How do I track whether ChatGPT mentions my business?
Short answerPick 10 to 20 questions a real buyer would ask, run them in ChatGPT, Perplexity and Google every month in the same way, and record whether you are named, in what position, and which sources are cited. Add the free signals: ChatGPT referrals in your analytics, Bing Webmaster Tools' AI Performance report, and asking every new enquirer how they found you.
Key points
- No AI platform gives you a full report of when it mentions your business, so tracking means testing a fixed set of buyer questions on a schedule.
- AI answers change from run to run, so repeat each question and look for patterns over months, not single results.
- ChatGPT tags many of the links it cites with utm_source=chatgpt.com, and Bing Webmaster Tools now reports Copilot citations, so some first-party data does exist.
- Paid trackers save time at scale, but for a single-location service business a careful monthly spreadsheet is usually enough.
Tracking AI mentions feels harder than tracking Google rankings because it is. There is no single position to look up, and no platform sends you a monthly report of how often it named you. What you can build is a simple, repeatable routine that tells you whether things are moving in the right direction, and that is what this guide sets out.
Why can't you look it up like a Google ranking?
A Google ranking is fairly stable: search the same phrase from the same place and you see roughly the same list. AI answers work differently in three ways that matter for tracking.
- Every answer is written fresh. Ask ChatGPT the same question twice and you can get two different shortlists. Neither is wrong; the tool generates a new answer each time.
- Answers depend on the person asking. Location, earlier messages, saved memory and the exact wording all change what comes back.
- There is no official dashboard for mentions. OpenAI, Perplexity and Google do not tell businesses how often they were named in answers. Google includes AI Overviews and AI Mode traffic in Search Console, but inside your ordinary web search figures, not as a separate count of mentions.
So the honest goal is not a precise number. It is a consistent sample, taken the same way every month, that shows whether you are named more often, described correctly, and cited from the right sources.
Which questions should you track?
Write down 10 to 20 questions, phrased the way a buyer would type them, not the way you would describe yourself. Cover four kinds:
- Category and place: "Who are the best [service] providers in [town]?"
- Problem first: the question a buyer asks before they know what they need, such as "How do I make a small kitchen feel bigger?"
- Your name: "What do you know about [your business]?" This checks whether the facts are right.
- Comparison: "[Your business] or [competitor]: which should I choose?"
Keep the wording fixed once you start. If you change the questions every month you are measuring your phrasing, not your visibility.
Sample business: Hartwell Joinery, bespoke kitchens, Leeds. A starting set might be: "Who makes the best bespoke kitchens in Leeds?", "Bespoke kitchen maker near Harrogate", "How much does a bespoke kitchen cost in Leeds?", "Is a bespoke kitchen worth it over a fitted one?", "What do you know about Hartwell Joinery?" and "Hartwell Joinery or Oak & Iron Kitchens?". Six questions covering buyers at every stage, each one repeated monthly.
How do you run a manual check?
Once a month, set aside an hour and run every question through each tool your buyers use. For most businesses that is ChatGPT with search switched on, Perplexity, and Google (both the AI Overview at the top of the results and AI Mode where it is available in your country).
- Use a temporary chat in ChatGPT, or a logged-out browser, so your own history does not shape the answer.
- Run each question at least twice, and note whether you appeared in both runs, one, or neither.
- Copy the answer and the cited sources into the sheet the same day; answers you do not record are gone.
Record the same columns every time:
| Column | What to write | Why it matters |
|---|---|---|
| Named? | Yes, no, or in one of two runs | The headline measure |
| Position | First, second, third, or in a longer list | Being named first is worth more than being named fifth |
| Who else | The businesses named instead of or beside you | These are the competitors the tool already trusts |
| Sources cited | The sites linked under the answer | The places you most need to be listed, correctly |
| Accuracy | Any wrong address, price, service or date | Wrong facts point to inconsistent listings |
Look at the trend across three months, not the result of one afternoon. A single run is noise; the same pattern three months running is a finding.
What do the platforms tell you for free?
Alongside the manual check, four free sources give you harder evidence.
Referral visits in your analytics
OpenAI says ChatGPT adds utm_source=chatgpt.com to many of the links it cites, so those visits show up in Google Analytics and similar tools as coming from ChatGPT. Visits from Perplexity, Copilot and Gemini usually appear as referrals from their own domains. This only counts people who clicked, which is a fraction of people who saw your name, but it is real traffic you can tie to enquiries.
Bing Webmaster Tools' AI Performance report
Microsoft added an AI Performance report to Bing Webmaster Tools in February 2026, in public preview. It shows how often your pages are cited in Microsoft Copilot and Bing's AI summaries, which pages, and the queries that led to them. It does not show clicks and does not cover ChatGPT, but it is the first first-party citation data a major platform has offered, and it costs nothing.
Google Search Console
Google confirms that appearances in AI Overviews and AI Mode are counted in the Performance report under the Web search type. You cannot separate them out, but rising impressions on question-style searches are a useful sign. The Search Console guide for owners shows where to look.
Asking every enquirer
Add "An AI tool such as ChatGPT" as an option on your enquiry form's "How did you hear about us?" question, and ask on the first call. Buyers who read your name in an answer and then searched for you directly will never show up as an AI referral in analytics. This is the only way to count them.
Are paid AI visibility trackers worth it?
A growing number of tools will run your questions across several AI platforms every day and chart your "share of voice". They save real time if you track hundreds of questions, many locations or many products. Subscriptions commonly run from tens to several hundred dollars a month, depending on how many questions and platforms you track.
Before paying, understand what they cannot do:
- Many query the platforms through programmatic access or simulated sessions, which may not match what a real buyer sees in the app.
- The questions are still the ones you choose. A tracker measuring the wrong questions precisely is still measuring the wrong questions.
- Their scores are samples from their own runs, not counts of real conversations. Treat a "visibility score" as a trend line, never as market share.
For a single-location service business, a monthly manual check plus the free data above usually tells you everything a tracker would. Revisit the question if the spreadsheet starts taking more than an afternoon.
What should you do with what you find?
The record is only useful if it changes what you do next. Read it like this:
- Not named anywhere: the tools either cannot find you or cannot tell what you do. Start with the fixes on the AI search visibility guide, beginning with plain statements on your own website.
- Named, but with wrong facts: your details disagree across the web. Correct every listing to one version.
- A competitor named with you absent: look at the sources cited under their name. Why ChatGPT recommends a competitor explains what those sources usually reveal.
- Named in Perplexity but not Google, or the reverse: each tool weighs sources differently. See how to get cited by Perplexity and how to appear in Google AI Overviews.
How often should you check?
Monthly is the right rhythm for most businesses. AI tools pick up website changes as they re-crawl, often within weeks, while corrected listings and new third-party mentions take one to three months to show. Checking weekly mostly records random variation and tempts you to react to it.
Keep the same questions, the same tools and the same method for at least six months before judging whether your work is paying off.
If you would rather have the first baseline done properly, a structured AI visibility audit sets the question list, runs it against your two closest competitors and tells you what to fix first. M/AFZAL's Growth Partner retainer includes the monthly check as standard, but the routine above works just as well run by you.
Straight answers.
The follow-up questions owners ask most.
- Can I see how many times ChatGPT has recommended my business?
- No. OpenAI does not publish mention or impression counts for businesses. The closest first-party signal is visits arriving from chatgpt.com in your analytics, which only shows people who clicked.
- Does it matter if I am logged in when I test?
- Yes. ChatGPT can use memory and past chats, which colours the answer. Test in a temporary chat or a logged-out session so the result is closer to what a new buyer sees.
- Why does ChatGPT give a different answer every time I ask?
- The answers are generated fresh each time and, with search on, built from whatever pages it retrieves that moment. Variation is normal, which is why one result proves very little.
- Should I track mentions in Gemini and Claude too?
- Track the tools your buyers actually use. For most businesses that means ChatGPT, Google and Perplexity first; add Copilot, Gemini or Claude if your buyers work in companies that use them.