Why ChatGPT doesn't mention your brand (and what to fix first)
Five common reasons ChatGPT leaves a brand out of its answers, how to check each one, and the order we fix them in.
You ask ChatGPT for the best tools in your category. It lists five competitors. You are not one of them. You try a different wording and get a different list, still without you.
Before fixing anything, get specific about the problem. "ChatGPT doesn't mention us" can mean it never names you, or it names you for some questions and not others, or it names you but describes you wrongly. Those are different problems. A handful of manual tests in your own account will not settle which one you have, because answers vary between runs and your chat history can shape what you see. You need the same set of buyer questions checked repeatedly over a few weeks before the pattern is reliable.
Once you have that, the causes usually fall into one of five groups. Here is each one and how to check for it.
1. OpenAI's crawlers can't reach your pages
ChatGPT answers some questions from what the model learned in training and others by searching the web at the time of the question. OpenAI's crawler documentation separates these. OAI-SearchBot is used to surface websites in ChatGPT's search features, and OpenAI states that sites opted out of it will not be shown in ChatGPT search answers. GPTBot crawls content that may be used to train future models. The two are controlled independently in robots.txt, so a site can allow one and block the other.
It is easy to block AI crawlers wholesale during a policy discussion and forget the rule is there. It is just as easy to never touch robots.txt and still sit behind a CDN or firewall that challenges unfamiliar bots.
How to check. Open yoursite.com/robots.txt and look for any rule naming OAI-SearchBot or GPTBot, plus any wildcard rule that would catch them. Then check your server or CDN logs for requests from OpenAI's user agents to the pages that matter, and look at the status codes. A 200 is fine. A 403, a challenge page or no requests at all is a problem. If you change robots.txt, OpenAI says it takes about 24 hours for its search systems to adjust, so don't retest the same afternoon.
2. There's no page that answers the question
Engines quote pages that answer the question being asked. If a buyer asks "which invoicing tools handle multi-currency for small agencies" and your site has a features page that mentions currencies in the ninth bullet, an engine has little reason to pick you over a competitor with a page built around exactly that question.
This is the first content gap we look for, because many sites are written as brochures. They describe the product in the company's own language and leave the buyer to work out whether it fits. Answer engines reward the opposite: the direct answer first, in the buyer's words, with specifics that can be quoted.
How to check. Take the prompts where you are absent and, for each one, try to name the single URL on your site that answers it. Be strict. If you need to combine three pages, or the answer is buried in a PDF, an image or a JavaScript tab, count it as missing. Then look at the pages ChatGPT did cite for those prompts and note how they are structured. You will often find comparison pages, use-case pages and plain FAQ answers that your site simply does not have.
3. You're absent from the sources ChatGPT cites
When ChatGPT searches, it builds its answer from a small number of pages, and it shows them as sources. For category questions those are often third-party pages: review sites, "best of" roundups, trade publications, Reddit threads, YouTube videos. If none of those pages mention you, the answer probably won't either, however good your own site is.
How to check. For each prompt where you are missing, list the cited domains and pages. Then check whether your brand appears on each one. Patterns appear quickly. The same two or three publications, or the same subreddit, may keep showing up across a whole group of prompts. Those are the outside sources worth working on, through genuine coverage, accurate listings and open participation in the communities concerned. We don't buy placements or post undisclosed promotion. It breaks most community rules, and it erodes the trust that made the source worth citing in the first place.
4. ChatGPT is confused about who you are
Some brands are left out because the engine does not have a clear picture of the company. This happens when the name is also a common word, when another company has a similar name, after a rebrand, or when your own descriptions disagree with each other across the site, LinkedIn and directory listings.
How to check. Ask ChatGPT direct brand questions: what does [Brand] do, who is it for, who are its competitors, where is it based. Repeat them several times. If the answers mix you up with another company, describe an old product, or put you in the wrong category, you have an entity problem. Fixing it is unglamorous: one consistent description of what the company does, used on your homepage, About page and the profiles you control; Organization structured data that matches that visible text and links to your official profiles; and clear notes on pages that mention a former name. Then fix the outside listings that still use outdated information.
5. You're testing the wrong prompts
Sometimes ChatGPT does mention you, just not for the questions you are checking. Your team may call the category one thing while buyers call it another. You may be testing "enterprise" prompts while your actual customers are mid-sized firms asking different questions. Or you are named in one country and absent in another, because the answer depends on the market.
How to check. Compare your test prompts with the words buyers actually use. Sales call notes, support tickets and Search Console queries are better sources than your positioning document. Look at the related sub-questions engines generate on the way to an answer, often called query fan-outs, because they show which phrasings the engine itself considers relevant. If your visibility jumps once the prompt set matches real buyer language, the problem was measurement, not the market.
What to fix first
Order matters, because some fixes are wasted until others are done.
Start by confirming the prompt set. If you are measuring the wrong questions, every later decision is built on a misleading number. Next, fix crawler access. It is usually the cheapest change, and no content work can help a page that OpenAI's search crawler cannot fetch. Then deal with entity confusion, since a page about a company the engine has misidentified is harder to place. After that, close the content gaps, revising existing pages before writing new ones. Outside citations come last in this list only because they are the slowest to influence, not because they matter least. Start that outreach early so it has time to land.
How we check all five in one place
We run these diagnostics on Promptwatch, the platform we run every score on, because it covers each check without stitching exports together. Prompt tracking monitors the real ChatGPT interface along with the other major engines, and shows query fan-outs, search volumes and personas for the prompt-mismatch check. Agent Analytics reads AI crawler logs from integrations such as Cloudflare, Vercel and Fastly, flags crawl errors, and traces the crawl-to-citation path for a page, which covers the access check. Content gap analysis lists the prompts where competitors are named and you are not. Citation analytics shows the cited domains and pages for each prompt, including Reddit and YouTube citations, which is how we find the third-party sources worth pursuing. For the entity check, reading the stored responses to brand prompts side by side, with sentiment attached, shows quickly whether the engine has the company right.
The value is less in any single report than in seeing which of the five causes applies to which prompts. That tells you what to fix first for your brand, rather than for a generic one.