Search Console’s Generative AI Reports: What They Show and How to Read Them

Many of us have been trying to work out how much of our Google visibility now comes from AI Overviews and AI Mode. The new Search Console AI Overviews report is the first official answer. On 3 June 2026, Google announced Search Generative AI performance reports. This is a separate view that counts how often links to your pages are shown inside Google’s generative AI features in Search and in Discover.

The short version: it shows impressions, split by page, country, device (Search only) and date. It does not show clicks, click-through rate or queries. And at launch it is being tested with a subset of sites, so many properties won’t see it yet.

Below I explain what the report has, what it leaves out, how it fits with the data you already have, and how I would turn it into decisions instead of just another chart.

What Google Announced on 3 June 2026

Google shared the news on the Search Central blog. Two reports are rolling out:

• Generative AI performance for Search, covering AI Overviews and AI Mode.
• Generative AI performance for Discover, covering generative AI features shown in the Discover feed.

Both sit next to the existing performance reports. They do not replace them. Google called them a dedicated view of visibility from generative AI features. It said it plans to add more metrics over time, but it gave no timeline.

The rollout starts with a subset of websites so Google can test and get feedback. Google did not publish how sites are picked or a date for wider access. Early access seems to be focused on UK site owners. Alongside this, Google is separately testing a control that lets UK sites opt out of generative AI features.

What the Search Console AI Overviews Report Actually Shows

Think of it as a performance report with one metric and four ways to slice it.

The Metric: Impressions

An impression is counted when a link to your site is shown to a user inside a generative AI feature. That is the only metric at launch.

For Discover, Google’s help page says the link has to be scrolled into view. A result is counted once per session, even if the user scrolls past it and back.

The Dimensions

• Pages: which of your URLs were shown. As in the rest of Search Console, data goes to the canonical URL (the main version of the page), not always the exact URL someone landed on.
• Countries: where the search or Discover view happened.
• Devices: desktop, mobile and tablet. This split is only in the Search report, not Discover.
• Dates: hourly, daily, weekly and monthly views. Dates follow Pacific Time, like the rest of Search Console.

The hourly view is more useful than it sounds. If you publish a news piece or update a key page, you can see within the same day whether it starts showing in AI features. I expect to use this view a lot after content updates.

Eligibility and Thresholds

The report may not appear at all if your site has too few generative AI impressions. Sites that are excluded from generative AI features won’t show data either, for the obvious reason that there is nothing to count.

Google’s help pages also say Search Labs experiments are not included.

What the Report Does Not Show

This is where most questions come from, so let me be clear.

• No clicks. You can see that a page was shown in an AI Overview or AI Mode answer. You can’t see how many people clicked through from that answer.
• No CTR. Without clicks there is no click-through rate.
• No queries. There is no query dimension. So you can’t see which searches triggered the AI answer that cited you.
• No average position. The report doesn’t tell you where your link sat in the AI answer.
• No split between AI Overviews and AI Mode. The Search report groups generative AI features together.
• Little history. It is a new report. You can’t compare with last year or even most of last quarter.

Google has said more metrics are coming. Until then, I treat this as a visibility report, not a traffic report.

How This Differs From the Web Search Type Data You Already Have

This is the part people most often get wrong. AI Overviews and AI Mode are not new to Search Console data. Google already counts them in the standard Performance report under the Web search type, mixed in with regular results.

So the new report doesn’t add impressions that were missing. It pulls out a slice that was already inside your totals.

A few old rules about how the Web data handles AI features are worth remembering when you compare the two:

• An AI Overview takes a single position on the results page. Every link inside it gets that same position. That is why position data has never told you where you sat inside an AI Overview.
• An impression in an AI Overview only counts when the link is actually visible. For example, after the user expands the overview or scrolls to the link.
• In AI Mode, a follow-up question counts as a new query.

If you want a refresher on how the two AI features behave differently for users, see my guide to AI Overviews vs AI Mode.

Why the Totals Won’t Simply Add Up

Don’t expect AI impressions plus “non-AI” impressions to equal your Web total. A page can appear in an AI Overview and in a regular blue link on the same results page. Search Console has its own rules for counting and removing duplicate impressions per property and per URL.

Chart totals and table totals can also differ inside the same report because of how data is grouped. That is normal Search Console behaviour, not a bug in your property. I get asked about this often, and the answer is almost always the same.

How to Find and Read the Report

If your property is in the test group, the reports appear in the Performance section of Search Console. There are separate entries for Search and Discover. If you can’t see them, the likely reasons are that your site isn’t in the first subset yet, or it doesn’t have enough AI impressions to pass the threshold.

Once you have access, here is how I would approach the first session.

1. Set the widest date range available. History is short. Start with everything you have and look at the overall trend before slicing.
2. Open the Pages tab and export it. Sort by impressions. Note which page types lead: guides, product pages, category pages, news.
3. Pull the same pages from the Web performance report. Use the same date range and export clicks, impressions and queries for those URLs.
4. Join the two exports in a spreadsheet. For each URL, work out what share of its Web impressions come from AI features.
5. Check Countries and Devices. Look for markets or devices where AI visibility is much higher or lower than your normal traffic mix.
6. Switch to hourly or daily view around changes. When you publish or update content, watch how fast it starts to collect AI impressions.
7. Repeat monthly and keep the exports. Build your own history, so comparisons mean more over time.

Remember the usual 1,000-row limit in the interface. For larger sites, filter by folder or page group so you don’t lose the long tail (the many low-traffic pages).

Using the Data to Make Decisions

Impressions alone don’t tell you if AI features help or hurt you. But together with your other data, they can answer some useful questions.

1. Find Pages With High AI Visibility but Falling Clicks

Take your joined sheet. Flag URLs where AI impressions are a large share of Web impressions and Web clicks have dropped. On these pages, the searcher may be getting the answer on the results page. This is the list I would look at first.

For those pages, ask if there is a reason to click: a tool, a calculator, original data, a comparison, a template, local stock. If the page only repeats a definition, AI features can sum it up without sending anyone through.

2. Find Important Pages With No AI Visibility

Make a list of your most important business pages and check whether they appear in the AI report at all. Not appearing isn’t always a problem, since many queries don’t trigger AI features. But it’s worth checking for informational pages on competitive topics.

Google says nothing special is needed to appear in AI features beyond normal Search eligibility. In practice, clear structure, direct answers near the top and well-supported claims make content easier to cite. My guide on writing content AI assistants can quote covers this in more detail.

3. Connect Visibility to Real Visits

Search Console won’t tell you about clicks from AI features, so pair it with your analytics. Watch landing-page sessions in GA4 for the URLs that gain AI impressions. Track conversions from those pages too.

You won’t get a perfect line from “shown in AI Overview” to “converted”. But you can see whether pages with growing AI visibility are holding up on engagement and revenue. If you also want to split out visits from ChatGPT, Perplexity and Gemini, see my guide to tracking AI traffic in GA4.

4. Report AI Visibility to Stakeholders Honestly

Managers and clients will ask, “How are we doing in AI search?” This report gives you a solid number for Google’s AI features. That is better than screenshots or third-party guesses.

Show it as visibility, next to clicks from the Web report, not as a replacement. If you already run a reporting dashboard, add a monthly AI impressions table from your exports. My walkthrough on building a free SEO dashboard in Looker Studio is a good base for that.

A Practical Example

Say you run a home insurance comparison site. The AI report shows your “what does buildings insurance cover” guide gets far more AI impressions than any other page, mostly on mobile in the UK. At the same time, the Web report shows that guide’s clicks have slipped while impressions stayed steady.

That mix suggests people are getting the basic answer in the AI feature. A sensible response is to make the click worth it. Add a cover checker, a comparison of what policies usually exclude, or clear paths to quotes. Then watch whether clicks and conversions for that page recover over the next few weeks.

Caveats to Keep in Mind

• It’s a test. Features, labels and definitions may change as Google gets feedback. Re-check the help page before drawing long-term conclusions.
• Not everyone has it. If a competitor talks about their numbers and you can’t see yours, it may just be the rollout, not your site.
• Thresholds hide small sites. Low-volume sites may not see the report even once it’s widely available.
• Fresh data can move. The most recent hours or days may be early figures that get revised.
• Impressions aren’t influence. A link counted in an AI answer may have been one of many sources. The report doesn’t measure how big your part was.
• Timing overlaps. The report arrived as the May 2026 core update finished rolling out. Don’t blame every swing in early AI data on AI features alone. Ranking changes from the update affect which pages get cited.
• Opting out removes data. If your site is excluded from generative AI features, it will show no AI impressions and won’t get traffic from those features.

What I Would Do This Week

1. Check whether your properties have the new reports, including any regional or ccTLD (country domain) properties.
2. If they do, export the Pages, Countries and Devices tabs. Save them with a date in the file name.
3. Join the Pages export with Web performance data for the same URLs and period.
4. Pick five pages with high AI visibility and five important pages with none. Review them one by one.
5. Write down any content or technical changes you make, so later data has context.

If you don’t have access yet, prepare anyway. Set up page groups, confirm your canonical URLs are correct, and make sure GA4 landing-page reports are clean. Then the data is useful the day it appears.

Conclusion

The Search Console AI Overviews report finally separates Google’s generative AI visibility from the rest of your search data. It is limited: impressions only, with no clicks or queries. But it’s Google’s own data, split by page, country, device and date. That is more than site owners had before, and I think it is worth using from day one.

Use it to spot which pages Google’s AI features rely on. Compare that with clicks from the Web report, and decide where your content needs to give people a stronger reason to visit. For Google’s own definitions, keep the Generative AI performance report help page bookmarked. A good next step is to pair this data with your analytics using my guide to tracking AI traffic in GA4.