To track your brand's visibility in AI search, ask the same buying questions on the same AI engines on a fixed schedule, and log who gets named and which sources get cited. Then add the two numbers Google now gives you for free: AI feature impressions in Search Console and AI Assistant visits in Google Analytics. That is how to track AI visibility in a way you can compare from one month to the next. This guide covers the method, what Google's data shows, and what a client or board report should contain.
The short answer
- Measure presence: how many answers name you, out of how many you asked. Your position inside one answer is noise.
- Keep the questions, the engines and the schedule fixed, and ask each question more than once.
- Search Console now shows AI Overview and AI Mode impressions, and Google Analytics has an AI Assistant channel. Neither tells you who the engines named instead of you.
How brands measure visibility in AI
We track three things, and we push back on any report that skips one.
- Presence. The share of answers that name you. "Named in 12 of 30 answers" is a measurement. "Ranked second in ChatGPT" is a screenshot.
- Competitors on the same questions. The businesses named instead of you, counted on exactly the same questions and runs, tell you whether you are winning or losing.
- Sources. The pages each engine cites, and whether those pages mention you. This is where the fixes come from.
Traffic from AI tools is context, since plenty of people read an answer and never click. If you have not started improving these numbers yet, our guide on how to improve visibility in AI search engines covers the fixes. This post is about measuring them.
How to track AI visibility yourself
You can monitor AI search visibility with a spreadsheet and an afternoon. The rule: between checks, the only thing that should change is the engines' answers.
- Fix the questionsTwenty to thirty questions buyers ask: finding a provider, a specific service, a problem, your reputation. Freeze the wording.
- Fix the enginesChatGPT, Gemini, Claude and Google's AI Overviews and AI Mode. Same account setup and location every time.
- Fix the cadenceMonthly or quarterly, in the same week of the month. Ask every question three times.
- Log every answerOne row per answer in a spreadsheet. All of them, not just the good ones.
| Log this | Why it matters |
|---|---|
| Date, engine, question, run number | Lets you compare like with like from one check to the next |
| Named you: yes or no | The count behind your presence number |
| Who else was named | Your competitors' presence, on the same questions |
| Sources cited | The sites the engines trust in your category: your outreach list |
| Was your own page cited? | Shows whether your site is a source or only a name |
Presence is the rows where you were named, divided by all rows. Work it out for each engine as well as overall, because ChatGPT and Google often disagree, and a blended number hides that.
The catch is volume. Twenty five questions on four engines, three runs each, is 300 answers per check. That is why tracking tools exist. Our comparison of AI visibility tools covers what they cost and how each one counts. If nobody will act on the data, a quarterly manual check of ten questions beats a dashboard nobody opens.
Why results vary from run to run
Ask the same question twice and you will often get different names, in a different order. Our audit page says it plainly: ask the same question three times and you can get different answers. Google says AI Overviews and AI Mode may run "multiple related searches across subtopics and data sources" before they answer, and that the two features "may use different models and techniques, so the set of responses and links they show will vary."
Single answers swing. Totals hold far better. One test we publish: two back to back runs of the same thirty question audit on Gemini put every named business within 1.2 points of itself. That is why we report presence across many answers and never a position in one.
Three habits cut the noise you add yourself:
- Same wording. Paste the questions from your sheet, never from memory.
- Same setup. The same account or a logged out window, the same location, the same model choice, every time.
- Repeat and flag. Ask each question three times, and mark the ones that swing instead of averaging them away.
One caveat: some tools and audits, ours included, ask the engines through their APIs, which do not behave exactly like the consumer app. Ask any vendor which one they measure.
What Search Console shows for AI Overviews and AI Mode
Google launched a Generative AI performance report in Search Console in June 2026, and says it reached all websites worldwide on August 31, 2026. It shows impressions in AI Overviews and AI Mode, which Google defines as how many times links to your site were shown in a generative AI feature on Google Search. You can break them down by page, country, device and date.
What the report does not show
- Clicks. Google lists impressions for this report. Clicks from AI features are counted in the regular Performance report under the Web search type, mixed in with the rest of your Search traffic.
- The questions. Google lists pages, countries, devices and dates, not the queries behind each impression.
- Mentions without a link. The report counts links to your site. An answer that names you but links somewhere else does not show up in it.
- Who was named instead, and anything outside Google Search. ChatGPT, Claude and the Gemini app are not in it.
Google's counting rules for the main report still apply. A click on a link in an AI Overview counts as a click. An AI Overview occupies a single position, and all its links share that position. A follow-up question in AI Mode counts as a new query. The report may not appear at all if your site has too few AI impressions.
Use it as a trend line next to your manual checks. Our guide to improving your visibility in Google AI Overviews covers how to get cited there in the first place.
AI referral traffic in Google Analytics
In May 2026, Google Analytics added an AI Assistant channel to its default channel group. When a visit's referrer matches a recognized AI assistant, GA4 sets the medium to ai-assistant and the campaign to (ai-assistant). Google's examples include ChatGPT, Gemini, Claude, Copilot, DeepSeek and Grok. It does not publish the full list.
To see it, open the Traffic acquisition report and find the AI Assistant row in the session default channel group. Switch to session source to see which assistant sent each visit, and compare their key events with organic search. Two limits to keep in mind:
- Google's own AI features are excluded. Google's definition leaves out AI Overviews and AI Mode, so those clicks sit in your Google search traffic. Search Console is the place to look for them.
- No referrer, no credit. The channel works from the referrer. A visit without one, such as a link a buyer copied out of an answer and pasted into a browser, cannot be credited to the assistant. Treat the number as a floor.
Reporting it to a client or a board
A board does not want 300 rows. It wants to know whether the business is recommended, next to whom, and what is being done. A report that a client or a board can read should show:
- One headline number: presence, named in X of Y answers, with the top competitors on the same questions.
- Each engine separately, so a gain on one cannot hide a loss on another.
- What changed since the last check: questions gained, questions lost, competitors that moved.
- The sources the engines cite, and which of them mention you.
- The fixes: what was done, what is next, and who owns each one.
- Traffic as context: AI Assistant visits and key events from Google Analytics, AI impressions from Search Console.
- A short method note: the questions, the engines, how many times each was asked, the dates, and whether the answers came from the app or an API.
- The raw answers, available to anyone who wants to check a finding.
Leave out a single "AI rank" and screenshots of one answer. Neither survives the question "what if you asked again?"
What our AI Visibility Audit report shows. It asks ChatGPT, Gemini, Claude and Google AI Overviews thirty buying questions for your category and city, three times each, and reports how often you are named against up to five competitors on the same questions. It ranks the sources the engines cite by domain, includes site readiness checks with fixes reviewed by our team, states the number of answers each engine returned, and prints its method and limitations. You get every raw answer as a spreadsheet and a 30 minute call to go through it. The first audit is back within two business days.
For tracking over time there are two re-check plans. Each re-check asks the same questions and adds a "Since last run" section: which questions you gained or lost, how you moved against competitors, and which sources changed.
| Plan | First audit (book by Dec 31) | Then | Audit price from January |
|---|---|---|---|
| One-time audit | $500 | No re-checks | $1,000 |
| Quarterly re-checks | $425 (15% off) | $300 every three months | $850 |
| Monthly re-checks | $375 (25% off) | $150 a month | $750 |
Prices as published on our AI Visibility Audit page, checked October 19, 2026. Re-check plans can be cancelled any time after the first re-check.
Where to start this week
- Today: write down twenty questions your buyers ask, in their words.
- This week: ask each one three times on ChatGPT, Gemini and Google. Log who is named and which sources are cited. That is your baseline.
- This month: open the Generative AI report in Search Console and the AI Assistant channel in Google Analytics, note both numbers next to your baseline, and put the next check on the calendar.
Free, about thirty seconds
See whether AI engines can read your site
Before you track who the engines name, make sure they can read you. The free AI Readiness Check reads your site and scores whether AI engines can find and understand it, then tells you what to fix first.
Run the free AI Readiness CheckWant the full picture? The AI Visibility Audit asks ChatGPT, Gemini, Claude and Google AI Overviews thirty of your buyers' questions and shows exactly who they name, with optional quarterly or monthly re-checks.
Common questions
How do I track AI visibility?
Pick twenty to thirty questions your buyers ask and ask them on the same AI engines (ChatGPT, Gemini, Claude and Google's AI features) on a fixed schedule. Ask each question more than once, and log whether you were named, who else was named and which sources were cited. Then report presence: how many answers named you, out of how many you asked.
How do I monitor AI search visibility without a paid tool?
Use a spreadsheet with one row per answer, and run the same questions monthly or quarterly. Add the two free Google sources: the Generative AI performance report in Search Console, which shows impressions of links to your site in AI Overviews and AI Mode, and the AI Assistant channel in Google Analytics, which shows visits from assistants such as ChatGPT. A paid tool mostly saves you the hours of asking.
How do I track brand visibility in AI Mode?
Search Console's Generative AI performance report counts impressions of links to your site in AI Mode and AI Overviews, by page, country, device and date. It does not show the question that was asked or who else was named, so also ask AI Mode your buying questions directly on a schedule and log the answers. Google counts a follow-up question in AI Mode as a new query.
Does Google Search Console show AI Overviews data?
Yes. Google launched a Generative AI performance report in June 2026 and says it reached all websites worldwide on August 31, 2026. It reports impressions in AI Overviews and AI Mode. Clicks from those features are counted in the regular Performance report under the Web search type, together with the rest of your Search traffic.
Can I see ChatGPT traffic in Google Analytics?
Yes. Since May 2026, Google Analytics sorts visits from recognized AI assistants into an AI Assistant channel, with the medium ai-assistant. Google's examples include ChatGPT, Gemini, Claude, Copilot, DeepSeek and Grok. Clicks from Google's own AI Overviews and AI Mode are not in that channel; they count as Google Search traffic.
How often should I check my AI visibility?
Monthly while you are actively publishing fixes, quarterly otherwise. Checking weekly mostly measures noise, because single answers change between runs. Whatever you choose, keep the questions and the engines the same so the numbers compare.
Sources
- Google Search Central Blog, Introducing Search Generative AI performance reports in Search Console (June 3, 2026)
- Search Console Help, Generative AI performance report (Search) and How clicks, impressions and position are counted
- Google Search Central, AI features and your website
- Google Analytics Help, Default channel group and release notes (May 13, 2026)
- Surf Rider Media, AI Visibility Audit method and pricing, checked October 19, 2026

