By Mustehsan Sheikh
To measure AI search traffic, you need to start from what Google actually hands you, not from what your dashboard seems to promise. Search Console now shows how often your pages appear inside its AI answers. It does not show how many of those appearances turned into visits. That gap shapes every decision in this post.
TL;DR
Search Console's Generative AI report counts impressions in AI Overviews and AI Mode.
No clicks, rate, position, or queries come with it.
Its impressions already sit inside your normal Web totals, so never add the two reports together.
Compare page-level clicks against AI impressions over the same dates, as correlation, not proof.
Pew found users clicked a normal result in 8% of visits with an AI summary, against 15% without one.
This post gives you a routine you can run every month, drawn from our search engine
optimization work for clients who need to report on AI visibility without inventing numbers.
What the report counts, and what it leaves out
Google announced the Generative AI performance reports on June 3, 2026, and rolled them out to all websites worldwide on August 31, 2026. The Search report covers AI Overviews and AI Mode. A separate report covers generative AI features in Discover. If you cannot find the Search report, the usual reasons are plain: your site has had too few impressions in AI features to fill it, or your site is excluded from those features in Settings.
One impression means a link to your site was shown to a user inside a generative AI feature. You can split that count by page, country, device, and date. You cannot split it by query, and you cannot split AI Overviews from AI Mode. Every link inside one AI answer shares that answer's single position, so position data would mislead even if Google gave it to you.
That is the whole metric set. There are no clicks, no click-through rate, and no average position in this report. Google has said it may add more metrics after hearing from site owners, but nothing has arrived yet. If your reporting template has columns for those numbers, they stay empty. Filling them with estimates and labelling them as Search Console data would be wrong, and it would poison every later comparison.
For a site owner, this report still answers a question that had no official answer before: are my pages being cited at all? A page cannot send a click from an AI answer it never appears in. That makes the report a starting point for measurement, not the measurement itself. Our search engine optimization work treats it exactly that way: a citation check first, then a traffic check from other data.
Why there is no AI click-through rate
A click-through rate needs a numerator and a denominator from the same set of events. Here the two numbers live in different tables with no shared key. The Generative AI report gives impressions for AI surfaces only. The classic Performance report gives clicks for all of Search, with AI clicks pooled inside the Web search type. Dividing one by the other produces a ratio, but the ratio describes nothing real, because the clicks in the numerator did not all come from the impressions in the denominator.
The same pooling happens in analytics. A visitor who clicks a link inside an AI Mode or AI Overview answer arrives from google.com, so GA4 groups that visit with normal Organic Search traffic. GA4 alone cannot split AI clicks from classic blue-link clicks. Any vendor report that claims a precise AI click-through rate from these two sources is estimating, not measuring. Ask what sits behind the number before you put it in a client report.
Position adds a second distortion. In the classic report, an AI Overview occupies one position, and every link inside it inherits that position. A page that appears only as a citation inside an answer can therefore show a strong average position while sending almost no visits. Read 
position next to clicks, never on its own, for any page with AI impressions.
position next to clicks, never on its own, for any page with AI impressions.Three readings that mislead
The first misreading is double counting. The AI impressions in the Generative AI report are already inside the classic Performance report's Web totals. The new report is a filtered view of data you already had, not a second pile of traffic. Add the two together and you count the same appearance twice. Treat the AI number as a share of your Search total, and say so in any chart that shows both.
The second misreading is treating an impression as a visit. An impression means a link was shown. It says nothing about whether the user noticed it, read it, or clicked it. Impressions can rise while clicks stay flat, and that is not a contradiction. It is the normal shape of a surface where most people read the answer and stop. Report impressions and clicks side by side, with different labels, and never merge them into one "AI traffic" line.
The third misreading is blaming a flat click-through rate on AI without a page-level check. Click-through rate moves for many reasons: a title change, a new competitor result, a seasonal query shift, a snippet that now answers the question in full. Before you attribute a drop to AI answers, check the page. Did its AI impressions rise over the same dates? Did its classic clicks fall on the same pages, or across the whole site? If the two movements do not line up page by page, the AI explanation is unproven, however tempting it feels.
A routine that gives you an answer
Start with eligibility and citations. Open the Generative AI report, set the date range to the last three months so a smaller site has enough data to read, and sort the Pages view by impressions. Write down your top pages. If the report is missing, check Settings, then Search generative AI: an excluded property gets no impressions and no traffic from these features, and Google says the setting is not used as a ranking signal elsewhere. A missing report on an included site usually means too few AI impressions so far, not a penalty.
Next, compare page by page. For each of your top cited pages, pull its clicks and click-through rate from the classic Performance report over the same dates, and watch the trend rather than one week's number. A page whose AI impressions climb while its clicks hold steady is being seen more and visited at the same rate. A page whose AI impressions climb while its clicks fall is the one to investigate first. Keep the language honest in your notes: this is correlation between two reports, not proof that an AI appearance caused a lost click.
Then turn to GA4 for the visit side. Filter the landing page report to Session default channel group equals Organic Search, and look at the same pages over the same dates. You are watching for movement in sessions, engagement, and conversions on pages whose AI impressions rose. Where GA4 shows no movement, say so. The honest sentence in a report is often "AI impressions rose and nothing else moved," and that sentence is worth more than a modelled click estimate dressed up as fact.
Finally, check citations yourself. Search your most important queries and look at what the AI answer actually cites and where your link sits: inline in the text, in a card, or in a carousel. Placement decides whether a click is plausible. A link buried at the end of a long answer behaves differently from one placed beside the sentence that answers the question. Note the placement next to the impression count, because the report itself never shows it.
One context note belongs here. On October 15, 2026, Google announced more and higher inline links in AI Mode and AI Overviews, a test slot for one or two official pages on task-based queries, expanded news carousels and Preferred Sources, and thumbnails on related links. Google shared no click data in that announcement either. More visible links are not more traffic until the clicks show up, so treat mid-October as a new baseline for your page-level comparisons rather than as a win.
What outside studies add
Search Console tells you about your own site. Outside studies tell you how users behave across many sites, and one study sets the stakes for this whole routine. Pew Research Center tracked 900 US adults through March 2025, covering 68,879 Google searches, of which 12,593 produced an AI summary. Users clicked a traditional result in 8% of visits when an AI summary appeared, against 15% without one. Links inside the summary itself were clicked in only 1% of visits to pages with a summary.
Read those numbers with their limits attached. The data comes from March 2025, before AI Overviews reached their current share of results, and from a US panel, so it does not describe every market. Google publicly disputed Pew's methodology, so attribute the finding to Pew, never to Google. And do not present 8% or 1% as an AI click-through rate. Pew measured visits and clicks across its panel; Search Console still gives you no AI click rate of your own. The study supports the shape of the problem, which is that summaries change click behaviour. It does not replace your own page-level comparison.
What to change once you know
The routine above sorts your pages into three groups, and each group gets a different response. Pages with rising AI impressions and steady clicks are doing their job: keep them current, because a cited page that goes stale loses its citation to a fresher source. Pages with rising AI impressions and falling clicks need a closer look at the page itself. Does the page answer the question fully in its first screen, leaving no reason to click? Does it offer a clear next step the answer cannot give, such as a calculator, a template, a price, or a booking step? Strengthen the reason to visit rather than rewriting the whole page.
Pages with no AI impressions at all need a different check. Confirm the site is included in Settings first. Then look at the page the way a citation picker would: does it state a direct answer early, back it with specifics, and cover the follow-up questions a user would ask next? Pages that bury the answer under a long introduction rarely get cited, however good the rest of the content is. This is also where our Ecovers project is a useful reference: a custom website built for search and AI answers from scratch, with a blog admin panel and automation behind it, where the structure of each page does the citing work instead of tricks.
When the data is thin, resist the urge to buy a tool that promises the missing number. No third-party tool can see Google's click split either. What a tool can do is track which answers cite you across more queries than you would check by hand, and that is worth paying for only once your own Search Console routine runs every month without fail. Reporting comes before rewriting. A team that knows which pages are cited, which are clicked, and which are neither will spend its next content budget in the right place.
Is your reporting showing AI visibility, or AI results?
If your monthly report still treats AI impressions as traffic, the next rewrite will chase the wrong number. We set up SEO and AEO reporting that separates what AI answers show from what they send, then build content around the pages that actually move.
See how we set up SEO and AEO reporting
Frequently asked questions
Does the Search Console Generative AI report show clicks?
No. The report shows impressions only, split by page, country, device, and date. Clicks from AI Overviews and AI Mode are counted in the classic Performance report, pooled inside the Web search type with every other click. There is no filter that separates them out, so no AI click-through rate can be calculated from Search Console data today.
Why does my Generative AI report not appear?
Google lists two usual causes. Your site may have had too few impressions in AI features to fill the report, which is common for smaller sites. Or your site may be excluded from generative AI features under Settings, then Search generative AI. An excluded site gets no impressions and no traffic from those features. Check the setting before assuming a technical fault.
Can I add Generative AI impressions to my normal search impressions?
No, because that counts the same appearances twice. The AI impressions are already inside the classic Performance report's Web totals. The Generative AI report is a filtered view of that data, not extra traffic. Report the AI number as a share of your Search total, and label both numbers clearly in any chart.
How do I track AI Mode traffic in GA4?
You cannot isolate it directly. Visitors who click from AI Mode or an AI Overview arrive from google.com, so GA4 places them in Organic Search with classic search visits. The workable routine is indirect: take the pages with rising AI impressions from Search Console, then watch those same landing pages in GA4 for movement in sessions and conversions over the same dates.
Does opting out of AI features hurt my normal rankings?
Google says the Search generative AI control is not used as a ranking signal for results outside those features. An excluded site keeps its normal listings but gets no impressions or traffic from AI Overviews, AI Mode, or AI features in Discover. Check your Generative AI report before switching, since opting out also removes the citation visibility you were measuring.
Do AI Overviews and AI Mode report separately?
No. The Search report covers both together, with no split between them. Discover has its own separate Generative AI report. If you need to know which surface cited a page, you have to check the answers yourself for your most important queries, since neither report breaks the two apart.
More from Buzzin Tech Hub
Google ranking drop diagnosis: check this before you rewrite
Google Discover vs Search traffic: diagnose by surface
BuzzinTech blog