Mentions
Does ChatGPT mention my brand?
You can answer that for a specific chat. You cannot answer it once, for every user, from a free web form. Here is a way to look that does not pretend otherwise.
Open a fresh chat. Turn off extra tools if you want a plain answer, or leave them on if that is how your buyers use the product — just write down which you did. Ask one prompt from the category (“best … tools”), one alternatives prompt, and one “what is {brand}” prompt. Copy the whole reply.
Paste it into a Benchline report. The scorer tells you whether the brand appears as a whole word, how far into the text the first mention sits, and whether it is item 1 or item 7 in a list. Repeat the same prompt in a new chat if the answer will change a decision. One screenshot is a story. Three chats are still a sample, and they are a better sample.
Run a free check
Brand, website, and an optional category. You get a shareable report with a citability score, a prompt pack, and a paste scorer.
What a mention is, and is not
A mention is the brand string in that answer. It is not a recommendation, not a citation, and not proof the model “knows” you. A late mention in a list of twelve is still a mention. The paste score separates “named first” from “named at the bottom” so you do not flatten those.
Other names in the same list are possible competitors. The guess is mechanical. If it flags a sentence fragment, ignore it. You can also type the competitors you already track so the scorer looks for those spellings on purpose.
Why we will not show a live ChatGPT rank
Scraping the ChatGPT website would break their terms and still would not be stable. Calling an API is allowed when you have a key, and it answers a different product. Free GEO tools explains that split. The citability crawl remains useful either way: if GPTBot cannot fetch the site, you have removed one path a future answer could have used. Read how those robots rules work before you flip them.
Questions
Can Benchline see my ChatGPT account?
No. Paste the text yourself. The scorer runs in the browser and does not upload the answer.
Why did two chats disagree?
Chats differ by model, memory, browsing, and the exact wording of the prompt. A single “yes” or “no” is a sample, not a census.
Is the OpenAI API the same as the ChatGPT app?
No. If this site is configured to call the API, the reply is labeled as an API reply. Do not treat it as what a logged-in user sees.
What prompts should I try?
Use the prompt pack on your report: category queries, “alternatives to {brand}”, “what is {brand}”, and “is {brand} worth it”. Category queries are where many brands are absent even when the branded query works.
Related guides
- Free GEO tools you can use this week
A map of free checks: citability crawls, prompt lists, paste scoring, and the paid suites we do not copy.
- What is generative engine optimization?
A plain definition of GEO, what you can change on your site, and what no tool can promise.
- What is llms.txt?
The community file at /llms.txt, what Benchline looks for, and a minimal example you can adapt.