Google Unveils Merchant Center AI Performance Insights Pilot: A New Frontier for Retailers in the AI Era

MOUNTAIN VIEW, CA – In a significant move that signals Google’s evolving approach to retail analytics in the age of generative AI, the tech giant last week launched a limited pilot program for its Merchant Center, offering select retailers unprecedented insights into how consumers interact with AI Mode and AI Overviews for shopping-related queries. This new feature, dubbed "AI performance insights," aims to bridge a critical data gap for e-commerce businesses navigating the rapidly changing landscape of AI-powered search.

The pilot, confirmed by independent SEO consultant Brodie Clark who gained early access through a client sub-account, marks the first instance of Google providing query-level data specifically for its AI-powered search surfaces. Clark promptly shared screenshots of the new interface, highlighting its potential to inform product feed optimization, even while acknowledging its inherent limitations.

While the introduction of this data is a welcome development for businesses managing product feeds, it comes with important caveats. Google groups the questions asked by users rather than listing them individually. This means retailers will gain insight into the "vocabulary of a category" – the common attributes and themes users inquire about – rather than the precise, long-tail queries themselves. This distinction is crucial: it’s enough to guide improvements in product listing attributes, but it won’t, for example, confirm whether a specific AI Mode interaction led to traffic on a retailer’s site.

This pilot represents a cautious yet notable step by Google to offer greater transparency into its AI search features, particularly as the integration of generative AI continues to reshape how users discover and research products online. For retailers, understanding these grouped query patterns could be a powerful tool for enhancing product visibility and relevance in an increasingly AI-driven marketplace.

A Timeline of Google’s AI Reporting Evolution

Google’s journey toward providing AI-related performance data has been a measured and often complex one, marked by announcements, limited tests, and ongoing industry demands for greater transparency. The Merchant Center pilot is not an isolated event but rather the latest chapter in this unfolding narrative.

From Marketing Live Announcement to Pilot Launch

The first official hint of these new capabilities came in May, during the annual Google Marketing Live event. At this conference, Google executives outlined their vision for integrating AI more deeply into advertising and commerce, with a specific mention of upcoming reporting tools. Roughly seven weeks later, in July, the "AI performance insights" pilot officially opened its doors to a limited number of U.S.-based Merchant Center accounts.

This pilot follows a series of related developments that highlight Google’s evolving strategy and the persistent calls from the SEO and publishing communities for more actionable AI data:

• June – Search Console AI Reports: Just weeks before the Merchant Center pilot, Google initiated testing of dedicated generative AI performance reports within Search Console. This initial rollout was restricted to a subset of UK sites. These reports provided impressions data, broken down by page, country, device, and date. Crucially, they conspicuously lacked both click data and query-level metrics, leaving many search professionals feeling that the most critical information was still withheld. This absence became a central point of discussion, underscoring the industry’s hunger for more granular insights into user behavior on AI surfaces.

• June – UK CMA Intervention: In the same eventful week that Search Console AI reports began testing, the UK’s Competition and Markets Authority (CMA) imposed a significant conduct requirement on Google. This mandate specifically addressed publisher controls and reporting related to AI search features. The CMA’s interpretive notes explicitly called for impressions, click-throughs, and click-through rates for search generative AI features, to be reported separately from general search. Google was given a nine-month window to implement these changes, signaling strong regulatory pressure for increased data disclosure, particularly concerning user engagement. The fact that current Google reports, including this Merchant Center pilot, still largely omit click data highlights the ongoing tension between regulatory demands and Google’s implementation timeline and strategy.

• July – Google’s Stance on Third-Party Tools: Approximately three weeks before the Merchant Center pilot went live, Google addressed Chief Marketing Officers (CMOs) directly, clarifying that third-party AI-visibility tools do not have access to its internal metrics. Google explicitly positioned Search Console and Merchant Center reporting as the authoritative "baseline" for tracking gains in AI visibility. This statement effectively underscored Google’s intention to maintain control over its AI data and channel reporting through its proprietary platforms, further emphasizing the importance of any new official reporting tools.

When these events are viewed in conjunction, the Merchant Center pilot appears as a calculated, incremental step within Google’s broader AI strategy. It’s a response to market demand and, indirectly, regulatory pressure, offering a piece of the puzzle to a specific audience (merchants with product feeds) while many questions regarding comprehensive AI performance data for all publishers remain unanswered. The phased rollout and the consistent absence of click data across different reporting initiatives suggest a deliberate, cautious approach by Google to managing the flow of information about its nascent AI search ecosystem.

Diving Deep into the AI Performance Insights Metrics

The "AI performance insights" report is housed within the familiar Google Merchant Center interface, accessible under the "Analytics" section, then "Products," and finally, a new "AI performance" tab. Google’s help documentation frames this report as a means to understand "how a brand is discovered across AI Mode and AI Overviews." For retailers, this represents a valuable opportunity to optimize their product data and potentially improve their standing in AI-driven shopping experiences.

Understanding the Nuances of the New Report

The report categorizes shopping questions into several insightful dimensions, each offering a unique lens into consumer intent and product interest:

• Query Type: This metric sorts shopping questions based on their fundamental nature. Google provides examples such as "searching by category," "researching product specs," or "looking for reviews." Understanding query types allows retailers to tailor their content and product descriptions to align with different stages of the buying journey. For instance, if a high frequency of "researching product specs" queries is observed for a particular product category, a merchant might prioritize detailed technical specifications, comparison charts, and elaborate feature descriptions in their product listings. Conversely, a prevalence of "looking for reviews" queries might prompt an emphasis on integrating customer testimonials and star ratings prominently.

• Query Frequency: Complementing query type, this metric indicates the popularity or volume of a given query type. High-frequency query types signal strong consumer interest in particular aspects of a product or category. Retailers can leverage this to prioritize optimization efforts, focusing on areas of highest demand. For example, if "durability" queries for outdoor gear show high frequency, ensuring comprehensive information about material strength and warranty in product feeds becomes paramount.

• Phase of Shopping Journey: This dimension groups questions by how far along the shopper is in their purchasing decision process. Are they in the initial exploration phase, comparing options, or ready to make a purchase? By understanding the dominant shopping phase, merchants can strategically refine their product information, calls to action, and even their broader marketing messages. An early-stage shopper might benefit from educational content and broad category descriptions, while a late-stage shopper needs clear pricing, availability, and shipping information.

• Product Terms: Perhaps the most directly actionable metric, "Product terms" reveals the specific vocabulary shoppers use when describing what they want. Google’s documentation offers examples like "maximum cushioning" and "arch support" for footwear. This data is invaluable for enriching product attributes in a retailer’s feed. If shoppers are consistently asking about "waterproof capabilities" for electronics, and a merchant’s product feed lacks this attribute, it presents a clear, actionable opportunity for improvement. Filling these attribute gaps directly enhances the product’s discoverability and relevance in AI-generated responses.

• Share of Voice: This metric aims to provide a competitive benchmark, comparing a brand’s AI impressions against those of its competitors. It offers a high-level view of a brand’s visibility within the AI search ecosystem relative to others in its category. While intuitively appealing, this metric comes with significant caveats, which will be discussed in more detail below.

Actionability and Practical Applications

The most potent takeaway from this report, as highlighted by the original article, is its direct utility for improving attribute completeness. For years, a common challenge for e-commerce businesses has been ensuring that their product data is as rich and comprehensive as possible. The "AI performance insights" report provides a clear, demand-driven signal from Google’s AI surfaces, indicating precisely which features or characteristics consumers are actively inquiring about. If shoppers in a particular category frequently ask about a specific feature that is currently absent from a merchant’s product feed, the fix is straightforward and impactful. Adding this attribute can significantly boost the product’s chances of being featured in relevant AI Overviews and AI Mode responses.

Beyond attribute completeness, understanding these metrics can inform broader content strategy. The "query type" and "product terms" data can guide the creation of more effective product descriptions, FAQ sections, and even blog posts or buying guides that directly address user questions and pain points. For instance, if "eco-friendly materials" frequently appears in product terms, a merchant can develop content highlighting their sustainable practices and product compositions.

Indirectly, these insights could also influence inventory planning or product development. A persistent demand for certain features or types of products, as revealed by the query data, might signal market trends that a retailer could capitalize on by adjusting their stock or even influencing future product designs.

In essence, while the data is grouped and not hyper-granular, it offers a strategic compass for merchants, directing their efforts towards optimizing the foundational data that powers their presence in the increasingly AI-centric shopping journey.

The Limitations and Ambiguities of Current AI Metrics

While the introduction of AI performance insights is a positive development, it’s crucial for search professionals and retailers to approach the data with a clear understanding of its limitations. The report, in its current pilot form, leaves several significant gaps and presents certain metrics in ways that require careful interpretation.

What the Data Doesn’t Tell You

The most fundamental limitation is that none of the metrics in Merchant Center’s AI Performance insights show you a real question anyone typed. Instead, as noted, the data is aggregated and grouped. This means retailers understand the shape of the demand – the prevalent themes and categories of inquiry – rather than the specific, verbatim queries themselves. While useful for identifying trending attributes, this grouping prevents the kind of granular keyword research possible with traditional Search Console data, making it unsuitable for generating precise keyword lists for paid campaigns or highly targeted SEO content.

Share of Voice Caveats: The "Share of Voice" metric, while seemingly offering a direct competitive comparison, is fraught with ambiguities:

• Fixed Competitor Set: Google calculates Share of Voice as a brand’s AI impressions divided by total impressions across that brand and its competitors. However, the critical detail is that retailers cannot change or customize this competitor set. It is defined by the competitors already available and recognized within Merchant Center. This lack of control means the "competitor" benchmark might not align with a retailer’s actual competitive landscape, potentially skewing the perception of their performance.
• Misleading Numbers: The metric can produce numbers that appear to reflect performance but are misleading. If a brand lacks sufficient AI impressions, its Share of Voice will display as a zero, which could be misinterpreted as poor performance rather than simply low visibility or limited data. Conversely, if an account has no competitors defined within Merchant Center (perhaps for highly niche products or new accounts), its Share of Voice will display as 100%. This seemingly perfect score is not indicative of market dominance but rather an absence of comparative data, and could be mistakenly celebrated as a strong quarter. Agencies, in particular, will face the challenge of explaining these nuances to clients, as a "zero" or "100%" can look starkly different on a performance slide than in a detailed conversation.

Scope Filters and Exclusions: The data presented in the AI performance insights is subject to specific filters and exclusions, which narrow its scope:

• Organic AI Traffic Only: The report is explicitly restricted to organic AI traffic. Crucially, paid ads traffic is not included. This means that retailers running shopping ads through AI Overviews or other AI surfaces will not see that performance reflected here, creating a fragmented view of their overall AI presence. This exclusion is significant, as it prevents a holistic understanding of how both organic and paid efforts contribute to AI visibility.
• Single Category Filtering: The product category filter allows users to analyze data for only one category at a time. There is currently no report that covers all categories collectively, making it cumbersome for large retailers with diverse product catalogs to gain an aggregated view of their AI performance across their entire business.
• Intent-Specific Queries: Insights only count conversational queries that specifically indicate shopping or brand intent. Any other types of conversational queries are not factored into the report. While this ensures relevance for e-commerce, it means the report doesn’t capture the full spectrum of user interactions with AI Mode or AI Overviews that might indirectly lead to shopping intent.

Missing Click Data: The most significant and persistent absence in all of Google’s current AI reporting initiatives, including this pilot, is click data. After more than a year of generative AI features being tested and rolled out, search professionals are still without direct metrics on how many users click through from AI Overviews or AI Mode responses to their websites. Impressions show how often a link to a product appeared, and John Mueller has clarified how these impressions are counted, with these rules also applying to the basis of "Share of Voice." However, impressions alone do not indicate engagement or traffic. The lack of click-through rates (CTR) and actual clicks makes it exceedingly difficult for retailers to quantify the direct business impact of their AI visibility.

Other Unanswered Questions: Individual queries remain inaccessible, and the precise details about the competitor set used for "Share of Voice" have not been shared. Furthermore, Google has not announced whether grouped query information will eventually be available in Search Console, particularly for sites that do not have product feeds, such as affiliate marketers or content publishers.

These limitations underscore that while the Merchant Center pilot is a step forward, it offers an incomplete picture. Retailers must exercise caution and context when interpreting these new metrics, understanding what they reveal and, perhaps more importantly, what they still obscure.

Broader Implications for Search Professionals and the Digital Ecosystem

The introduction of AI performance insights in Merchant Center has far-reaching implications, not just for the retailers directly involved in the pilot, but for the broader digital marketing community, including agencies, affiliate marketers, and content publishers. It highlights a growing disparity in data access and signals Google’s strategic priorities in the evolving AI search landscape.

Navigating the Evolving Landscape of AI Search

For Merchants with Access: For the fortunate few retailers included in the pilot, two significant workflow changes emerge immediately:

1. A New Demand Signal: The grouped query data provides a fresh, AI-specific demand signal. This is distinct from anything available in Search Console and offers a direct input for prioritizing product feed optimizations. Merchants can now see what attributes are missing or underrepresented in their listings based on what AI users are actively seeking. This empowers them to make data-driven decisions to enhance product discoverability.
2. Mandatory Reporting Metric: The "Share of Voice" metric, despite its complexities, will inevitably become a line item in monthly performance reports. This necessitates a proactive approach from merchants and their agencies to understand, explain, and contextualize this metric for stakeholders, especially given its potential for misinterpretation due to the fixed competitor set and reporting nuances (zeros for low impressions, 100% for no competitors).

For Merchants Without Access: The majority of merchants, who are not yet part of this limited US pilot, remain at a disadvantage. Their AI reporting capabilities are still restricted to the impression data found in Search Console, which lacks any query dimension. This creates an uneven playing field, where some businesses can proactively optimize for AI search while others must rely on more traditional, less direct signals.

For Non-Product-Feed Sites (Affiliates, Review Sites, Editorial Teams): This segment of the digital ecosystem faces an even wider data gap. Affiliate sites, review platforms, and editorial teams publishing buyer guides often compete directly with brands for visibility in AI Mode answers. However, unlike merchants, they do not receive these grouped shopping-query insights. Their AI reporting remains confined to the impression data in Search Console, which covers the same two AI surfaces (AI Mode and AI Overviews) but completely lacks any query dimension. This means that while they are directly impacted by AI search, they are effectively flying blind when it comes to understanding what users are asking AI about in their respective niches. The disparity in reporting capabilities is stark and could significantly impact their ability to adapt their content strategies for the AI era.

For Agencies: Marketing agencies face a unique set of challenges. The "Share of Voice" metric, while attractive for inclusion in client decks, poses a significant communication hurdle. Its reliance on a fixed, non-editable competitor set and its propensity to display misleading "zero" or "100%" values means agencies must invest considerable effort in educating clients about its true meaning and limitations. Presenting these numbers without proper context risks misrepresenting performance and undermining trust. Agencies will need to develop robust narratives to explain why a zero doesn’t necessarily mean a bad quarter, or why a hundred isn’t always a sign of market dominance.

Google’s Strategy and Future Outlook

The "filing decision" – where Google chooses to place specific data reports – is often indicative of its underlying philosophy. As SEJ contributor Slobodan Manic argued previously, placing AI visibility within Search Console on purpose implied that AI visibility is search visibility, and thus belongs in the primary search tool. The Merchant Center pilot doesn’t entirely refute this; rather, Google’s CMO guidance names both dashboards as first-party reporting. This suggests a nuanced view: grouped query information relevant to product feeds logically lands in the dashboard merchants use for managing those feeds, while broader search performance remains in Search Console. This compartmentalization might reflect Google’s internal engineering priorities and its approach to segmenting data based on its most direct application.

Looking ahead, Google has stated that the pilot will expand to Australia, Canada, India, and New Zealand in the coming months. This broader rollout will provide valuable data on how these metrics hold up across different accounts, languages, and product categories, offering a more comprehensive test of the system’s utility and scalability.

The open question, however, remains whether any of these new, more granular metrics will eventually cross over into the Search Console reports for the wider publishing ecosystem. Google indicated in June that it would add metrics to Search Console reports "over time," but without specifying which ones or when. This leaves content publishers in a state of uncertainty, eagerly awaiting more actionable data.

A nearest "fixed point" in this evolving landscape is the CMA’s nine-month implementation window for engagement reporting. This mandate covers clicks and click-through rates for UK publishers in relation to search generative AI features. While this applies to a different dashboard, a different audience, and a different jurisdiction than the current Merchant Center pilot, it sets a precedent. The CMA’s intervention highlights the potential for regulatory pressure to drive greater transparency and data disclosure from Google, especially concerning the most sought-after metric: user engagement (clicks). Whether this regulatory push will eventually lead to similar comprehensive reporting for all publishers globally remains to be seen, but it certainly puts a spotlight on the missing pieces of Google’s AI data puzzle.

The continuous tension between Google’s desire to maintain control over its proprietary data (and potentially protect user privacy by aggregating queries) and publishers’ and merchants’ desperate need for actionable data to optimize their online presence defines the current state of AI search analytics.

Conclusion: A Step Forward, But Many Miles to Go

The Google Merchant Center’s "AI performance insights" pilot is undeniably a significant step forward in providing retailers with a clearer understanding of how their products are being discovered in the age of generative AI. By offering aggregated insights into the types of questions consumers ask AI Mode and AI Overviews, and the product terms they use, Google is empowering merchants to enhance their product data, improve attribute completeness, and strategically refine their content. This represents a tangible, albeit limited, demand signal that can directly influence product feed optimization and potentially boost visibility.

However, as highlighted by Brodie Clark, "In its current form, similar to the recent rollout of AI reporting in Search Console, there isn’t a great deal of actionability behind the data, though it is good to see at least some form of query data being included – something that has been lacking in GSC." This sentiment captures the dual nature of this pilot: a welcome glimmer of insight, yet still far from the comprehensive, granular data that search professionals truly need.

The most glaring absence remains click data. Without knowing how many users actually click through from AI-generated responses to their sites, retailers and publishers alike are left unable to fully quantify the return on investment for their AI visibility efforts. Furthermore, the limitations surrounding "Share of Voice," the exclusion of paid traffic, and the lack of comprehensive, cross-category reporting underscore that this pilot is just a preliminary offering.

Crucially, the disparity in reporting capabilities between merchants with product feeds and other vital segments of the online ecosystem, such as affiliate sites and editorial publishers, remains stark. These non-product-feed sites compete for the same AI Mode answers but are denied access to the grouped query insights, leaving them with only impression data from Search Console. This creates an uneven playing field and hinders their ability to adapt to the evolving AI search landscape.

As Google continues to integrate AI into its core search experience, the demand for greater transparency and more actionable data will only intensify. While the Merchant Center pilot offers a valuable, albeit imperfect, tool for a select group of retailers, the broader digital marketing community will continue to press for more comprehensive, consistent, and granular reporting – especially the long-awaited click data – to truly navigate and thrive in the era of AI-powered search. The journey towards full AI search visibility has certainly begun, but many miles remain before a complete and equitable data landscape is achieved.