Search Console doesn't separate clicks from AI Overviews, but there's a way to measure impact

Google Search Central documentation shows that pages shown in AI Overviews are counted together with traditional results, under the Web search type in Search Console. Without dedicated segmentation, measuring the impact of generative search requires cross-referencing indirect signals.

Google Search Central documentation shows that pages shown in AI Overviews are counted together with traditional results, under the Web search type in Search Console. Without dedicated segmentation, measuring the impact of generative search requires cross-referencing indirect signals.

What Google's documentation says, and what it doesn't say

After Google began showing AI-generated answers at the top of part of its search results, many people in technical SEO started looking for a way to isolate, within Search Console, how much traffic comes specifically from these summaries. The common expectation is a dedicated "search appearance type" filter for AI Overviews. The official Google Search Central documentation, "AI features and your website," updated on December 10, 2025, does not confirm this feature.

In short: according to the text itself, sites that appear within AI Overviews and AI Mode have clicks and impressions added to Search Console's general search traffic, under the "Web" search type in the performance report. There is no separate appearance type to distinguish what comes from an AI-generated summary from what comes from a traditional result.

This changes the angle of any generative search visibility analysis: you can't apply a filter and read "clicks from AI Overviews" in a ready-made report. The work of measuring becomes more like interpreting a pattern than reading a direct number.

How these clicks are counted today in the performance report

In the Search Console performance report, the search type filter has options such as Web, Image, Video, and, in some cases, Shopping and News. AI Overviews and AI Mode didn't get their own option in this list: they fall under "Web," mixed in with traditional text results.

This has a direct consequence for anyone trying to justify investment in GEO (generative engine optimization): a spike in impressions without a proportional increase in clicks can be a sign that the page is appearing as a supporting source within an AI-generated summary, but the report doesn't confirm this with a single filter click. It's pattern reading, not label reading.

Step-by-step guide to isolating generative search signals without a dedicated filter

Since there's no native segmentation, approximating a GEO diagnosis in today's Search Console depends on combining indirect signals. A reasonable roadmap for a technical SEO team to follow:

1. Open the performance report and filter by "Web" search type (this is where all AI Overviews traffic is included, together with traditional results).
2. Compare the impressions curve with the clicks curve per page over time. A jump in impressions without an equivalent jump in clicks is the most common pattern associated with "zero-click" features, such as AI Overviews.
3. Cross-reference this breakdown with engagement data in Google Analytics (time on page, scroll rate, pages per session), since the documentation associates clicks on pages with AI Overviews with longer sessions.
4. Use URL Inspection to confirm that the page is indexable and free of blocks from robots.txt, nosnippet, or noindex, since these are the only technical requirements cited for appearing as a supporting link in an AI summary.
5. Make sure the page's structured data matches the visible text on the page, one of the SEO fundamentals cited in the documentation. Tools such as the Enhancements report and the Rich Results Test can help with this check, even though they aren't specifically mentioned by Google in this context.

None of these steps delivers an isolated number for "AI Overviews clicks." They deliver circumstantial evidence, which is quite different from a ready-made filter in the report.

The trade-off the documentation itself admits

It's worth reading how Google describes the quality of these clicks, even without being able to isolate them:

We've seen that when people click from search results pages with AI Overviews, these clicks are higher quality (meaning, users are more likely to spend more time on the site).

(translation) We've found that when users click on search results pages that include AI Overviews, these clicks tend to be higher quality, meaning users tend to spend more time on the site.

(author) Google Search Central, official Google documentation

It's a strong statement about traffic quality, but without the numerical detail (how much more time, for which type of query) that would allow turning this into a measurable goal. It's evidence in favor of investing in GEO, not a metric you track on a dashboard.

Structured data: a basic requirement, not a trigger for specific optimization

A direct "before and after" performance comparison of structured data adjustments, filtered only for pages that appear in AI Overviews, isn't possible today through native Search Console, because that filter simply doesn't exist. What the documentation guarantees is more modest: structured data that matches the visible text on the page falls under the same list of SEO best practices, with no additional specific technical requirement for appearing in AI features.

This means that adjusting a page's schema markup doesn't change the game on its own. What matters is making sure that schema doesn't lie, because the text itself cites the need to ensure structured data matches the visible text on the page, as an item on a list of fundamentals that already applied before AI features existed in search.

Measuring the impact of this falls back on the same indirect signals from the earlier step-by-step guide: comparing impressions and clicks under the "Web" type before and after fixing the schema, and cross-referencing with session time in Analytics.

How to control the presence (or absence) of content in these features

For those who want to limit or prevent a page from being summarized within AI Overviews or AI Mode, the documentation points to the same directives as always: nosnippet, data-nosnippet, max-snippet, and noindex. There's no exclusive tag for AI search features; the control is the same one that already existed for snippets in traditional results.

To limit the use of content for training other Google AI systems, outside of search, the way to do it is Google-Extended, declared in robots.txt. These are different mechanisms, with different purposes, and it's worth not confusing one with the other when writing the site's crawling policy.

What remains open

Google's own documentation admits that AI Overviews "don't appear frequently," which already limits the volume of data available for any analysis, even if Search Console opened a dedicated filter tomorrow. Adding to this is the absence of any sign that such a filter is on the product roadmap: the text treats inclusion within "Web" as the current measurement model, not as a temporary limitation.

For technical SEO teams, the practical consequence is treating GEO as a reading of trend and engagement quality, not as a KPI that can be isolated by a filter. Third-party tools that try to estimate presence in AI Overviews via search results scraping fill part of this gap, but none of them has access to the same data Google uses internally to generate the performance report.

Source: Google Search Central, "AI features and your website" (official documentation), https://developers.google.com/search/docs/appearance/ai-features