Track your brand in ChatGPT
Firemark asks ChatGPT your buyer questions on a schedule and records whether it names your brand, whether it links your site, and which competitors it names instead. ChatGPT is included in every plan, from Lite at $29 a month.
How Firemark tracks ChatGPT
We collect ChatGPT answers from the ChatGPT web app, logged out, from a US location, in English. We do not use the ChatGPT API.
Each prompt in your basket is asked once per run. The answer text and its sources are captured, then scored for your brand and for each competitor you track. If a capture comes back under 300 characters with no sources, we treat it as a failed capture and ask again, so a stub reply is never recorded as your brand being absent.
ChatGPT can answer the same prompt differently from one ask to the next. We ask each prompt once per run and report the counts as measured. To read movement, look at the trend over several runs. A movement is only reported after three comparable runs and a change of at least 5 points.
- Logged outNo personal account or history shapes the answer.
- US location, EnglishThe same location and language on every run, so runs can be compared.
- Checked before each runThree prompts are tried first. If ChatGPT returns nothing, it is left out of that run and shown as short instead of counted as zero.
Firemark reads ChatGPT's web app in a logged-out session from the US, so the answers are what a first-time visitor would see, and it reports how often your brand is named and linked across your prompts.
What you see for ChatGPT
For ChatGPT you see how many of your prompts named you, how many also linked your site, where you rank against your competitors and which sites ChatGPT cited.
- Named and linked, counted separatelyNamed means your brand is in the answer text. Linked means your site is among the sources. A link on its own never counts as a mention.
- Rank against competitorsYour rank among the 1 to 10 competitors you track, run after run.
- Funnel stageEvery prompt carries a stage from top to bottom of the funnel, so you can see whether ChatGPT names you when buyers are still researching or only when they compare.
- SourcesThe sites ChatGPT cites, split between answers where you are named and answers where a competitor is named and you are not.
- ChatGPT's share beside the othersYour share on each engine, with a 95% range once you track 20 or more prompts.
How ChatGPT answers to 125 prompts split by verdict
Each square is one answer. Demo data.
Prompt: What is the best CRM for a small retail team?
For a small retail team, Tool A and Your Brand come up most often. Your Brand suits teams that want customer data and in-store tools in one place, while Tool A is stronger on email.
Illustrative text written for this page. In the product, a snippet is a sentence of up to 500 characters taken from a real answer.
What is different about ChatGPT
ChatGPT is the one engine we read logged out, and it is in every plan, so it is the first engine most brands track.
- We read the app, not the APIThe ChatGPT API and the web app are different products, and answers can differ between them. The web app is where buyers ask, so that is what we read.
- Signed-in sessions are not measuredA person who is signed in to ChatGPT, with their own settings and history, may get different answers. We measure the logged-out experience and say so.
- A card is not a mentionChatGPT can put a single link under a vendor heading as a card, and we read that line as part of the answer. If your name appears only in a list of sources, such as a Sources heading or a block of link-only lines, the answer is not counted as naming you.
- Same cadence as every engineChatGPT runs with the others on your schedule, daily, weekly or monthly up to your plan, overnight US Eastern time.
How do I see if ChatGPT mentions my brand?
Ask ChatGPT your buyer questions and read whether your brand is in the answer, then repeat that across many prompts, because one answer can differ from the next. Firemark does this on a schedule and totals the result.
Checking by hand gives you one answer to one question from your own session, which may be shaped by your history and settings. Firemark asks every prompt in your basket from a logged-out US session, scores each answer for your brand and each competitor, and reports how many answers named you, how many linked your site and who ranks above you.
If you are looking for a ChatGPT rank tracker, note that ChatGPT answers have no fixed positions. The count of answers that name you, and your rank among the competitors you track, are the closest equivalent.
Plans that include ChatGPT
All four plans include ChatGPT, so you can start on Lite at $29 a month.
| Plan | Price | Engines | ChatGPT |
|---|---|---|---|
| Lite | $29 a month | 3 engines | Included |
| Starter | $79 a month | 4 engines | Included |
| Growth | $199 a month | 5 engines | Included |
| Pro | $499 a month | 5 engines | Included |
Lite also covers AI Overviews and AI Mode. Starter adds Gemini, and Growth and Pro add Perplexity.
Limits we state
Tracking ChatGPT (web) covers United States English only, and it is a sample of answers, not a census of everything people ask.
- United States and English only. Other countries and languages are not tracked.
- The counts describe the prompts you track, asked once each per run. They do not describe every question anyone could ask.
- We read ChatGPT logged out. Signed-in sessions, memory and custom instructions are not measured.
- Each prompt is asked once per run, so a single answer is one sample of a range ChatGPT could give.
- Claude is not tracked yet. It is planned for Pro.
Track the other engines beside ChatGPT
Each engine has its own counts, so you can see where you are named on one and missing on another.
ChatGPT tracking FAQ
The basics
Which AI engines does Firemark track?
ChatGPT (web, logged out), Perplexity, Google AI Overviews, Google AI Mode and Gemini, in US English. Lite covers ChatGPT, AI Overviews and AI Mode, Starter adds Gemini, and Growth and Pro cover all five. Claude is coming.
Why is ChatGPT tracked logged out?
Firemark reads ChatGPT's web app in a logged-out session from the US, so no personal account or history shapes the answer. It does not use the ChatGPT API. Single answers vary from one ask to the next, so counts are reported as measured.
ChatGPT gives different answers each time. How can the counts mean anything?
Each prompt is asked once per run, and the counts are reported as measured. One answer can differ from the next, so the useful reading is your share across many prompts and the trend over several runs. A movement is only reported after three comparable runs and a change of at least 5 points.
Why do you read the ChatGPT web app and not the API?
Buyers use the web app, and the API is a different product whose answers can differ. Firemark reads the web app and does not offer an API route for ChatGPT.
Does tracking include answers for signed-in ChatGPT users?
No. Answers are read from a logged-out session in the US. A signed-in person, with their own settings and history, may see different answers, and we do not measure that.
Does ChatGPT cite its sources, and can I see them?
ChatGPT can show sources under an answer. Firemark captures them and counts your brand as linked when your site is among them. The Sources screen lists the sites ChatGPT cited, split between answers that name you and answers that name a competitor and not you. A link on its own never counts as a mention.
Plans and billing
Which plans include ChatGPT?
All four. Lite at $29 a month covers ChatGPT, Google AI Overviews and Google AI Mode, and the larger plans add Gemini and Perplexity.
Find out if ChatGPT names you
Start tracking your brand, or open the demo to see a full report first.