How to Track and Measure Your Brand’s Visibility in ChatGPT

To track your brand’s visibility in ChatGPT, ask it a fixed set of the questions your buyers ask, the same way each time, and record four things per answer: whether your brand is named, where it sits against competitors, how it is described, and which sources ChatGPT cites. Repeat on a schedule and compare.

What “visibility in ChatGPT” actually means

Visibility in ChatGPT comes down to four signals. Your brand is named in the answer or it is not. When it is named, it sits somewhere in a list or a recommendation. It is described in a particular way, accurately or not, warmly or not. And when ChatGPT searches the web, it cites particular pages, which may or may not include yours.

Two modes of answering need to be kept apart. ChatGPT can answer from what the model learned in training, or it can run a live web search and build the answer from the pages it finds. OpenAI’s help page on ChatGPT search says it may search the web automatically when a question would benefit from current information, and that answers which use search may include citations, with a Sources button listing the pages behind them. Answers from the model’s own knowledge carry no such list.

The search mode is the one you can influence fastest. What the model learned in training changes only when a new model is trained. A search-based answer changes as soon as the pages it reads change, so corrected profiles, better pages and fresh coverage on the sites ChatGPT already relies on can reach the answer long before the next model does.

Signal	What to record	Why it matters	What moves it
Named	Yes or no, for every prompt	A buyer who never sees your name cannot put you on a shortlist	A consistent presence on the sources ChatGPT reads for your category
Ranked	Position in the list or recommendation, or “only name given”	A brand listed fifth reads differently from a brand listed first or recommended alone	How often, and how favorably, those sources rank or recommend you
Described	Sentiment (positive, mixed, negative) and every factual error	A wrong price or an old dispute repeated as current shapes the buyer’s view before your own site does	Correcting the facts on the pages ChatGPT draws from, and the balance of reviews and coverage
Cited	Every domain and page listed under Sources	These are the pages actually shaping the answer, and your list of places to work on	Pages of your own that answer the question directly, and a presence on the third-party pages already cited

How to see your brand’s visibility in ChatGPT today (step by step)

You can see where you stand in an afternoon. Ask the way your buyers ask, and the same way every time, so that next month’s answers compare with this month’s.

1. Build 15 to 30 prompts from real buyer questions. Take them from sales calls, support tickets, onboarding emails and your own search data, and sort them into four groups:
   • Category: “best X for Y”, for example “best payments app for freelancers in Europe”.
   • Comparison: “X or Y”, your brand against the competitor buyers mention most often.
   • Problem: “how do I fix Z”, the problem your product solves, asked without naming any brand.
   • Brand: “is X legitimate”, “what does X do”, “X reviews”.
2. Open a clean session. Memory, custom instructions and past chats all shape what ChatGPT tells you. OpenAI’s memory help page explains that a Temporary Chat lets you choose, before it starts, whether it uses your memories and custom instructions, and that it does not create or update memories. Choose neither. A signed-out browser session is another clean option, and OpenAI states that people who are not signed in can use web search as well.
3. Use the same wording every time. Paste each prompt from your list rather than retyping it. A changed word is a changed prompt.
4. Run each prompt with search on and off. Ask it once as written and note whether ChatGPT searched: citations and a Sources button appear when it did. Then ask it again with Search selected from the tools menu, so that every prompt has a search-based answer on record next to the plain one.
5. Save the answer, the date and the sources. Copy the full answer text, the date, the mode and every domain under Sources. A screenshot is a backup. A text record is what you can compare.

The result is a snapshot: which prompts name you, in what position, in what terms, and on whose evidence. Its value comes from repeating it.

How to track brand mentions in ChatGPT for a person

For a founder, an executive or anyone else tracked by name, start with disambiguation. Names are shared, and ChatGPT will answer about the wrong person, or blend two people into one, without saying so. Before you read sentiment, check identity.

Add one or two identifiers to the brand prompts (the company, the role, the city) and run each one with and without them. Record whether ChatGPT identifies the right person, whether it mixes in facts from someone else with the same name, and which sources it cites for the identification. If the plain-name prompt returns someone else, that is the first finding, and it comes before any question of tone.

The prompt groups change a little for a person. Category prompts become “who are the leading experts in X” or “who runs Y”. Brand prompts become “who is [name]”, “is [name] trustworthy” and “what happened with [name]”, which is what a cautious investor, partner or employer asks before a first meeting.

How to measure brand visibility in ChatGPT: the metrics

Five measures turn saved answers into numbers you can compare month to month, each calculated across the whole prompt set, never from a single answer.

Mention rate

The share of prompts in which ChatGPT names you. Calculate it for each prompt group as well as overall. A brand that appears in every brand prompt and in no category prompt is known but not recommended, which calls for different work from a brand nobody has heard of.

Position

Where you appear when ChatGPT gives a list or a recommendation: first, second, fifth, or the only name given. Record the average position across the prompts that name you, and how often you are first.

ChatGPT share of voice

Share of voice is your mentions against those of the competitors ChatGPT names, counted across the same prompt set. If ChatGPT names you 6 times and three competitors 18 times between them, your share is 25 percent. Building the competitor set and comparing the figure across engines is covered in our guide on how to measure AI share of voice.

Sentiment and accuracy

Mark each answer that names you as positive, mixed or negative. Then list every factual error separately: pricing that has changed, a discontinued product presented as current, an old dispute described as open, a competitor’s feature attributed to you. Errors deserve their own list, because ChatGPT states them in the same confident voice as the facts it gets right.

Cited sources

List the domains under Sources for every search-based answer, count how often each one appears across the set, and mark which ones are yours. This shows which sites ChatGPT leans on for your category, whether any of your own pages are among them, and where an error in an answer came from.

A worked example (invented for illustration)

A payments app for freelancers runs 24 prompts in search mode: 6 brand, 10 category and 8 problem and comparison prompts. ChatGPT names it in 9, a mention rate of 38 percent: all 6 brand prompts, 3 of the 10 category prompts, and none of the rest. When it appears in a list, it sits fourth on average. Two answers quote a transfer fee the app dropped a year ago, and both cite the same review site, whose comparison page still shows the old fee. The category answers cite two comparison sites and one forum thread again and again, and none of the app’s own pages.

The reading: known but rarely recommended, misreported because of one stale page, and absent from the pages the category answers are built on. The order of work is to get the fee corrected on the review site, secure a place on the two comparison sites, and publish a page of its own that answers the category question directly.

How to track brand performance in ChatGPT responses over time

ChatGPT visibility tracking is a schedule plus a log. Monthly is enough for most brands. Move to weekly during a launch, a rebrand, a funding announcement or a crisis, when the sources about you are changing fast and you need to see whether the answers follow.

One answer proves nothing. ChatGPT can give different answers to the same prompt on different runs, and a search-based answer depends on what the search returned that day. That is why every metric above is a share across the set. A single category answer that drops your name tells you very little. A mention rate that falls across several category prompts tells you something.

Keep the log simple enough that someone keeps filling it in, one row per prompt per run:

• Prompt: the exact wording, taken from the fixed list.
• Date of the run.
• Mode: model knowledge or search.
• Named: yes or no.
• Position: a number, or “only name given”.
• Sentiment: positive, mixed or negative, plus any factual error in a short note.
• Sources: every domain listed.

Decide in advance what counts as a real change. A workable rule: a shift counts when it appears across several prompts in the same group and holds for two runs in a row. A move in one prompt, or a move that reverses the next month, is variation. A new domain in the citations deserves a note even on its own, because it shows where the next change in the answers may come from.

ChatGPT brand monitoring works best next to your search data rather than apart from it, since buyers use both and the answers draw on the open web. The case for measuring the two together is set out in what brands should be measuring now.

Tracking competitors in ChatGPT

Your prompt set already tracks competitors, because every category and comparison answer names them. Read those answers for three things.

• Who appears. List every competitor ChatGPT names for your category prompts, how often and in what position. The list can include names your sales team rarely hears.
• Which sources they win on. When a competitor is named and you are not, open the sources behind that answer. Those pages (comparisons, review sites, directories, articles, forum threads) are where the competitor has a presence you lack.
• What ChatGPT says about them. The strengths it attributes to them are the claims a buyer will hold you against.

The second item becomes your work list. Sort the cited pages into two groups. In the first are pages where you could reasonably appear: a comparison that leaves you out, a directory you never joined, a review site with an empty profile. That group becomes outreach and profile work. In the second are pages on topics where you have published nothing worth citing. That group becomes your content plan.

Manual tracking or a tool: what a tool to track brand visibility in ChatGPT should do

A spreadsheet and a fixed prompt list are enough for one brand in one market and one language. Once you track several markets, several assistants or more than a few dozen prompts, a tool saves real time. The category changes quickly, so judge any tool against what the method needs:

• Fixed prompt sets. You define the prompts, and the tool runs them unchanged, run after run.
• Several engines. ChatGPT is one of the assistants your buyers use, and the same prompt set should run on the others too.
• Search mode captured. Each answer is stored with a record of whether it came from a live search.
• Cited sources stored. The domains and pages behind each answer are kept, so you can see which sources matter and when they change.
• History and export. Past runs stay available and export to a file, so the trend survives a change of tool.
• Markets and languages. Prompts run from the countries, and in the languages, your buyers use.

Whatever a tool reports, the judgment stays with you: which cited sources matter for your buyers, which errors to correct first, and what to change on which page.

What to do with what you find

ChatGPT leaves you out. The gap is usually one of entity and authority: ChatGPT lacks consistent facts about who you are, or you are absent from the sources it cites for your category. Closing it is entity and authority work, on your own site and on the third-party pages your competitors win on.

ChatGPT gets you wrong. Arguing with the answer changes nothing, because the error lives on a page ChatGPT reads. Find that page in the citations, have it corrected, and make sure the accurate version exists on pages ChatGPT trusts. Correcting the record at the sources is the core of AI reputation management.

The tone is wrong. Lukewarm or negative answers usually reflect the balance of reviews, coverage and discussion about you. How that balance feeds the answer is explained in how sentiment shapes AI recommendations.

You need the whole picture. ChatGPT is one of six engines that a full online reputation audit reads, alongside Claude, Gemini, Grok, Perplexity and Copilot, together with Google results, reviews and authority signals.

Google works differently, with AI Overviews sitting on top of a ranking you can read, and our separate guide shows how to track your brand in Google AI Overviews.

If ChatGPT leaves you out or gets you wrong, send us your name and we will tell you what it says, which sources it relies on and what would change the answer.

Frequently asked questions

How can I track my brand’s success in ChatGPT?

Run a fixed set of 15 to 30 buyer prompts every month in a clean session, with and without search, and log whether your brand is named, its position, how it is described and which sources are cited. Success shows as a rising mention rate and share of voice across the set, fewer factual errors, and your own pages appearing among the cited sources.

How do I track my brand’s performance in ChatGPT responses over time?

Keep the prompt list fixed and rerun it on a schedule: monthly for most brands, weekly during a launch or a crisis. Log each run with the date, mode, mention, position, sentiment and sources. Judge changes on shares across the whole set, and count a shift as real only when it appears across several prompts and holds for two runs in a row.

How to see brand visibility in ChatGPT?

Open a Temporary Chat set to ignore your memories and custom instructions, or use a signed-out session, then ask the questions your buyers ask: category, comparison, problem and brand questions. Record whether you are named, where, how you are described, and which sources ChatGPT lists when it searches. The pattern across 15 to 30 prompts is your visibility.

Where can I find tools to track my brand’s visibility in ChatGPT?

Tracking tools come from SEO software companies and from specialist AI visibility startups, and the category changes quickly. Judge any of them on six points: fixed prompt sets, several engines, search mode captured, cited sources stored, history with export, and coverage of your markets and languages. Whatever the tool, deciding which sources to work on stays with you.

How do I track what ChatGPT and other AI tools are saying about my brand?

Run one prompt set across every assistant your buyers use, such as ChatGPT, Gemini, Claude, Perplexity and Copilot, on the same day in each. Log the same fields for every engine: named or not, position, description, errors and cited sources. Comparing the engines shows whether a problem sits in one assistant or in the sources they all read.

Is there a tool that tracks how often my brand gets recommended in ChatGPT?

Yes. Several AI visibility tools run your prompts on a schedule and report how often ChatGPT names or recommends your brand, usually as a mention rate or a share of voice against competitors. Check that the tool lets you set the prompts, records whether each answer used search and stores the cited sources, because a recommendation count alone does not show what to change.