Search Console and AI Overviews: What the Data Shows (and Hides)

Search Console shows AI Overviews and AI Mode traffic, but mixed into the Performance report's "Web" search type, with no separate filter. To estimate impact, group queries by intent, filter question-style queries with regex, compare click-through rate at stable position across periods, and check a sample of results pages by hand. Our view: treat the output as an estimate with clear caveats.

Facts are from Google documentation accessed 30 Oct 2026. Points marked "Our view" are interpretation; examples are illustrative.

On this page

• What does the data show and hide?
• Step 1: How do you export queries?
• Step 2: How do you group queries?
• Step 3: Which regex filters help?
• Step 4: How do you check CTR at stable position?
• Step 5: How do you validate with manual checks?
• Step 6: How do you build the monthly sheet?
• Step 7: How do you pair it with analytics?
• What should you tell stakeholders?
• Which pitfalls should you watch for?
• What can you do when CTR falls?
• How do you handle seasonality and other changes?
• How do you choose which pages to act on?
• Exercise
• Checklist
• From MentionX data: what Search Console can't show
• Worked example: Ledgerline reads its Search Console data
• How do you set up the export?
• How do you build query groups?
• What does CTR at stable position tell you?
• How do you explain this to stakeholders?
• What else should you track?
• How do you pick pages for the next sprint?
• How often should you review the data?
• Which regex filters are worth saving?
• How do you account for seasonality?
• Common questions
• Related
• Sources

Checked on 31 October 2026. "Our view" marks MentionX opinion. Ledgerline is a fictional company used for illustration.

What does the data show and hide?

Google says AI feature traffic is included in the Performance report under the "Web" search type (Google).

Step 1: How do you export queries?

Open Performance, Search results, set search type to Web and pick the last three months. Export queries with clicks, impressions, CTR and position. Our view: also export pages, so you can see which URLs are affected.

Step 2: How do you group queries?

Add a column for intent in a spreadsheet:

Step 3: Which regex filters help?

In the Performance report's query filter, choose custom regex. Examples: ^(how|what|why|which|can|should|is|does)\b for questions and (\S+\s+){6,} for longer queries. Our view: save these as a checklist so the same filters run every month.

Step 4: How do you check CTR at stable position?

Use date comparison, for example this month versus the same month last year. For each group, compare CTR where average position changed little. Our view: falling CTR at stable position on question queries is consistent with answers being satisfied on the results page, which may include AI Overviews. It isn't proof; seasonality and other SERP features also change CTR.

Step 5: How do you validate with manual checks?

Search a sample of 20 queries per group in a clean browser session. Record whether an AI Overview appears and whether your page is linked in it. Our view: note the date and location; results vary.

Step 6: How do you build the monthly sheet?

Step 7: How do you pair it with analytics?

Google suggests looking at conversions and engagement in tools such as Google Analytics, not just clicks (Google). Google also describes clicks from pages with AI Overviews as "higher quality" (Google). Our view: compare conversion rate for landing pages hit by question queries before and after.

What should you tell stakeholders?

• AI features are included but not separable.
• Which query groups show falling CTR at stable position.
• What the manual sample found.
• What you're doing about it.

Which pitfalls should you watch for?

• Reading position changes as AI Overview effects without a sample check.
• Comparing months with different seasonality.
• Mixing branded and question queries.
• Ignoring Discover and News, which are separate search types.

What can you do when CTR falls?

Our view: check whether your page is linked in the AI Overview for those queries. If not, compare the linked pages with yours: are they more specific, more current, or more original? Google says unique, non-commodity content likely matters most (Google). If you are linked, focus on what happens after the click: clear next steps and useful depth.

How do you handle seasonality and other changes?

Our view: always compare against the same month a year earlier where you have the data, not only the previous month. Note anything else that changed: site migrations, title rewrites, new SERP features, algorithm updates listed on Google's Search Status Dashboard, or your own content changes. Keep a simple change log beside the monthly sheet. When CTR moves, check the log first. If several changes happened at once, say so in the report rather than attributing the whole effect to AI Overviews.

How do you choose which pages to act on?

Our view: rank pages by clicks lost at stable position in the question group, then by business value. A guide that feeds sign-ups matters more than a definition page with high traffic but little conversion. Start with five pages and recheck them after two months.

Exercise

Export three months of Web queries, group them into question, commercial and branded, and calculate CTR change at stable position for each. Check ten question queries by hand for AI Overviews.

Checklist

• Web search type selected
• Queries and pages exported
• Intent groups defined
• Regex filters saved
• CTR compared at stable position
• Manual SERP sample recorded

From MentionX data: what Search Console can't show

Method. Read-only aggregate queries over MentionX scans from 22 August to 29 October 2026 (UTC): 2,217 engine runs (2,154 completed) across 18 anonymised projects and six engines.

Search Console reports Google traffic only, and doesn't separate AI features. It can't show how often other engines mention you. In our sample, mention rates varied by engine:

And by prompt intent:

Our view: AI Mode's mention rate sat in the middle of the engines we tracked. Search Console can help you infer how AI features affect clicks, but you need separate tracking to know whether Google's AI features, or any other engine, mention your brand.

Limits. A convenience sample of mostly B2B software projects, run through engine APIs, which can behave differently from the consumer apps. Perplexity, Copilot and DeepSeek figures come from a single day (26 October 2026), so treat them as a snapshot, not a trend.

Worked example: Ledgerline reads its Search Console data

Ledgerline is a fictional accounts-payable tool. Figures are illustrative.

Ledgerline notices clicks on its guides falling while impressions hold steady. It exports queries for the last six months and groups them into how-to, definition, comparison and brand queries.

For definition and how-to queries where its average position is stable, CTR has dropped. The team manually checks a sample of those queries and sees AI Overviews on most of them. Brand and comparison queries show little change. Ledgerline reports that AI features probably reduced clicks on informational queries, checks whether its pages are cited in those overviews and focuses its next content work on comparison pages and on making its guides more citable.

How do you set up the export?

Use the Performance report with the Web search type. Export queries and pages for a period before and after the change you're investigating. For larger sites, use the API or a connector to get more rows than the interface shows. Keep the exports in a folder with dates, so you can rebuild the analysis later.

How do you build query groups?

Start with simple rules: queries containing "what is" or "meaning" are definitions; "how to" are how-to; "vs" or "alternative" are comparisons; your brand name marks brand queries. Use regex filters where possible. Then review a sample of each group by hand and adjust the rules. Groups make trends clearer than individual queries.

What does CTR at stable position tell you?

If a query's average position stays roughly the same but CTR falls, something else on the results page is taking clicks. AI Overviews are one likely cause, but not the only one: ads, other features and changes in searcher behaviour can all play a part. Treat the result as evidence, not proof, and confirm with manual checks.

How do you explain this to stakeholders?

Say clearly what the data shows and what it doesn't. Search Console shows clicks and impressions from Google, including AI features, but not which feature produced them. It doesn't show mentions in AI answers or other engines. Present your inference, your manual evidence and your plan. Our view: honest framing builds more trust than claiming precise AI-traffic numbers you can't measure.

What else should you track?

Track brand mentions and citations in AI answers separately, across Google's AI features and other engines. Watch branded search volume, direct traffic and conversions, which can rise even when clicks on informational queries fall. Together, these show whether visibility is shifting rather than disappearing.

How do you pick pages for the next sprint?

Prioritise pages with falling clicks on queries where buyers are close to a decision, such as comparisons and product questions. For informational pages, focus on being cited in AI answers and on capturing visitors who do click with clear next steps.

How often should you review the data?

Monthly is enough for most teams, with a deeper review each quarter. Note any major Google updates or launches in your log, so you can connect changes in the data to events. Keep the query groups stable between reviews so trends stay comparable.

Which regex filters are worth saving?

Save filters for question words such as what, how, why and which, for comparison terms such as vs, versus, alternative and compare, and for your brand and product names. Saved filters make monthly exports faster and keep groups consistent between reviews. Test each filter on a sample to make sure it doesn't catch unrelated queries.

How do you account for seasonality?

Compare the same period year on year where you can, as well as month on month. Many categories have predictable peaks and dips. Note launches, campaigns, site changes and Google updates in the log, so a drop caused by seasonality isn't blamed on AI features. Where year-on-year data isn't available, use a set of control queries without AI features as a rough comparison.

Keep the export settings the same each month: the same date range length, the same filters and the same country and device splits. Changing settings between exports makes trends impossible to read, and it is the most common reason two people get different answers from the same property.

Common questions

Can I filter AI Overviews in Search Console? No. AI feature traffic is included in Web search without a separate filter.

Does a falling CTR always mean AI Overviews? No. Ads, other features and seasonality can also cause it.

Should I track other engines too? Yes. Search Console covers Google only.

Related

• Forecasting traffic loss from AI Overviews

Sources

• Google, AI features and your website; Succeeding in AI search
• Search Console Help, Performance report metrics

All accessed 30 Oct 2026.