Yes, partly. Google Search Console now has a Generative AI performance report showing impressions for your client's pages in AI Overviews and AI Mode combined, by page, country, device and date. It shows no clicks and no split between the two features, so report visibility, not traffic.
That limit is the whole job. Most of what gets written about this subject is a news recap of the launch; a client does not need a recap, they need a page of their monthly report that says what was measured, what moved, and what the agency did about it. That page can be built line by line, with set wording for the months when a number is missing.
What the Google report contains, and what it leaves out
Google announced the Search Generative AI performance reports on 3 June 2026 and, according to the Search Console help page, rolled them out to all websites worldwide by 31 August 2026. The report counts how many times links to the client's site were shown in a generative AI feature on Google Search. Google's help page lists AI Overviews and AI Mode as the features included, and says the list may grow over time. There is a separate report for generative AI features in Discover, which is a different surface and should get its own line.
The ways to slice the data are four: pages, countries, devices (desktop, tablet or mobile) and dates, with hourly, daily, weekly and monthly granularity according to the announcement. You can filter by search type: web text-based searches, or web multimodal searches, meaning image-led ones such as Lens. There is an export button for the chart and table data.
Now the gaps, which matter more to an agency than the features. The report as documented offers impressions as its measure. It does not list clicks, click-through rate, average position or a query table. It does not split AI Overviews from AI Mode. Clicks from AI features are not lost, though: Google's AI features guidance says sites appearing there are counted in the overall Search Console data, reported in the ordinary Performance report under the Web search type. They are simply not separable from every other click.
Three smaller points catch people out. The chart counts a property once per search, so if two of the client's pages appeared in the same AI answer, the chart total counts one impression, while the page table counts each page. Chart and table totals will therefore differ, and Google says as much in its discrepancy note. The table has the same 1,000-row limit as the standard Performance report. And dates run on Pacific Time, so a monthly figure will not line up exactly with a client's own calendar-day analytics.
The Complete Excel Course with AI Integration
From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.
By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner
Buy nowA filled-in example: one client, one month
An illustration with invented numbers, for a bookkeeping practice that is the agency's client. In the Generative AI performance report for the month, set to Web: text-based, the property shows 41,200 impressions. The standard Performance report, same month, same search type, shows 212,000 impressions and 5,830 clicks. Last month the AI report showed 33,900.
The first instinct is to write "AI Overviews account for 19% of our visibility". Do not. The two reports count differently (one impression per property per search in the chart, with the AI report limited to AI features), and the AI report may overlap with the classic one, since Google says the data is also included in the overall Performance report. A safe sentence is: "Links to your site were shown 41,200 times in Google's AI features this month, up 21.5% on last month's 33,900. Clicks from these features cannot be isolated, so we track the clicks that matter through the pages involved."
The percentage is a quick sum: 41,200 minus 33,900 is 7,300, and 7,300 divided by 33,900 is 0.2153. Showing that sum in your working notes, not on the client page, saves an argument later.
Next, the page table. Say the top three pages by AI impressions are a guide to year-end deadlines (14,900), a page on sole trader record-keeping (9,200) and the services page (1,100). Those pages tell you something the totals cannot: the AI features are quoting informational pages, not the page that sells. That becomes an action, not just a number: add a clear, linked next step on the two guide pages, then check next month whether visits to the services page from those guides rose in the client's analytics.
The reporting page, line by line
One page, six lines, the same order every month. A client who sees the same layout for a year learns to read it.
| Line | Source | What you say | What you never say |
|---|---|---|---|
| 1. Google AI impressions | Generative AI performance report, Web text-based | Times links were shown in AI Overviews and AI Mode combined, with the month-on-month change | "Traffic from AI Overviews" or any split between the two features |
| 2. Top pages in AI features | Same report, Pages tab, top five | Which pages Google's AI features linked, and the role each page plays | That these pages "rank" first in AI answers |
| 3. Countries and devices | Same report, Countries and Devices tabs | Only where it changes a decision, such as a mobile-heavy audience | Percent shares from a table that is cut at 1,000 rows without checking |
| 4. Clicks and conversions | Performance report plus the client's own analytics | Total search clicks, then what visitors did on the pages named in line 2 | A click figure labelled as AI-driven |
| 5. Other AI surfaces | Bing Webmaster Tools AI Performance, if verified | Citations and grounding queries, as a separate line | A sum of Google impressions and Bing citations |
| 6. What we changed | Agency work log | The two or three edits made to pages in line 2, and the date of each | A cause for a rise unless the dates support it |
Notice the pairing in lines 2 and 4. The AI report tells you which pages were shown; the analytics tell you what happened after. Joining them by page URL is the one piece of analysis here that is genuinely yours rather than Google's.
Setting up the monthly export in ten minutes
Do this once per client property and then repeat only the export.
- Open the client's property in Search Console and find the Generative AI performance report. If it is not listed, jump to the section on missing reports below.
- Set the search type to Web: text-based, the date range to the last full calendar month, and the granularity to monthly. Add a second view for the previous month so the comparison comes from the same filters.
- Open the Pages tab, export, and save the file with the property name, the month and "ai-pages" in the file name. Repeat for Countries and Devices only if line 3 will be used.
- In the standard Performance report, filter to the same pages and export clicks and impressions for the same period.
- Paste both exports into the agency's reporting sheet, one tab each, and keep the date range in a cell at the top of each tab.
If the client has many pages, remember the 1,000-row limit. A property with thousands of URLs will have its tail cut off, so say "top 1,000 pages" in the notes tab rather than implying the table is complete.
Using an AI assistant to draft the commentary, not the numbers
An assistant is good at turning a pasted table into three plain sentences. It is bad at arithmetic you have not checked and it will happily describe a rise as a cause. Give it the numbers, tell it what it may not claim, and read the result with the report open beside it. If you want the wider reporting workflow, How Marketing Agencies Use AI to Automate Client Reporting covers the monthly routine this page slots into.
Write the "Google AI visibility" paragraph of a monthly client report.
Use only the figures below. Do not add any figure, cause or forecast.
Rules: Google's report measures impressions in AI Overviews and AI Mode
combined; there are no clicks and no split between the two features.
Never say "traffic from AI Overviews". Maximum 90 words. Plain English.
This month impressions: 41,200. Last month: 33,900.
Top pages: year-end guide 14,900; record-keeping guide 9,200; services 1,100.
Our changes: added a next-step link block to the two guides on 9 October.A realistic first answer from a tool reads: "Links to your site appeared 41,200 times in Google's AI features this month, up from 33,900. Your two guides drew most of the visibility. After we added a next-step block on 9 October, the guides began sending readers towards your services page." That last sentence is the problem. The assistant has turned a date and a rise into a result, and nothing you pasted says visitors moved. Edit it to "We added a next-step block to both guides on 9 October; we will report visits to the services page from those guides next month." Every draft from an assistant needs this one question asked of it: which sentence claims a cause?
Reporting the Bing side without muddying the Google side
Microsoft introduced an AI Performance report in Bing Webmaster Tools on 10 February 2026, as a public preview; check how the tool labels it now before you describe it to a client. It measures something different: citations, meaning pages displayed as sources in AI-generated answers across Microsoft Copilot, AI summaries in Bing and select partner integrations. Its dashboard shows total citations, average cited pages per day, page-level citation counts, a timeline and grounding queries, which are the phrases the AI used when retrieving the content. Microsoft states the grounding queries are a sample of overall citation activity, and that citation counts do not indicate ranking, authority or placement within an answer.
So it is the one place where you can see query-like language, and the one place where the unit is a citation, not an impression. Report it in its own box. A short wording that works: "Bing's AI Performance report recorded 310 citations across 18 pages. Grounding queries suggest the AI associates the practice with deadline questions." The numbers there are invented for the illustration; what is real is the structure and the caveats. Check that the client's site is verified in Bing Webmaster Tools before promising this line, because the report cannot be built from nothing.
For a client whose buyers research in a work setting, where Copilot is often built into the office suite, this line can matter more than the Google one. For a client selling to the public, Google AI Overviews matter more. Choose which line gets the larger font accordingly.
When the Google report is missing
Google's help page says the report may not appear where a site has not received enough impressions in generative AI features. For a small local client this is common. Handle it in three steps.
- Write "Below Google's reporting threshold this month" in line 1. This is a true statement, not a failure.
- Do not fill the gap with a number from a third-party tracker unless the client has asked for it, and then label it as a sampled estimate from that tool, with the tool named. Trackers query a set of prompts; they are not Google's own count. For those, Best AI Overviews Tracking Software: What a Local Business Needs explains what they can and cannot see.
- Replace the line with a leading indicator the agency controls: how many client pages meet the basic eligibility Google describes (indexed, and eligible to show with a snippet), and how many have been fixed this month.
Google's own guidance also says there are no extra requirements to appear in AI Overviews or AI Mode beyond being indexed and snippet-eligible, and that the usual SEO best practice applies. That is a useful sentence to quote to a client who has been sold a special "AI optimisation" package. It also means line 6 of the report ("what we changed") should list ordinary improvements, not mysterious ones.
Three client conversations and the wording for each
The client who wants AI Overviews "traffic"
Say: "Google does not report clicks from AI features separately. What we can show is how often your pages appeared there and how total clicks and enquiries moved. I'd rather give you a number we can defend." Follow with the page-level join from the earlier example.
The client whose rival "ranks first in AI"
Say: "AI answers show a set of links that varies by question and by person, and Google's own report does not give a position. So a screenshot of a rival is one result for one query on one day." Offer to track a fixed list of ten questions monthly, and record the date, the wording and who is cited, labelled as manual spot checks and nothing more.
The client whose clicks fell while impressions rose
This is the awkward one, and it can be real: an answer that satisfies the reader can reduce clicks. Do not claim cause from the data you have. Say: "Impressions in AI features rose 21.5% while clicks on those guides fell 6%. Both can be true when people read the answer without clicking. We're checking whether the pages that lost clicks also lost enquiries, which matter more." Then actually check, using the client's analytics by page.
Mistakes that make an AI visibility report unsafe
- Adding two reports together. Google impressions plus Bing citations equals nothing. Different units, different counting, different audiences.
- Reading chart totals against table totals. A client who adds up the page table and finds it exceeds the chart will assume you made an error. Put the one-line explanation about property-versus-page counting in the footnote of the first report you send, then leave it out.
- Quoting preliminary data. Google marks the newest data as preliminary with a dotted line. Export for a full month that ended at least a couple of days ago.
- Claiming AI Overviews caused a lift in enquiries. You have impressions and total clicks. A cause needs a test, such as a change made to one page with its neighbours left alone.
- Trusting a draft from an AI assistant. Every adjective in the draft is a claim. Read for "driving", "boosting", "resulting in".
For the page-level work this reporting points to, SEO for AI Search: A Client Audit Checklist for Agencies is the companion tutorial, and if the client also buys ads, the pre-spend checks in Ads in AI Overviews: A Pre-Spend Checklist for Ad Agencies and Clients belong in the same quarterly review.
How to check the report is working after three months
After three monthly reports, put them side by side and ask four questions. Did the client ask the same question twice? If so, the page is not answering it. Did every line have a source named? Did any sentence claim a cause without a dated change behind it? And did the client act on anything in line 2? A report that nobody acts on is decoration. If the answer to the last question is no, shorten the report to lines 1, 2 and 6 and spend the saved time on the pages themselves.
Keep a private changelog per client with the date of every page edit. When a rise appears, you can then say "the guide was updated on the 9th and impressions rose in the following weeks", which is an honest description of timing without claiming cause. Rather than rebuild all this by hand each month, an agency can keep a template sheet with the six lines, the two export tabs and the prompt above pre-filled, so the monthly task becomes: export, paste, read, edit, send.
One last thing worth stating to the client once, in the first report: the data is Google's and Microsoft's, not the agency's, the definitions can change, and both companies have said they expect to add to these reports. Keep the template flexible, and re-read the vendors' help pages each quarter before the review.
Questions clients ask about AI Overviews numbers
Why can't I see clicks from AI Overviews in Search Console?
Google's Generative AI performance report documents impressions, with pages, countries, devices and dates as the ways to slice them. Clicks from AI features sit inside the ordinary Performance report under the Web search type, mixed with every other result. Tell clients that AI-driven clicks are included in totals but cannot be separated.
Does the report separate AI Overviews from AI Mode?
No. Google's help page lists AI Overviews and AI Mode as the features included, and the report offers no dimension or filter that splits them. Label the line in your reports as 'AI Overviews and AI Mode combined' so nobody quotes it as an Overviews-only figure.
What if the client's property has no Generative AI report?
Google says the report may be missing when a site has not received enough impressions in generative AI features. Report that plainly as 'below the reporting threshold', check again next month, and do not substitute a third-party estimate without labelling it as one.
Should we report Bing citations to a client who only cares about Google?
Yes, in a short separate line. Bing's AI Performance report covers citations in Microsoft Copilot and Bing summaries, which is a different audience and a different metric. It is useful context for B2B clients, but never add it to the Google impressions figure.
How often should an agency report AI visibility?
Monthly is enough for most retainers. Impressions move slowly, the table is limited to 1,000 rows like the main Performance report, and weekly numbers invite clients to react to noise. Keep a quarterly page-level review for the pages that actually earn citations.
Further reads
- How to Track Whether AI Search Is Sending You Customers — Tie AI visibility to enquiries using the client's own analytics.
- AI Content Workflow for Agencies: From Brief to Approved Draft — Turn what the report shows into page updates the agency can approve.
- PR for AI Search: How Agencies Earn Citations in ChatGPT Answers — How earned coverage turns into citations in AI answers.
- How to Choose an SEO Agency That Understands AI Search — The client's side of this conversation, from their buyer's view.
- How AI Overviews Are Changing Local Search for Small Businesses — How Overviews change the results page for local clients.
- How to Get Your Business Recommended by ChatGPT and AI Search — What clients can do on their own pages and profiles.
- Best AI SEO Agency for Hotels? Questions to Ask Before You Sign — Judge an AI SEO agency for a small hotel by its proof, contract terms and reporting, and compare it with doing the steps yourself.
- AI SEO for Restaurants: Can You Get Recommended by ChatGPT? — How a restaurant can test what ChatGPT and Google AI say about it, then fix listings, menu text, reviews and crawler settings, and what stays out of its hands.
- Google AI SEO Agency: Questions to Ask Before You Hire One — A buyer's checklist for hiring a Google AI SEO agency: proof to demand, reports to expect, AI Overview tracking, contract clauses and red flags.
- Semrush AI Visibility vs Ahrefs Brand Radar: Which Suits an Agency? — A neutral, costed head-to-head of Semrush AI Visibility and Ahrefs Brand Radar for an agency tracking 3, 10 or 25 clients, plus a test to run.
- AI Keyword Clustering for SEO Agencies: Validate Clusters Before Sign-Off — Embed or prompt-cluster a keyword list, validate a sample against search results, and send the client a map they can approve, with costs per 10,000 keywords.
- AI SEO Tools for Local SEO: Citations, Reviews and Map Reports — A priced shortlist of local SEO tools for agency clients, compared on citations, review replies and map-pack reports, plus a trial test to run.
- AI Schema Markup Generator: Validate Client Structured Data Before Launch — A pre-launch routine for AI-drafted JSON-LD: prompts that stop invented data, then validator, Rich Results Test and Search Console checks.
- Best AI for Content Analysis: Audit a Client's Article Library — A five-stage audit of a client's article library: export the URLs, cluster by topic, list gaps and overlaps, then rule on merge, refresh or prune.
- Content Refresh With AI: Update Old Articles and Track Rankings — A client-by-client method for choosing old articles, refreshing them with AI, and measuring the ranking change against pages you left alone.
- How to Advertise on ChatGPT: Approval, Budget and Reporting for Agencies — Run ChatGPT ads for a client with sign-off, a capped test budget, tracking and a weekly report, based on OpenAI's own Ads Manager documentation.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Google Search Central documentation and announcements (AI features and your website; Search Generative AI performance reports, 3 June 2026), Google Search Console Help (Generative AI performance report), Bing Webmaster announcement (AI Performance public preview, 10 February 2026).