grouped query data, Google has moved the needle, offering merchants a way to speak the "language" of the AI.
However, the gap between what merchants need to know—the direct impact on bottom-line revenue—and what they are permitted to know remains wide. As the industry awaits further integration and the eventual inclusion of click-through data to satisfy global regulatory pressures, the current pilot serves as a vital, if imperfect, window into the future of search. Marketers who adapt their feeds to these new insights will likely gain an early-mover advantage, even as the landscape continues to shift beneath their feet.
