Google Search Console now shows your AI overviews and AI mode impressions

What Google actually launched

On 3 June 2026, Google announced new Search Generative AI performance reports in Search Console, with separate views for Search and for Discover. The stated aim is to help site owners understand their visibility inside generative AI features on Search, such as AI Overviews and AI Mode. You can read Google's full announcement on the Search Central blog.

The report carries a Beta label, and Google has been clear that it is still deciding which metrics to add. That matters for how much weight you put on it. This is the first official window into AI visibility, not a finished analytics product.

For two years, the only honest answer to "are we showing up in AI answers" was that we could test and estimate it. Now there is a number that comes from Google itself.

Where to find it in Search Console

Open Search Console, go to Performance, and look for Generative AI in the left-hand menu. It sits alongside your usual search results report, and Discover has its own Generative AI entry underneath.

Source: Google Search Central blog, licensed under CC BY 4.0.

If it is not there, nothing is broken. The rollout has been staged, so some properties have it while others do not, and it is common to see it on one site in your account and not another. Data also only starts accumulating from the point Google begins collecting it for your property, so early graphs can look sparse.

One more thing worth knowing before you panic about double counting: this data was already flowing into your overall performance report. The new view isolates it, so it does not add new traffic.

What the report shows

The Search version of the report covers impressions from AI Overviews and AI Mode. Google documents the full scope of the report in its Search Console help centre.

You can break the data down by:

• Pages, so you can see which URLs are being surfaced.

• Countries, which is useful if you run a bilingual site.

• Devices.

• Days, so you can spot when something changed.

Experiments in Search Labs are excluded, because they are still in active development. Generative AI features in Discover are reported separately.

An impression here means a link to your site was shown to someone inside a generative AI feature. It is a visibility signal, nothing more.

What it does not show, and why that matters

This is the part most coverage skims over. The report currently leaves out:

• Queries.

• Clicks.

• Click-through rate.

• Average position.

• Where in the answer your citation appeared.

• Which passage of your page was used.

• Anything about conversions or revenue.

So you can see that Google showed your page inside an AI answer, but not what people asked to trigger it, and not whether anyone clicked through. Google has said it is working with site owners on which additional metrics would be useful, so some of this may arrive later.

Be careful with the conclusions you draw. A large impressions number is not proof of business value, and a small one is not proof that AI search is ignoring you. Treat it as one input, not a verdict.

How to read your numbers without fooling yourself

The most useful reading is relative, not absolute.

Start with the Pages tab and ask which types of page earn impressions. In our experience, practical explainer content and clear service pages tend to surface more often than general brand pages, though that will vary by sector. Look for the pattern rather than the total.

Then check whether the pattern matches your commercial priorities. If your blog is pulling most of the AI visibility while your service pages get none, that is a content gap worth a conversation.

Finally, watch the trend line alongside your normal performance report. If AI impressions rise while clicks stay flat, that is worth investigating rather than celebrating. Do not report AI impressions to a client or a board as though they were traffic.

What to do about it on a Framer or Webflow site

There is no lever that guarantees a citation, and anyone selling you one is guessing. What you can do is remove the reasons a page gets skipped. On the Framer and Webflow builds we work on, that usually means:

• Answering the question directly, near the top of the page, in plain language, before the background and the sales pitch.

• Using genuine question-led headings that match how people actually ask, rather than clever ones.

• Keeping structured data accurate and current, so the page is easy to parse.

• Making sure pages are fast, indexable and not hidden behind interactions or scripts.

• Reviewing older posts, because outdated pages are easy to pass over.

• Writing content that says something specific. Generic advice has nothing to quote.

None of that is new. It is the same discipline that has always made content easy to understand, now measured on a new surface.

Frequently asked questions