How to Track Brand Mentions in ChatGPT (Free Method + Tools)

Short answer: ChatGPT has no built-in brand-mention dashboard, so tracking means asking it the questions your customers ask and logging what comes back — systematically, on a schedule, across the queries that matter to your revenue. Do it manually with a prompt panel, or automate it with an AI visibility tracking tool.

Here is the exact process we use, the free version and the tool-assisted version, and what to do with what you find.

Why this is worth tracking at all

When ChatGPT answers “who should I hire for X,” it names specific businesses. If you are one of them, that is a referral channel you are not measuring. If a competitor is and you are not, that is market share quietly moving. Either way, you want eyes on it — the same way you watch your Google rankings.

The free method: a monthly prompt panel

1. Build your question list

Write 10 to 20 questions a real customer would ask, in plain language: “best [your service] in [your city]”, “who should I hire for [problem]”, “[competitor] vs alternatives”. Pull them from sales calls, support tickets, and the queries already sending you traffic.

2. Ask, twice, in a clean session

Run each question in a fresh ChatGPT conversation — chat history and memory personalize answers, and you want what a stranger sees. Ask each question at least twice on different days: answers vary between sessions, so a single run can mislead you.

3. Log five things

Were you mentioned at all?
What position — first name offered, or an afterthought?
What did it say about you — and is it accurate?
Which competitors were named?
Did it cite a source — and whose page was it?

4. Repeat monthly and watch the trend

One snapshot is trivia; a monthly series is intelligence. Track mention rate across your panel over time — that number responding to your work is how you know your AI recommendation efforts are landing.

The tool-assisted method

AI visibility platforms — Semrush AI tracking among them — run prompt panels for you at scale, track your share of AI answers against competitors, and alert you on changes. Worth it once you are tracking more than a couple dozen prompts or multiple AI engines. Our guide to measuring AI search visibility compares the approaches across ChatGPT, Gemini, and Perplexity.

What to do with what you find

Not mentioned: your entity footprint is thin — fix consistency, schema, and third-party presence first
Mentioned but described wrong: the sources ChatGPT reads are stale or inconsistent — update your site, profiles, and directory listings
Competitor cited from a roundup page: earn placement on that exact page — it is already in the answer path
Mentioned and accurate: protect it — keep reviews flowing and content fresh

A worked example: the monthly panel for a local plumber

Here is the method applied to a concrete business, so you can copy the shape.

Five money prompts: “best plumber in [city]” / “who should I call for a burst pipe in [city]” / “[city] plumber reviews — who’s reliable?” / “cheapest way to fix a water heater — should I call a plumber?” / “best plumbing company near [neighborhood]”.
Two competitor prompts: “compare [you] vs [rival]” and “is [rival] any good?” — you learn what the engines believe about the category, not just about you.
One source prompt: “what sources would you check to pick a plumber in [city]?” The answer tells you which directories and review sites the engine trusts — that list is your citation to-do list.

Run all eight on the first business day of the month, same wording, logged in one spreadsheet: date, engine, prompt, named brands, sources cited. Twenty minutes. After three months you have a trend line; after six you know exactly which fixed citation moved you into the answers.

Frequently asked questions

Why do I get different answers than my customers?

Chat history, memory, location, and model version all shift answers. Always test in a fresh session, and remember you are sampling a distribution — not reading a fixed ranking.

Can I see how often ChatGPT actually mentions my brand to real users?

No — OpenAI does not publish that data. Panel testing is a proxy, but run consistently it tracks direction reliably, and citation traffic shows up in your analytics as referral visits from chatgpt.com.

How many prompts should I track?

Start with 10 to 20 high-intent questions. Depth beats breadth: five revenue-critical prompts tracked monthly beat fifty random ones tracked once.

Should I track Gemini and Perplexity with the same panel?

Yes — same prompts, separate columns. The engines disagree often, and the disagreement is information: it tells you which source each one leans on.

What if my brand never appears at all?

Start with the source prompt above. Get listed and reviewed on the sources the engine names, fix your entity consistency, and give it a quarter. Absence is almost always a data problem, not a popularity problem.