Blog 02 / 09

AI search

Why is my website not appearing in ChatGPT search?

A missing ChatGPT mention rarely comes down to one piece of code. First separate search crawling from model training, then check whether your important pages are accessible, specific and genuinely useful enough to support an answer.

In brief:

To give a website the opportunity to appear in ChatGPT search, allow OAI-SearchBot, make important pages publicly accessible and indexable, and publish specific information that can support an answer. GPTBot controls potential training use; it does not control ChatGPT search visibility.

One prompt is not a reliable visibility test. Results can vary with the wording of the question, location, timing and whether web search is used. Begin with the questions a prospective customer would genuinely ask, then inspect whether the response cites current web sources.

OAI-SearchBot, GPTBot and ChatGPT-User are different

User agentWhat it doesWhat a site owner should know
OAI-SearchBotCrawls pages for ChatGPT search featuresAllow it in robots.txt and avoid blocking OpenAI’s published search crawler IPs if search visibility is wanted
GPTBotCrawls content that may be used to improve generative AI modelsTraining access can be allowed or disallowed independently
ChatGPT-UserVisits pages for certain user-initiated requestsIt is not an automatic web crawler and does not determine search inclusion

A business can allow OAI-SearchBot while disallowing GPTBot. The two choices do not need to match.

Why might a website not appear in ChatGPT search?

  1. The search crawler is blocked. A robots.txt rule, hosting firewall or security service may prevent access.
  2. The page is not indexable. A noindex directive, login wall, broken canonical link or server error can keep useful content out of search systems.
  3. The important information is hard to read. Company details held only in images, scripts or closed interface elements are less dependable than clear page text.
  4. The page is too vague. A polished slogan cannot answer questions about the service, location, process, suitability or cost.
  5. The business is described inconsistently. Conflicting names, addresses, phone numbers or service areas make the entity harder to identify.
  6. There is little reason to cite the page. Generic copy that repeats common advice offers less value than first-hand detail, transparent limits or a useful explanation.

What should you check first?

1. Crawl access

Review robots.txt for a rule that blocks OAI-SearchBot or all crawlers. Then check the hosting and security layer. A permissive robots.txt file does not help if requests are refused elsewhere.

2. The important pages

Open the home, service, about and contact pages without signing in. Each should load successfully and state the relevant information in ordinary text. Check canonical links, robots directives and status codes as part of the same review.

3. The substance of the content

Explain who the service is for, what it includes, the area covered, the process, meaningful limitations and the next step. This is useful for conventional search, AI search and—more importantly—the reader.

4. Identity and trust

Use a consistent business name and contact information. Add a genuine byline or publisher, a clear review date and links to primary sources when a factual claim may change.

Will allowing the crawler guarantee a mention?

No. Allowing access makes inclusion possible; it does not create an entitlement to a citation or recommendation. OpenAI notes that a robots.txt change can take roughly 24 hours to be reflected by its search systems, but that is an adjustment period, not a promise that the website will appear.

Likewise, adding structured data or an “AI optimisation” file cannot turn a thin page into a dependable source. The useful work is less mysterious: make the website accessible, specific and worth referring to.

How should you monitor progress?

Keep a small set of realistic, non-branded questions and review them periodically rather than every day. Record whether web search was used, which sources appeared and whether the website’s ordinary search visibility and enquiries are improving. AI answers are volatile; a single screenshot is not a strategy.

If the technical and content foundations need attention, our SEO and AI search service sets out the practical work we can do without promising a named citation.

Source: OpenAI: overview of crawlers and user agents.

Discuss your website