Rank tracking in Google is a solved problem. Pick a keyword, pick a location, get a position. AI search doesn't work like that. There's no position one in ChatGPT, the answer changes from one run to the next, and Google only started reporting AI Overview data to site owners in 2026.
That doesn't make it unmeasurable. It means measuring it differently. This guide covers what to track, the four ways to track it, and a setup that works for a small business without an enterprise budget.
Key takeaways
- Track how often you're named across many runs, not whether you appear in one test.
- Search Console now has a generative AI report for Google's AI features. It shows impressions by page, but no queries or clicks.
- GA4 can show AI referral traffic and leads with a custom channel. It's free and takes ten minutes.
- Paid tools automate prompt sampling across engines. They're worth it once you track more than a couple of dozen questions.
- Report AI visibility alongside leads, so it never becomes a vanity metric.
The short answer
To track AI visibility, build a fixed list of the questions your customers ask, run each one several times a month in ChatGPT, Gemini, Perplexity and Google, and record how often you're named and linked. Add Search Console's generative AI report for Google's AI features and a custom AI channel in GA4 for traffic and leads. Use a paid tracker once the manual sampling gets too big to run by hand.
Why AI visibility is harder to track than rankings
Three things make AI tracking different from rank tracking:
- No fixed positions. An AI answer names a handful of businesses in a paragraph. Being second in the paragraph isn't the same as ranking second.
- High churn. Ahrefs tracked ChatGPT citations daily for 20 days and found 59% were gone by day two. One test is a coin flip.
- Personalisation and location. Answers shift with location, previous conversation and whether the tool searched the web for that answer.
So the unit of measurement isn't a position. It's a rate: out of all the times this question was asked, how often were you in the answer? My breakdown of ChatGPT ranking factors explains why the answers move so much.
What to measure
| Metric | What it tells you |
|---|---|
| Mention rate | Share of runs where your brand is named in the answer |
| Citation rate | Share of runs where your site is linked as a source |
| Share of voice | Your mentions as a share of all brand mentions in your question set, compared with competitors |
| Sources cited | Which third party sites the AI keeps linking to for your topic, and so where you need a presence |
| Accuracy | Whether the AI describes your services, prices and areas correctly |
| AI referral traffic and leads | Whether any of it turns into enquiries |
Mention rate and leads are the two that matter most. The rest explain them.
Method 1: Manual prompt sampling (free)
This is where everyone should start, because it teaches you more about your market than any dashboard.
- Write 10 to 20 questions your customers would actually ask. Mix recommendation questions ("best roofer in Tamworth"), problem questions ("why is my flat roof leaking") and cost questions ("how much does a new roof cost UK").
- Run each one three times in ChatGPT, Gemini, Perplexity and Google, in a logged out or private session, ideally on different days.
- Record four things per run: were you named, were you linked, which competitors were named, which sources were cited.
- Repeat monthly with exactly the same questions, so the trend means something.
A simple spreadsheet works. Ten questions, four engines, three runs is 120 checks, which takes about an hour. Past that it's worth automating.
Method 2: Search Console's generative AI report (free)
In June 2026 Google started rolling out generative AI performance reports in Search Console. You'll find them as a tab under the Performance report. They show how often your pages appeared in Google's AI features: AI Overviews and AI Mode in Search, with a separate report for AI features in Discover.
| Shows | Doesn't show |
|---|---|
| Impressions in Google's AI features | Clicks or click through rate |
| Which of your pages appeared | Which queries triggered them |
| Country, device and date breakdowns | Where in the AI answer you appeared |
| Data from 18 May 2026 onwards | ChatGPT, Perplexity or any non Google tool |
It's limited, but it's the only first party data you'll get from Google on this. Use it to see which pages Google trusts enough to use in AI answers, then strengthen those pages and build similar ones. If the report isn't showing for you, Google's explanation is that your site may not have enough data yet. The details are in Google's announcement on the Search Central blog.
Method 3: AI referral traffic in GA4 (free)
When someone clicks a link inside ChatGPT, Perplexity or Gemini, the visit usually arrives with a referrer you can identify. Set up a custom channel group in GA4 so these visits get their own channel rather than being lumped into Referral:
- Go to Admin, then Data display, then Channel groups.
- Copy the default channel group.
- Add a channel called AI Referral and move it above Referral.
- Set the condition to Source matches regex:
chatgpt.com|chat.openai.com|perplexity.ai|gemini.google.com|copilot.microsoft.com|claude.ai
Two blind spots to know about. Clicks from AI Overviews and AI Mode show up as ordinary Google organic traffic, so they won't appear in this channel. And some AI visits arrive with no referrer at all and land in Direct. Rising branded searches in Search Console are a useful second signal, because people often see a business in ChatGPT and then Google it by name.
Method 4: AI visibility tracking tools (paid)
Paid trackers automate method one. You give them a list of prompts, they run them on a schedule across several AI engines and report mention rates, citations and share of voice against competitors. The category has grown fast. Some of the established options:
| Tool | Good for |
|---|---|
| Ahrefs Brand Radar | Brand mentions across a large prompt database, covering AI Overviews, AI Mode, ChatGPT, Gemini, Perplexity, Copilot and Grok |
| Semrush AI Visibility Toolkit | Teams already using Semrush for SEO who want AI data in the same place |
| Peec AI, Profound, Otterly | Dedicated AI trackers built around your own custom prompt lists |
Pricing changes often in this category, so check current plans directly. When comparing, ask four questions: which engines does it cover, how many prompts does the plan include, how often does it run them, and can you add your own prompts rather than relying on its database.
Treat every tool's numbers as samples, not facts. They're running the same kind of checks you would by hand, just more of them.
A setup that works for a small business
| Layer | What | Cost | How often |
|---|---|---|---|
| Question set | 15 fixed questions across recommendation, problem and cost | Free | Set once, review quarterly |
| Sampling | Manual runs across four engines, or a tool once it gets too big | Free to start | Monthly |
| Google AI features | Search Console generative AI report | Free | Monthly |
| Traffic and leads | GA4 AI Referral channel plus key events | Free | Weekly glance, monthly report |
| Brand demand | Branded impressions in Search Console | Free | Monthly |
Put the results on the same page as your normal SEO numbers. My SEO report example shows the one page layout, and an AI section fits under the rankings table: mention rate this month, last month, and the competitor who's named most.
Tracking only tells you where you stand. To move the numbers, start with my guide on how to rank in ChatGPT and AI Overviews.
FAQs
How do I track my brand in ChatGPT?
Build a fixed list of customer questions, run each one several times a month in ChatGPT and record how often your brand is named and linked. Add a custom AI Referral channel in GA4 to see whether ChatGPT sends you traffic and leads.
Does Google Search Console show AI Overview data?
Yes. Since June 2026, Search Console has been rolling out a generative AI performance report showing impressions in AI Overviews and AI Mode by page, country, device and date. It doesn't show clicks or the queries behind them.
What is a good AI mention rate?
There's no universal benchmark, because it depends on how many businesses compete for your questions. Track your own rate against your closest competitors and aim to grow your share of voice month on month.
Are AI rank tracking tools accurate?
They're as accurate as their sample. AI answers change between runs, so tools report rates across repeated checks. Use them for trends and competitor comparisons rather than treating any single result as fact.
Can I track AI Overview clicks in GA4?
Not separately. Clicks from AI Overviews and AI Mode are counted as Google organic traffic. Search Console's generative AI report shows impressions in those features, but not clicks.
Where this leads
If this is the bit you want handled rather than explained:
- GEO services. Your question set tracked every month across ChatGPT, Gemini and AI Overviews.
- AI SEO. The fixes that move the numbers once tracking shows where you stand.
- A free first pass. I will run your ten most important questions and send you who gets named.
