Skip to content

Why Your Brand Is Not Showing Up in ChatGPT

Six reasons a company is missing from ChatGPT answers, in the order worth checking, starting with the free one most people never look at.

You typed your company name into ChatGPT, or asked it who to use for the thing you sell, and you were not there. Your competitor was.

Here are the six reasons that happens, ordered by how quickly you can check them and how often they turn out to be the answer.

1. You are blocking the crawlers (check this first, it is free)

Open yourdomain.com/robots.txt and look for these five names:

  • GPTBot (OpenAI)
  • OAI-SearchBot (OpenAI search)
  • ClaudeBot (Anthropic)
  • PerplexityBot (Perplexity)
  • Google-Extended (Google AI)

If any of them is followed by Disallow: /, that engine has been told not to read your site, and nothing else on this list matters until you change it.

This is more common than it sounds. The blocks get added during AI-scraping panics, by security vendors as a default, or by an agency that made a reasonable call two years ago and never revisited it. Nobody tells the person watching the revenue number.

It takes two minutes to check and is the best-documented lever in the whole field.

2. You might be showing up, and checked wrong

The most common false alarm.

Ask an AI engine the same question twice and you can get different companies named. These systems sample. They are not lookup tables returning a fixed ranking. If you asked once and were absent, you have one data point from a distribution, not a finding.

There is a second version of this mistake: asking in a chat where you have already been talking about your company. The model leans on the conversation, tells you what you want to hear, and you conclude you are fine. Use a temporary or incognito chat.

Ask each question at least three times, in clean sessions, across more than one engine. The full method is here, it is free, and it takes an afternoon.

3. The pages the engine reads do not mention you

This is the usual answer for commercial questions, and the one people least want to hear.

For “who should I use for X”, ChatGPT is not evaluating your website and forming a view. It is reading a small set of third-party pages, often the same fifteen to twenty URLs for an entire category, and assembling a shortlist from them. Typically:

  • “Best X” and “top ten X” listicles, frequently written by one of the vendors
  • Review platforms like G2 and Capterra
  • Reddit threads where someone asked your buyer’s exact question
  • YouTube explainers
  • Industry directories and roundups

If you are on none of those, you will not be in the answer regardless of how good your site is. If a competitor is on six of them, they get named constantly.

We proved this on ourselves the expensive way. We published three rounds of content on our own domain. Google impressions went up more than fivefold. Our share of AI answers went down, from 12.8% to 11.1%. Publishing on your own site does not buy you a place on someone else’s.

The fix is getting accurately represented on those sources, and the first step is finding out which ones they are, by writing down every source cited across thirty answers.

4. What the engine knows about you is wrong

Sometimes you are present and the problem is worse than absence. The engine describes a product you discontinued, quotes pricing from two years ago, attaches you to the wrong category, or repeats a criticism from a thread you never saw.

An engine that is unsure about you will drop you from a shortlist rather than risk being wrong. Ask directly: “what does [your company] do?”, “who is [your company] for?”, “what do people say about [your company]?” Then go and correct whatever is stale at the source. That is usually faster than earning a new mention and it removes a reason not to recommend you.

5. Your own pages do not say what you do plainly enough

Your site matters. It matters after the four things above, and in a specific way: the engine has to be able to lift a clear claim about you without guessing.

The most common failure is vagueness. “Solutions for modern business” gets you named on nothing. “Built for warehouses running 500 to 50,000 SKUs” gets you named on the queries that convert.

The second most common is the pronoun problem. “It integrates with most major ERPs” is useless once that sentence is pulled out of your page and dropped into an answer, because the model does not carry your paragraph’s subject with it. Name the thing in every claim you want quoted. More on this in optimizing content for LLMs.

The third is depth. We audited our own site and found 23 of our 27 articles were under 800 words, while every page of ours that reached the first page of Google on a contested term was over 1,100. Short pages are cheap to produce and easy to skip over.

6. Nobody is being named, and that is good news

Ask your category’s main question and read the answer carefully. If the engine hedges, describes a general approach, and names nobody in particular, you have not lost. The category is unsettled and the definitive source has not been written yet.

This is the best position on this list. It means the third-party page the engines will eventually rely on does not exist, and you can be the one who writes it. That is a much cheaper path than displacing an incumbent from six listicles.

What this does not need

It does not need more blog posts on your own domain. That is the instinct, it feels productive, and on the evidence it is the slowest lever available.

It does not need llms.txt. It is recommended everywhere and no major engine honours it. Google has said explicitly that it ignores it. We grade it as refuted in our evidence register.

And it does not need a subscription before you know whether you have a problem. The manual check is free.

The order to work in

  1. Check robots.txt for the five crawler names. Today, two minutes.
  2. Ask your ten buyer questions three times each across four engines, in clean chats, and write down every source cited.
  3. Correct anything the engines say about you that is wrong or stale.
  4. Claim your review-platform profiles and ask real customers for reviews.
  5. Contact the publishers of the two or three listicles that keep getting cited.
  6. Then improve your own pages: specific, quotable, deep enough to be worth citing.

Steps 1 to 4 cost close to nothing and are where the early movement is. Most companies start at step 6, which is why most companies wait a long time to see anything.

If you would rather see the diagnosis before doing any of it, run the free visibility check. Send your domain and two or three competitors and we will tell you which of the six above is actually your problem.

See where you stand in AI answers

We’ll run your domain through our tracker and send back a free one-page visibility snapshot — no call required.