Business Hub tells you whether ChatGPT can read your website. Ours says “Access confirmed”. If yours says something else, you’ll want to know what it means and what to do about it.
We can’t show you the failure screens, because we’ve never seen one. What we have is the code that the Business Hub page loads, which names six other statuses and three failure reasons.
This guide lists them, gives our reading of each, and sets out the checks we’d run. We don’t know how this status relates to what ChatGPT uses in its answers. A site its crawlers can’t read is worth fixing either way.
Get the free ChatGPT Profile Audit Kit
Where to find the status
Open the Overview page in Business Hub and look for Website access. Ours read “Access confirmed”, with the date it was last updated.
The statuses
| Status | What it means | What to do |
|---|---|---|
| Access confirmed | ChatGPT read your website | Nothing. Note the date. |
| Couldn’t read website | It tried and failed | Find the failure reason below |
| Access blocked | Your site refused it | Check robots.txt and your firewall |
| Not read yet | No attempt has finished | Wait, or request a check if your Overview page offers one |
| No website | The profile has no website | Add one in the Contact tab |
| Check requested | A check is queued | Wait |
| Checking | A check is running | Wait |
We’ve only seen “Access confirmed” ourselves. The other status names come from the page code, and the meanings beside them are our reading of those names. The wording on your screen may differ slightly.
The three failure reasons
The code names three reasons a read can fail. The names are OpenAI’s. The explanations and fixes are our reading of them.
robots_blocked
Your robots.txt file tells OpenAI’s crawlers to stay out.
How to check: open yourdomain.com/robots.txt and look for rules that disallow these user agents:
GPTBotOAI-SearchBotChatGPT-User
Also look for a blanket Disallow: / under User-agent: *. That blocks everything that doesn’t have its own rule.
How to fix it: remove the disallow rule, or add an explicit allow for those agents. OpenAI documents its crawlers and how to control them in its overview of crawlers.
This is the first cause we’d check. A block can come from a rule someone added on purpose, or from a plugin or security setting that added it for you.
robots_unreachable
ChatGPT couldn’t fetch your robots.txt file at all.
How to check: load yourdomain.com/robots.txt in a browser. You want it to load, or return a clean “not found”. A server error or a timeout is the problem.
How to fix it: find out why the file errors. Common causes are a misconfigured redirect, a server error on that path, or a firewall challenging the request.
fetch_failed
The robots.txt check passed, and the page itself couldn’t be fetched.
How to check:
- Does the website address on your profile load, with the full https address?
- Does it redirect several times before it lands?
- Does your firewall or bot protection challenge automated visitors?
- Does the page need JavaScript, a cookie banner, or a login before any content appears?
How to fix it: correct the address in the Contact tab, shorten the redirect chain, and ask whoever manages your firewall whether it’s blocking OpenAI’s crawlers.
Requesting a re-check
The code shows that a re-crawl can be requested from Business Hub. We haven’t seen the control ourselves. Two details are worth knowing:
- A re-crawl also visits eligible linked pages, not only the address on your profile.
- Requests have a cooldown.
The code names five outcomes for a request. The names are OpenAI’s. The meanings are our reading of them:
| Outcome | What it means |
|---|---|
| requested | The check is queued |
| already_fresh | Your site was read recently, so nothing was done |
| blocked | The request couldn’t go ahead because access is blocked |
| busy | Try again later |
| cooldown | You asked too recently |
So fix the cause first, then request the check. Asking again while the block is still in place won’t change the result.
A quick diagnosis
- Does the profile have a website? If not, add one.
- Does robots.txt load? If not, that’s
robots_unreachable. - Does robots.txt block GPTBot, OAI-SearchBot, or ChatGPT-User? If so, that’s
robots_blocked. - Does the page load quickly, without a challenge or a login? If not, that’s
fetch_failed. - Fix, then request a re-check if your Overview page offers one.
What “Access confirmed” doesn’t prove
It tells you ChatGPT read your site. It doesn’t tell you:
- Which pages it read.
- Whether it understood your prices, services, or hours.
- Whether it will mention you.
Access is the floor. What’s on the pages still has to be accurate and clear. Add Organization schema that matches your profile, and Product or Offer schema on your product and pricing pages.
One limit on testing this yourself: you can request a page while identifying as one of OpenAI’s crawlers, and that shows what your robots.txt and server allow. It doesn’t prove what a firewall does to requests from OpenAI’s own addresses. If your status says blocked and your robots.txt is clean, the firewall is the next place to look.
Website access is one line on the Overview page. Our ChatGPT business profiles guide covers the rest of the profile.
Get the free ChatGPT Profile Audit Kit
Sources and method
The statuses, failure reasons, re-crawl behavior, and request outcomes come from the code the Business Hub page loads in the browser, read on November 11, 2026. Code shows what has been built, which isn’t always what is switched on, and it can change without notice. “Access confirmed” is the only status we’ve seen on our own profile. The causes and fixes are our diagnosis from standard crawling practice. We don’t know which of OpenAI’s crawlers Business Hub uses for this check.