How ChatGPT decides which websites to show
When a question calls for it, ChatGPT can search the web and answer with links to its sources. OpenAI explains that citations appear inside the answer, and that there is also a “Sources” list with other related links.
For that job it uses its own crawler, OAI-SearchBot, which according to OpenAI’s documentation is used to surface websites in ChatGPT’s search features. OpenAI also makes clear that any public website can appear in ChatGPT search.
What does not exist is a list you can sign up to, or a position you can buy. Results and citations can be incomplete or out of date, so it is best not to treat them as a stable ranking.
Step 1: do not block OpenAI’s crawlers
This is the simplest step and the most important one. OpenAI uses several crawlers with different jobs, and each one is controlled separately in your website’s robots.txt file.
| Crawler | What it is for | If you block it |
|---|---|---|
| OAI-SearchBot | Showing websites in ChatGPT search | You do not appear in its answers (at most, as a navigation link) |
| GPTBot | Collecting content that may be used to train models | You signal that your content should not be used for training |
| ChatGPT-User | Visiting a page when a user asks for it in ChatGPT | robots.txt rules may not apply, because a person starts the visit |
Open your robots.txt file (it lives at yourwebsite.com/robots.txt) and check there is no “Disallow: /” rule for OAI-SearchBot or for all crawlers (“User-agent: *”).
If you want to make it explicit, add “User-agent: OAI-SearchBot” followed by “Allow: /”.
Decide separately what to do about GPTBot. Allowing it or not is a decision about training and has no effect on whether you appear in search.
Check that your firewall, security plugin or CDN is not blocking these crawlers. OpenAI also recommends allowing requests from its published IP ranges.
How long changes take to show
OpenAI says its systems take about 24 hours to adjust after you update robots.txt. That is the timescale for crawl permission, not a promise of appearing.
Step 2: write pages that answer, not pages that decorate
An assistant that searches the web needs passages it can read, understand and attribute to a source. A website full of vague phrases (“comprehensive tailor-made solutions”) gives it little to cite. A page that answers one specific question does.
- One page per service, saying what is included, who it is for, which area you cover and how to hire you.
- A direct answer at the top of each page: two or three sentences that make sense out of context.
- Real questions from your customers (timescales, requirements, what happens if…) answered in visible text, not hidden in images or a PDF.
- Facts that can be checked: prices or how they are calculated, opening hours, address or service area, terms and conditions.
All of this is also good SEO, and that is no coincidence: AI that searches the web starts from pages that can be crawled and read. If your website does not have that foundation yet, start with how to rank your website on Google.
Step 3: make your details say the same thing everywhere
An assistant cross-checks sources. If your website says you open Monday to Friday, your Google listing says something else and an old directory shows a different phone number, the picture becomes confusing. Go through these, in this order:
Business name, address or service area, phone number and email, written identically on your website, your Google listing and your profiles.
Services and opening hours kept up to date in all those places.
A contact page and an “about us” page with real, verifiable details.
If you serve customers in person, a complete, up-to-date Google Business Profile.
What not to do
Do not invent reviews, awards, clients or figures to look more citable. Besides being misleading, it is exactly what cross-checking sources exposes.
How to check whether it is working
There is no official “ranking in ChatGPT” dashboard, but there are useful clues. OpenAI says ChatGPT adds the parameter utm_source=chatgpt.com to the links it sends to websites, so in your analytics tool you can see the visits that arrive from there.
- Look in your analytics for the source “chatgpt.com” and which pages receive those visits.
- If you have access, check your server logs to see whether OAI-SearchBot visits your site.
- Ask ChatGPT what a customer would ask (“X company in my city”) and note which sources it cites. Repeat on different days, because answers change.
There is a step-by-step method in how to tell if ChatGPT or Gemini recommend your business.
What nobody can promise you
Appearing in ChatGPT search depends on many factors you do not control: what is asked, which other sources exist and how each answer is generated. Anyone who guarantees you a mention, or a specific position, is promising something that is not in their hands.
What is in your hands is the technical and editorial side: do not block, explain things well and keep your details up to date. That is the line of work of our GEO and AI visibility service, always without guaranteeing mentions.
Frequently asked questions
Do I have to pay to appear in ChatGPT?
No. OpenAI’s documentation says any public website can appear in ChatGPT search and describes no paid sign-up. What you control is not blocking OAI-SearchBot and offering clear, reliable content that is easy to read and to attribute.
Does blocking GPTBot remove me from ChatGPT?
No. GPTBot and OAI-SearchBot are different crawlers and are configured separately. Blocking GPTBot signals that your content should not be used to train models; to appear in search, what matters is not blocking OAI-SearchBot.
How long does it take for my website to appear?
There is no guaranteed timescale. OpenAI only says its systems take about 24 hours to adjust to a change in your robots.txt. Whether a page ends up cited depends on each question and the sources available, and it may not happen.
Is a llms.txt file any use?
OpenAI does not list it as a requirement in its crawler documentation. It is a community proposal, not a standard adopted by search engines. You will find the detail in the guide on what llms.txt is.
