Google Now Reports Your AI Mode Visibility. Here's How to Read the Numbers

AnalyticsAI searchSEOGEO
by Anton S
Google Now Reports Your AI Mode Visibility. Here's How to Read the Numbers

For two years, merchants asked Google a simple question: when my products show up in AI Overviews or AI Mode, how would I know? The answer was mostly "you wouldn't." AI impressions were folded into regular search totals with no way to separate them.

That's changed. Google now has two reports for this, one in Merchant Center and one in Search Console, and both got upgrades in September. On September 16 Google expanded Merchant Center's AI performance insights to Australia, Canada, India, New Zealand and the US. On September 24 Search Console added image-based searches to its reporting.

Neither report is complete, and they measure different things. For most Shopify stores, the Merchant Center one is the more useful of the two.

Report one: AI performance in Merchant Center

This one is built for product sellers. Google says it shows how your products are discovered across AI Mode, AI Overviews and the Gemini app.

Where to find it: in Merchant Center, go to Analytics, then Products, then the "AI performance" tab at the top. Per Google's help page, it covers English-language queries for accounts in the five countries above. It only counts organic AI traffic. Paid ads are excluded.

The main metrics:

  • Share of voice: your brand's impressions divided by the total impressions for you and your competitors on related queries.
  • Competitors' average share: the benchmark to compare against.
  • Frequency: how common a type of search, term or attribute is.
  • Products showing: how many of your products appear for the top terms and intents.

It also sorts queries into three shopping stages: Discovery (general browsing), Evaluation (comparing specs) and Ready to Buy.

The stage split is the most useful part. A store with healthy share of voice in Discovery and almost none in Evaluation has a specific problem: Google mentions it when people browse, but drops it once they start comparing details. That usually comes down to missing attributes, the specs and materials a comparison depends on.

Read the fine print before you celebrate

Two details in Google's help page can mislead you.

First, if you haven't defined competitors, your share of voice shows 100%. That's not a win. It means there's nothing to compare against. Set up competitors before you read anything into the number.

Second, data updates daily but with a few days' lag, and a dash means no impression data at all. Don't panic over a dash in your first week. Panic a bit if it's still there after a month.

Report two: Generative AI performance in Search Console

Google introduced this report on June 3. It covers AI Overviews and AI Mode in Search, plus AI features in Discover, and it's still rolling out to a subset of sites. If you don't see it yet, you're not doing anything wrong.

What it gives you, per PPC Land and SEO Hacker's walkthrough: impressions, broken down by page, country, device and date.

What it doesn't give you is a long list:

  • no clicks, so no click-through rate
  • no queries, so you can't see which questions triggered your page
  • no split between AI Overviews and AI Mode

So the report tells you which of your URLs Google's AI features use, and how often. That's still worth knowing. If your blog guide on "how to choose a running shoe" gets AI impressions and your actual running shoe collection gets none, you've learned where Google thinks your expertise sits.

The September addition: image searches

On September 24 Google added web multimodal reporting to Search Console. It covers searches made with Google Lens, Circle to Search on Android, image uploads to Google Search, and Chrome's "Search this image." You'll find it as a new search type filter in both the Performance report and the Generative AI report.

According to Relevant Audience's breakdown, the standard Performance report shows clicks, impressions, CTR and position for multimodal traffic, but no query data, because the searches are images rather than words. Google also confirmed this traffic "wasn't previously in the counts," so these are new numbers, not old ones moved into a new bucket.

For stores selling visual products (furniture, fashion, home decor), this is the first real look at how often people find you by pointing a camera at something. If the numbers are meaningful, it's a direct argument for better product photography and descriptive image file names and alt text.

What these reports can't see

Both reports cover Google only. They tell you nothing about ChatGPT, Perplexity, Copilot or Meta's new Muse agent, all of which now recommend products and send shoppers to stores.

They're also delayed and summarized. Merchant Center shows share of voice by category and stage, not which competitor beat you on a specific prompt. For that you still need to run the questions yourself, or use a tracker built for it. (Seokai's AI visibility tracking does this for ChatGPT, which is the gap Google's reports leave open.)

What Google says moves the numbers

In the same September 16 post, Google gave a rare number: retailers that follow its feed best practices see about a 5% conversion increase. It's pushing three newer attributes:

  • video links for visual ad formats
  • conversational attributes, which in testing with lululemon were used in 50% of AI Mode recommendations
  • loyalty data, so member pricing shows up in results

The lululemon number stands out. Half of AI Mode recommendations drew on the extra descriptive attributes the retailer supplied. That's Google telling you, about as directly as it ever does, that rich product data gets used.

What to do

  1. Open Merchant Center's AI performance tab this week. If you don't see it, check that your account is in an eligible country and your feed is active.
  2. Define your competitors before you read share of voice. Pick three to five stores you really lose sales to.
  3. Look at the stage split. Weak in Evaluation? Add specs, materials, dimensions and compatibility to your Shopify product data and feed.
  4. Check Search Console for the Generative AI report. If you have it, list the pages getting AI impressions and compare them with the pages you want there.
  5. Turn on the multimodal filter if you sell visual products, and watch it through the holidays.
  6. Write down your baseline now. Note share of voice and AI impressions in early October so you have something to compare against after BFCM.

These reports arrived late and they're thin, but they give you real numbers instead of guesses. Start recording them now and you'll have a meaningful trend by spring.

Share this Story

Blurred blog main image. Google Now Reports Your AI Mode Visibility. Here's How to Read the Numbers

Related Blogs

Pastel 3D pie chart with one slice pulled out, beside a small speech bubble
AI searchGEOAnalyticsBest practices

Share of Model Is the New Market Share

Your rank report cannot tell you whether an AI assistant recommends you. Share of model can. Here is how to define it, instrument it, and read the number without fooling yourself.

July 26, 2026 by Anton S

Read More
Minimal 3D illustration of a central AI assistant node on a pastel platform, representing generative engine optimization.
GEOAI searchSEO

What Is GEO (Generative Engine Optimization)? A 2026 Guide for Shopify

Shoppers are starting to ask AI assistants what to buy instead of scrolling Google. Here's what Generative Engine Optimization means for your Shopify store in 2026, and a concrete checklist to start showing up in AI answers.

May 1, 2026 by Anton S

Read More
Amazon Blocked Meta's Shopping Agent. Shopify Let It In. Where Do You Stand?
AI searchShopifyE-commerceGEO

Amazon Blocked Meta's Shopping Agent. Shopify Let It In. Where Do You Stand?

In one September weekend Amazon and Shopify took opposite sides on AI shopping agents. Here's what each was protecting, and how a Shopify merchant should set their own agent policy.

October 2, 2026 by Anton S

Read More