Merchant Center AI Performance Insights: How to Read the Report and Act on It

What AI Performance Insights Actually Is

On September 16, 2026, Google announced in its holiday shopping update for retailers that AI performance insights in Merchant Center are available to businesses in Australia, Canada, India, New Zealand and the United States. Google's own help documentation describes the report as a view into how your brand is discovered across its generative AI surfaces, specifically AI Mode and AI Overviews in Search (an earlier help page also mentions the Gemini app).

Until now, merchants have had decent reporting on classic Shopping ads and free listings, and almost nothing on conversational discovery. This report is Google's first attempt at closing that gap. It is a benchmarking tool, not a conversion report: it tells you how visible your brand is relative to others, not how many orders AI Mode sent you.

The Four Metrics You Will See

Google's documentation lists a small set of metrics. Understanding exactly how each is defined prevents the most common misreading, which is treating share of voice like market share.

• Your share of voice is your AI impressions divided by the total impressions across you and your competitors for related queries. It is a relative number: if a competitor gets more visible, your share falls even if your own impressions stay flat.
• Your competitors' average share is the same calculation for your defined competitor set, based on the competitors Merchant Center already knows about. A gap between your share and this number tells you whether you are above or below the market norm.
• Query frequency shows how popular specific query types, product terms, shopping journey phases, or product attributes have been. Think of it as a demand signal that helps you decide what to prioritize.
• Query type classifies what shoppers were doing, for example searching by category, researching product specs, or looking for reviews, independent of how far along the buying journey they were.

One important limit from Google's documentation: the insights only cover conversational search queries that are mainly shopping-intent and brand-related queries. Everything else is excluded. So this is a partial picture of AI-driven discovery, not a complete one.

Filters and What They Mean for Interpretation

The dashboard uses the standard Merchant Center filter bar with product category, time period, countries where you are eligible, and traffic. Google notes there is no single report that spans all categories, so you analyze one product category at a time.

That design has a practical consequence. A merchant with five product categories has five separate share-of-voice readings, and they will often disagree. A store can be strong in one category and nearly invisible in another. Resist the urge to average them. Instead, rank your categories by revenue and read the report for your top two or three first.

A Practical Way to Read the Report

Here is a short routine that keeps you from drowning in dashboards.

1. Start with the gap. Compare your share of voice to the competitor average in your best-selling category. Above average means defend; below means diagnose.
2. Check the query types. If shoppers in your category mostly ask spec-research or review-style questions and your listings have thin descriptions, that is a data quality gap, not a bidding gap.
3. Look at query frequency for attributes. If a specific attribute such as a material, size or compatibility term is trending and you do not submit it, add it.
4. Pick one change per category. Make one data improvement, wait for a few weeks of data, then compare. Changing five things at once leaves you unable to learn which one worked.
5. Log it. Keep a simple spreadsheet of date, category, change made and share of voice before and after.

The Product Data Levers That Plausibly Move AI Visibility

Google has said that product data is a foundational input to its AI-powered formats, and its holiday guidance for retailers stresses keeping the feed accurate and adding conversational details. It has not published a ranking-factor list for AI Mode, and nobody outside Google can claim to know one. What you can do is strengthen the inputs Google has told you it reads.

• Titles that say what the product is. Plain, specific titles with the product type, key attribute and brand beat marketing phrases.
• Descriptions that answer questions. Shoppers talking to AI Mode ask about fit, material, compatibility, care and use cases. If the answer is in your description, it can be used. If not, it cannot.
• Complete attributes. Color, size, material, gender, age group, GTIN, and product highlights where relevant. Missing attributes are the most common avoidable gap we see in scans.
• Accurate price and availability. A mismatch between feed and landing page is both a policy risk and a trust problem. See our guide to price accuracy violations.
• Merchant reviews and brand information. Seller ratings and a consistent brand presence help Google associate your products with your brand. Our piece on seller ratings and reviews compliance covers the requirements.

For deeper background on how generative results use structured data, read Google Shopping and AI Overviews in 2026 and structured data for AI and voice assistants.

Why Compliance Comes Before Visibility

A product that is disapproved or an account that is suspended has no share of voice to measure. The most expensive mistake in this whole area is spending effort on conversational optimization while a policy problem sits unresolved in the Needs Attention tab.

Before you invest time in the insights report, confirm that your account is clean: no active misrepresentation flag, no landing page mismatches, working shipping and return policy pages, and accurate contact details. Google's September consolidation of its Shopping policies did not add restrictions, but it is a good moment for a clean-up; see our breakdown of the policy consolidation.

A free scan at gmcunbanned.com runs 57 checks against the issues that most often trigger suspensions, which makes it a fast way to confirm the foundation before you build on it.

What the Report Cannot Tell You

• It does not report clicks, conversions or revenue from AI surfaces. Pair it with your analytics. Our GA4 measurement setup guide helps you see what traffic actually arrives and converts.
• It does not explain why a competitor is ahead. It shows the gap; you supply the hypothesis.
• It is limited to eligible countries and to conversational queries.
• It is a relative metric, so market-wide growth in AI usage does not show up as growth in your share.

Treat it as a compass, not a scoreboard. Its most valuable role is telling you which category deserves attention first.

Quick Action Checklist

Frequently Asked Questions

Is AI performance insights available in every country? No. Google's September 16, 2026 announcement named Australia, Canada, India, New Zealand and the United States. Availability may widen, so check Merchant Center rather than assuming.

Does a higher share of voice mean more sales? Not directly. It measures relative visibility in conversational results, not conversions. Use analytics to connect it to revenue.

Can I change who my competitors are? Google's documentation says competitor data is based on the competitors currently available in Merchant Center. Check your account's competitive visibility settings for what you can adjust.

Will fixing my feed immediately change the numbers? Expect a lag. Feed changes need to be processed and then reflected in query data, so judge results over several weeks, not days.

Where should I start if I have limited time? Fix any open policy issues, then improve titles and descriptions on your top-revenue category, and check the report again after a few weeks. A free scan at gmcunbanned.com will surface the policy and landing page issues first.