Search this question and nearly every result opens the same way: check your robots.txt, you have probably blocked GPTBot.
The data does not support that advice. A BuzzStream analysis published March 19, 2026, covering 4 million citations across 3,600 prompts, found that sites blocking GPTBot were still cited 88.2% of the time, and sites blocking OAI-SearchBot were still cited 82.4% of the time. The study's conclusion was that the relationship between crawler access and AI citation is far weaker than most publishers believe.
So if blocked crawlers are not usually the reason, what is? Here are the actual causes, in the order worth checking them.
Reason 1: You are not in Bing's index
This is the highest-value check and the one most guides bury below robots.txt.
ChatGPT's live search does not run on Google. It leans on Bing's index, and the correlation is stark: roughly 87% of ChatGPT citations match Bing's top results, while only about 10% of pages ChatGPT cites rank in Google's top 10. You can be dominant in Google and invisible in ChatGPT.
What to do: verify your site in Bing Webmaster Tools, submit your sitemap there, and confirm your priority pages actually show as indexed. Bing Webmaster Tools lets you import directly from Google Search Console, so this is typically fifteen minutes of work.
Tip
Fast diagnostic: search site:yourdomain.com in Bing, not Google. If your pages are missing there, you have found your problem, and no amount of content restructuring will fix it until Bing has you.
Reason 2: You are blocking the wrong bot (and it does not matter much anyway)
OpenAI runs multiple agents that do different jobs, and conflating them causes most of the confusion here:
| Agent | Purpose | Blocking it means |
|---|---|---|
| GPTBot | Crawls content to train OpenAI's models | Your content is excluded from training data |
| OAI-SearchBot | Powers ChatGPT's live search results | Your pages are excluded from search retrieval |
| ChatGPT-User | Fetches a page when a user's question requires it | Live user-initiated fetches are refused |
These are independent. Blocking one has zero effect on the others.
Here is the telling pattern: when sites block something ChatGPT-related, it is overwhelmingly GPTBot — one analysis counted 108 full blocks against GPTBot versus 3 against OAI-SearchBot. Publishers were protecting their content from training, which is a legitimate business decision, and search access was never what they restricted.
Meanwhile GPTBot accounts for 12.5% of AI bot traffic and OAI-SearchBot only 2.2%, and blocking of GPTBot among the top 1,000 websites has climbed from 5% in early 2023 to 25% today.
What to do: open yourdomain.com/robots.txt and read it. If you want training excluded but search included, disallow GPTBot and explicitly allow OAI-SearchBot and ChatGPT-User. If you want maximum visibility, allow all three. Just do not expect unblocking alone to produce citations — the data above says it usually will not.
Reason 3: Nothing outside your site says anything about you
This is the cause that actually explains the 88.2% figure. If sites that block crawlers still get cited, the citation is frequently not coming from a crawl of their pages at all. It is coming from somewhere else describing them.
Models lean on sources they treat as consensus: Reddit threads, industry publications, comparison pages, review sites. Reddit is the single most-cited domain across every major AI platform. If your category is discussed in those places and your brand never appears, the model has nothing to retrieve about you regardless of how good your own pages are.
This is also why the fix is slower than a technical one. You cannot ship third-party presence in an afternoon.
Watch out
Do not solve this by manufacturing mentions. Google's guidance explicitly names seeking inauthentic mentions as ineffective, and spam systems target it. Bought brand-drops get filtered; genuine participation compounds. The shortcut here is the one reliable way to make the problem permanent.
We run this diagnosis end to end — Bing indexation, crawler configuration, and a prompt panel showing which competitors get named where you do not. You keep the findings whether or not you work with us.
Reason 4: Nothing is broken — this is just the default
Before assuming a fault, calibrate. One index found roughly 94.8% of sites never appear in AI answers at all.
Absence is the statistical norm, not a symptom. This matters diagnostically: if you approach it as "something broke," you will spend weeks chasing fixes for problems you do not have. The accurate framing is that you have not yet cleared a bar that almost nobody clears, and that bar is mostly about Bing indexation, genuine authority, and third-party presence.
Reason 5: Your content is not extractable
If you clear the gates above and still are not cited, the remaining variable is usually the writing itself.
ChatGPT cites specific, quotable statements. Long narrative paragraphs that build toward a conclusion give it nothing to lift. Pages that answer a question in one clean sentence get cited even when they rank poorly.
What to do: lead each section with a direct one-to-two sentence answer, make sentences self-contained, and attach a statistic with a named source to each meaningful claim. Our guide on how to rank on ChatGPT covers this in depth.
The diagnostic order
Work top to bottom and stop when you find the problem:
- Search
site:yourdomain.comin Bing. Missing? That is your answer. Verify in Bing Webmaster Tools and submit your sitemap. - Read your robots.txt. Identify which OpenAI agent is blocked, and decide deliberately whether you want training excluded, search excluded, or neither.
- Check for a bot filter above robots.txt. Cloudflare and similar WAFs can block AI agents before robots.txt is ever consulted — check your firewall rules and any "block AI scrapers" toggle.
- Confirm content renders server-side. If it only appears after JavaScript executes, assume some agents never see it.
- Ask ChatGPT about your category, not your brand. If it names three competitors and not you, that is a third-party presence problem, not a technical one.
- Audit extractability. Does any single sentence on your key page answer a buyer question on its own?
- Wait weeks, not days. Engines re-crawl on their own schedule, and ChatGPT lags Perplexity noticeably.
The honest summary
The reason your website is not showing up in ChatGPT is most likely that you are not in Bing's index, that nothing outside your own site describes you, or simply that you are among the ~95% of sites that never appear. It is much less likely to be your robots.txt, despite that being the first thing almost everyone will tell you to check.
That distinction matters because the fixes are not equally expensive. Bing indexation is an afternoon. Third-party presence is a quarter. Knowing which one you actually need is the difference.
For the full playbook once you have diagnosed the cause, read how to rank on ChatGPT. For the cross-engine theory, our generative engine optimization guide, and for the Google side specifically, how to show up in AI Overviews — where the eligibility gate works the same way. If you are weighing whether to hire help, the seven agency tests are worth reading first.