In brief
To track brand mentions in AI search, run a fixed set of real buyer questions through ChatGPT, Perplexity, Gemini, Claude and Google on a schedule. Record whether you are named, where, in what tone, who else appears and which pages are cited. You can do this by hand in a spreadsheet, or with a tool once the list grows.
Key points
- Tracking starts with a fixed set of real buyer questions, not a list of keywords.
- Record mention, position, sentiment, competitors, recommendation and cited pages for every answer.
- Answers vary from run to run, so ask each question more than once and watch trends.
- A spreadsheet works for a few dozen questions checked monthly.
- A tool helps once you track many questions, several engines, locations or clients.
- Search Console, Bing Webmaster Tools and GA4 add free evidence alongside your own tracking.
The steps at a glance
- 1
Write your prompt set
List 20 to 50 real questions buyers ask, in their words. Cover research, comparison and choosing a supplier, plus a few questions that name your brand.
- 2
Choose the engines
Pick the engines your buyers use, usually ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews, with AI Mode measured through Search Console.
- 3
Decide what to record
For each answer record whether you are mentioned, your position, sentiment, competitors named, whether you were recommended and which pages were cited.
- 4
Run the questions by hand
Use temporary or logged-out sessions, ask each question two or three times, and log every answer in a spreadsheet with the date and engine.
- 5
Move to a tool when the list grows
Once you track more than a few dozen questions, several locations or clients, use a tool that runs the prompts on a schedule and reads each answer the same way.
- 6
Add free data sources
Export the Search Console generative AI report, check Bing Webmaster Tools AI Performance, and watch AI referral visits in Google Analytics 4.
- 7
Review the trend and act
Each month, compare mention rate, position and citations against the last period. Find the questions and pages where competitors win and work on those first.
What does tracking brand mentions in AI search mean?
Tracking brand mentions in AI search means checking, on a regular basis, what AI engines say when buyers ask about your field. You want to know whether your brand is named, how prominently, in what tone, and which pages the engine used. You also want to see who is named instead of you.
A brand mention in an AI answer is different from a mention on social media or in the press. It happens privately, inside one person's chat, and it is not published anywhere. The only way to see it is to ask the same questions yourself, in a consistent way, and record the answers.
Which AI engines should you track?
Track the engines your buyers use. For most UK businesses that means six.
- ChatGPT, with search on, because it cites pages when it searches.
- Perplexity, which searches the web for most questions and numbers its citations.
- Google AI Overviews, which appear above normal results for many questions.
- Google AI Mode, Google's separate chat-style search.
- Google Gemini, which many people use through Android and Google apps.
- Anthropic Claude, popular with professional and technical users.
If budget or time is short, start with ChatGPT and Google AI Overviews. Add the others once the routine is working.
What should you record for each answer?
Record the same fields every time, so answers can be compared across weeks and engines. The table shows what to capture and why.
| Field | What it means | Why it matters |
|---|---|---|
| Mentioned | Whether your brand is named at all | The basic visibility count |
| Position | Where your brand first appears, first, second and so on | Being named first carries more weight |
| Sentiment | Positive, neutral or negative tone | A mention with a warning is not a win |
| Recommended | Whether the engine suggests choosing you | Shows intent, not just awareness |
| Competitors | Every other brand named | Shows who you are up against in AI answers |
| Cited pages | The URLs the engine linked | Shows where the answer came from |
| Accuracy | Whether facts about you are right | Wrong prices or places cost enquiries |
Position lets you work out share of voice weighted by position. Cited pages tell you where to act, because they show which sites shape the answer.
How do you track brand mentions in AI search by hand?
You can track by hand with a spreadsheet and a couple of hours a month. It is the best way to learn how the engines behave before paying for anything.
- Write your prompt set. List 20 to 50 real questions buyers ask. Take them from sales calls, enquiry forms, reviews and forums. The prompt research guide covers where to find them.
- Mix the question types. Include "best" and "near me" questions, comparisons, problem questions and a few that name your brand directly.
- Set up a clean session. Use a temporary chat in ChatGPT and a logged-out or private window elsewhere. This reduces the effect of your own history.
- Ask each question two or three times. Answers vary, so one run can mislead.
- Log every answer. Use one row per answer, with the date, engine, question and the fields from the table above. Paste the cited URLs too.
- Repeat monthly. Keep the questions the same so the trend means something.
The manual method has limits. It is slow, and it is easy to record answers slightly differently each time. Your location and account settings can also creep in. OpenAI says ChatGPT may use your approximate location and saved memories when it turns your question into searches.
How do you track brand mentions with a tool?
A tool does the same job on a schedule and reads every answer the same way. It is worth it once you have more than a few dozen questions, several engines, more than one location or clients to report to.
A good tool will run your prompt set across the engines you choose, weekly or more often. It will record mention, position, sentiment, competitors, recommendation and cited pages for every answer. It should let you run the same prompts for different countries or regions if you sell in more than one place. It should also show trends over time rather than one monthly score.
When comparing tools, it helps to ask how they collect answers. Answers gathered through the engines' APIs are consistent and repeatable. They can differ a little from what a person sees in a consumer app, because personal history and location are removed. That is fine, as long as the tool is honest about it.
Which free sources help you track AI mentions?
Three free sources add evidence alongside your own tracking. None of them tells you the full answer, but each one shows part of the picture.
- Google Search Console. The generative AI performance report shows impressions from AI Overviews and AI Mode by page, country, device and date.
- Bing Webmaster Tools. The AI Performance report shows how often your pages are cited in Microsoft Copilot and Bing's AI summaries, and which pages.
- Google Analytics 4. Visits from chatgpt.com, perplexity.ai, gemini.google.com and other AI tools show up as referrals. The GA4 guide explains how to separate them.
These show what happened on your site. Your own tracking shows what the answer said, including the many times you were not named at all.
How do you track AI mentions for a business with several locations?
Run the same questions for each place you serve. A buyer in Perth and a buyer in Inverness asking for "the best kitchen showroom near me" may see very different answers.
By hand, that means adding the town to each question, such as "best kitchen showroom in Perth", and keeping a separate tab per location. With a tool, look for one that can set a location for each run, rather than just adding the town to the wording. Either way, report share of voice per location as well as in total. A strong total can hide a town where you never appear. The local SEO and AI search guide covers the local side in more depth.
What should you do with the tracking results?
Tracking is only useful if it leads to action. Each month, it helps to work through three questions.
Where are we missing? Sort the questions by how often you are named. Look at the ones where you never appear but competitors do. Open the cited pages and see what they have that you do not.
Which sources keep coming up? List the domains cited most often. If the same directory, review site or comparison article appears again and again, make sure you are on it with correct details.
What is wrong? Read the answers that mention you and check the facts. A wrong price, an old address or a service you no longer offer is worth fixing quickly, at the source page that states it.
Then pick two or three actions for the month and check whether the numbers move. The guide to measuring AI visibility explains how to read those numbers fairly.
Why does your brand appear in some AI answers but not in similar ones?
Because each wording leads the engine to search for slightly different things. "Best wedding photographers in Glasgow" and "affordable wedding photographers near Glasgow" may bring back different pages, and so different names.
The model also adds some randomness each time it writes. And the pages it finds change as sites are updated. This is normal. It is why tracking uses groups of related questions and repeated runs, and why the trend matters more than any single answer.
How often should you track AI brand mentions?
Weekly or fortnightly suits most businesses using a tool. Monthly is enough by hand. Checking daily mostly adds noise, because single answers vary so much.
Check sooner after a big change, such as a rebrand, a new location, a price change or a major content update. Those are the moments when wrong or out of date details most often creep into answers.
How Axiom GEO helps
AI visibility tracking in Axiom GEO runs your prompt set on a schedule across ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews. Each answer is read for mention, position, sentiment, competitors, recommendation and cited pages. Competitor share of voice records every brand named, whether it was recommended and why, per engine and over time. Local AI visibility runs the same prompts per country or region and splits share of voice by market. Prompt suggestions propose new questions to fill gaps in your set.
Sources
These are the external pages referred to in this guide.