Skip to content
Chalk Labs / AI SEO
How-to guide

How to Use Search Console to Find AI Overview Queries

By Rahil Jain, founder of Chalk LabsUpdated How we research

The short answer

Open the Generative AI performance report in Search Console. It shows impressions from AI Overviews and AI Mode by page, country, device and date. It does not list queries. To find the queries, take the pages with AI impressions and check them in the normal Performance report. Treat the match as a clue, not proof.

  • Since 31 August 2026 every site can get the Generative AI performance report, if it has enough AI impressions.
  • The report shows impressions by page, country, device and date. Google's help page lists no queries and no clicks for it.
  • AI Overview and AI Mode clicks still count in the normal Performance report, mixed in with other web search data.
  • To find queries, filter the normal report by the pages that appear in the AI report.
  • Bing Webmaster Tools adds grounding queries for Copilot, but Microsoft says they are only a sample.

What does Search Console show about AI Overviews and AI Mode?

It shows how often links to your pages appeared in them. Google announced Search Generative AI performance reports in June 2026 1 and says it rolled them out to all websites worldwide as of 31 August 2026 2. The report covers AI Overviews and AI Mode on Search, and has a separate version for Discover 2.

The data is also still counted inside the overall Performance report 1. So you have two places to look, and they answer different questions.

The generative AI report groups data by four dimensions: pages, countries, dates and devices 2. You can filter by search type, either text-based or multimodal, which covers image searches such as Google Lens 2. Dates are in Pacific Time, and you can export both the chart and the table 2.

Do it step by step

0 of 8 done0%

  1. Sign in to Search Console with a property you own or manage, and open the Generative AI performance report 2. If it is missing, the site probably lacks enough impressions, or you may have excluded it from AI features 2.

    Tip: Use a Domain property if you have one. It covers all subdomains in one view.

  2. Choose the longest range available, then pick Web: text-based to start 2. Note the first date that has data. Everything before it is unknown, not zero.

    Tip: Remember that dates are in Pacific Time 2.

  3. Open the Pages tab and sort by impressions. Copy the top 10 to 20 URLs into a sheet. These are the pages Google's AI features choose to link to for your site.

  4. Open Performance, add a Page filter for one URL, and read clicks, impressions, click-through rate and position 5. Record them next to the AI impressions from step 3. Repeat for each URL.

    Tip: Use the same date range in both reports.

  5. With the page filter on, open the Queries tab. Add a Custom (regex) query filter using the patterns in the templates 4. Copy the question-style and long queries into your sheet.

    Tip: If the list looks short, remember the table caps at 1,000 rows and drops anonymized queries 5.

  6. In the AI report, open the Countries and Devices tabs 2. Look for a country or device where AI impressions are high but clicks in the normal report are low. That is where the answer may be doing the visitor's work for them.

  7. Verify your site in Bing Webmaster Tools and open AI Performance 7. Copy the grounding queries and the most-cited pages into the same sheet. Mark them as a sample, because Microsoft says that is what they are.

    Tip: Compare which pages appear on both sides. Those are your strongest assets.

  8. Add one row per page per month using the log template. Choose a single page to improve using the reading table, make the change, and review it next month. Do one thing at a time so you can see what worked.

Your ticks are saved in this browser only.

Copy-ready templates

Regex: question-style queries

regex
^(how|what|why|which|when|where|who|can|does|do|is|are|should|will)\b

In Performance, Add filter, Queries, Custom (regex), Matches regex. The default match is partial and case-insensitive, so ^ anchors the start .

Regex: long, conversational queries (8 or more words)

regex
^(\S+\s+){7,}\S+$

A heuristic for queries that look like full prompts. It is our shortcut, not a Google label. RE2 supports this pattern .

Regex: only pages in one folder

regex
^https://www\.example\.com/blog/

Use with a Page filter set to Custom (regex). Replace the domain with your own.

Monthly AI visibility log

markdown
| Month | Page | AI impressions (GSC) | Clicks (GSC, all web) | CTR | Top question-style queries | Bing citations (sample) | Action taken |
|---|---|---|---|---|---|---|---|
| 2026-11 | /pricing | | | | | | |
| 2026-11 | /guide/x | | | | | | |

Fill one row per page. Keep the same pages each month so trends are comparable.

What does the generative AI report not show?

It does not show queries. Google's help page lists pages, countries, dates and devices as the dimensions, and describes the data as impressions 2. It describes no clicks and no click-through rate for this report. We found no mention of either, so do not assume they exist.

It also leaves out Search Labs experiments, which Google says are still in active development 2. It does not cover ChatGPT, Perplexity, Claude or Copilot. And the usual limits of the Performance report apply, including the 1,000-row table limit 2.

That means the report answers 'which of my pages are being shown in Google's AI answers, where and when'. It does not answer 'what did people type' or 'did they click'. The next sections cover how to get closer to those two answers.

How does Search Console count clicks and positions from AI answers?

A click on an external link in an AI Overview or AI Mode counts as a click, and standard impression rules apply 3. An AI Overview occupies a single position in the results, and every link inside it gets that same position 3. So an AI Overview link cannot be compared with a classic blue link by average position.

AI Mode works differently. Its position follows the same method as a normal results page, and each follow-up question counts as a new query 3. All impressions, positions and clicks in the follow-up answer are credited to that new query 3.

Our reading of this: average position may move without any ranking change, and odd long phrases in query lists may come from follow-up questions.

How do you find the queries behind your AI impressions?

Use the page as the link between the two reports. In the generative AI report, switch to the Pages tab and note the URLs with the most AI impressions 2. In the normal Performance report, add a Page filter for one of those URLs, then open the Queries tab. The queries listed are those that brought that page into any Google web result, AI or classic.

You cannot say which query came from an AI Overview. What you can say is that the page is shown in AI answers and that these are its queries. For pages that earn most of their impressions from AI features, the match is strong. For pages with many classic rankings, it is weak.

Narrow the list with a regex filter. Search Console supports a Custom (regex) filter for queries and pages using RE2 syntax, with partial matching unless you anchor the pattern 4. A filter for questions, or for queries of eight words or more, shows the conversational end of the list. It is a heuristic, not a label from Google.

Know the limits of the list. The table shows at most 1,000 rows, and anonymized rare queries are left out of the table but kept in chart totals, unless you apply a query filter 5. Totals can therefore differ when you filter.

Can GA4 or other tools separate AI Overview traffic?

Not in GA4. Google Analytics defines its AI Assistant channel as visits from sources like ChatGPT, Gemini, Copilot or Grok, and says it excludes Google's AI Overviews and AI Mode 6. Those visits arrive through Google Search, so analytics treats them as normal organic traffic. Search Console is the only first-party place to see them.

For Microsoft's side, Bing Webmaster Tools has an AI Performance report, launched in public preview on 10 February 2026. It shows total citations, average cited pages, grounding queries and page-level citation counts across Copilot and AI summaries in Bing 7. Microsoft says the grounding queries are a sample of overall citation activity, and that citation counts do not show ranking or importance 7.

Paid tools can add a view of ChatGPT and Perplexity. Google warns about tools that promise ranking success or claim to use internal Google metrics, because no third-party tool has access to Google's internal systems 8. Use them if they help your workflow, and check their advice against Google's own guidance 8.

What should you do with what you find?

Match each pattern to one action. The table gives a reading and a first move. These rules are our suggestions, not Google's, and they work best with at least a month of data.

A Semrush walkthrough shows why the comparison is worth making. On its own test site, the first refresh surfaced a guide that AI answers cited for 50 prompts and that earned zero Google clicks 9. That is one vendor's test site, not a benchmark. It does show that being cited and being clicked are different things.

Reading AI impressions against clicks
What you seeLikely meaningFirst move
High AI impressions, normal clicksThe page is shown and still gets visitsProtect it. Refresh facts and keep the answer clear
High AI impressions, very few clicksThe AI answer may cover the questionAdd something the answer cannot give: a template, tool, price or example
Few AI impressions, strong classic rankingThe page may not answer the follow-up questionsAdd sections that answer the related questions
No AI impressions, page not indexedA basic eligibility problem 4Fix indexing first
AI impressions drop suddenlyCould be a feature change, a seasonal dip or a page changeCheck the date against your edits before reacting

What are the common mistakes?

The first mistake is calling impressions 'traffic'. An impression means a link was shown, not that anyone visited. The second is reading average position in AI answers like a rank, when every link in an AI Overview shares one position 3.

The third is judging a page too early. Google says the newest data can be preliminary and may change in the next few hours 2. Give each change at least four weeks of data before you decide it worked.

The fourth is chasing every fan-out question with a new page. Google's guide says making separate content for every variation of how people search, mainly to manipulate AI responses, violates its scaled content abuse policy 10. Improve the existing page that answers the group of questions.

When is this not worth your time?

Skip it if your site gets only a handful of Google impressions a month. The report may not appear, and any pattern will be noise. Spend the time on getting pages indexed and answering real customer questions first.

Also skip a weekly routine. A monthly look is enough for most small teams, because the data is slow to move and easy to over-read.

Frequently asked questions

Can I see exactly which queries triggered an AI Overview for my site?

Not directly. The generative AI report groups impressions by page, country, device and date, and does not list queries 2. You can filter the normal Performance report by the same pages and read their queries, but that mixes AI and classic results.

Do clicks from AI Overviews show up in Search Console?

Yes. Google says a click on an external link in an AI Overview counts as a click, and AI Mode clicks count too 3. They sit inside the normal web search data, not in a separate AI clicks column that we could find in Google's help pages.

Why is my report empty or missing?

Google says the report usually does not appear when a site has not received enough impressions in generative AI features, or when the site was excluded from them 2. Check your settings, and make sure your key pages are indexed and snippet-eligible 4.

Does Search Console cover ChatGPT or Perplexity?

No. The report covers Google's AI Overviews and AI Mode only 2. For other assistants you need your own analytics referrals, server logs, or a paid tracking tool. Microsoft's Bing Webmaster Tools covers Copilot and Bing AI summaries 7.

Is the Search Console AI data enough to judge my AI SEO?

It is a good start for Google, not the whole picture. It shows impressions, not wording, not sentiment, and nothing from other assistants. Pair it with a small tracked prompt list and Bing's report, and treat each as one partial view.

About the author

Rahil Jain

Rahil Jain is an AI SEO (answer engine and generative engine optimisation) consultant and co-founder of Chalk Labs.

LinkedIn · Method and credentials

Sources

  1. Introducing Search Generative AI performance reports in Search Console · Google Search Central BlogOpened October 20, 2026. Supports: Announcement in June 2026; data also stays in the overall Performance report.
  2. Generative AI performance report (Search) · Google Search Console HelpOpened October 20, 2026. Supports: What the report includes, its dimensions, search types, limits, preliminary data, missing-report reasons, export and the Labs exclusion.
  3. What are impressions, position, and clicks? · Google Search Console HelpOpened October 20, 2026. Supports: How clicks, impressions and position are counted for AI Overviews and AI Mode, including the single position and follow-up questions.
  4. Performance report (Search results): Advanced filtering and comparison · Google Search Console HelpOpened October 20, 2026. Supports: Custom regex filters for queries and pages; RE2 syntax; partial matching; case-insensitivity.
  5. Performance report (Search results): Troubleshooting data discrepancies · Google Search Console HelpOpened October 20, 2026. Supports: 1,000-row table limit and omission of anonymized queries from the table.
  6. [GA4] Default channel group · Google Analytics HelpOpened October 20, 2026. Supports: GA4's AI Assistant channel excludes Google's AI Overviews and AI Mode.
  7. Introducing AI Performance in Bing Webmaster Tools Public Preview · Microsoft Bing Webmaster BlogOpened October 20, 2026. Supports: Launch on 10 February 2026; metrics offered; grounding queries are a sample; counts do not show ranking.
  8. Optimizing your website for generative AI features on Google Search · Google Search CentralOpened October 20, 2026. Supports: Measure with the Generative AI performance report; caution about third-party tools; eligibility requirements.
  9. How to build an AI visibility dashboard with Semrush · Semrush BlogOpened October 20, 2026. Supports: On the author's test site, a guide cited for 50 AI prompts earned zero Google clicks.
  10. Optimizing your website for generative AI features on Google Search (scaled content abuse note) · Google Search CentralOpened October 20, 2026. Supports: Creating separate content for every variation or fan-out query, mainly to manipulate responses, violates the scaled content abuse policy.
How this page was made. Built on Rahil Jain's AI SEO method and drafted with AI research assistance. Every statistic links to the page we read it on (opened October 20, 2026 or nearby). Published October 20, 2026; last updated October 20, 2026. Chalk Labs sells AI SEO services, so we have a commercial interest in this topic. Spotted an error? Tell us and we will fix it.