You can't rank in ChatGPT the way you rank in Google. There is no public position to win, answers vary and no one controls them. You can make sure ChatGPT is allowed to read your site, that your site is clear, and that other sources describe you correctly. That improves your chances. It guarantees nothing.
Is there such a thing as a ChatGPT ranking?
Not in the way the word suggests. There's no list you climb and no report that says you're in position three. When ChatGPT searches the web, OpenAI says it ranks results using multiple factors meant to help people find relevant, reliable information, and adds: "Placement is not guaranteed." It doesn't publish those factors.
What the user sees is a written answer, shaped by their question and sometimes by their saved memories. Two people asking the same thing can get different answers. Anyone showing you a "ChatGPT rank" is showing you their own test method, which can be useful, but isn't an official position.
What does OpenAI actually require?
OpenAI documents access requirements, not a ranking formula. These are worth checking today:
- Allow OAI-SearchBot. Sites opted out of this crawler aren't shown in ChatGPT search answers, apart from possible navigational links.
- Check your host or CDN. OpenAI's help page says your host or CDN must allow traffic from its published crawler IP addresses. Security settings can block it without you knowing.
- Know the difference with GPTBot. Disallowing GPTBot signals that content shouldn't be used to train models. It's a separate setting, so you can block training and still allow search.
- Allow time. Search systems can take about 24 hours to adjust after a robots.txt change.
What can you influence?
This part is inference, not OpenAI policy. It follows from how the product works: ChatGPT turns a prompt into queries, retrieves results and writes from them. So it should help to be findable and easy to quote.
| Lever | Why we think it helps | How sure |
|---|---|---|
| Clear pages that answer buyers' questions | They are what retrieval can find and use | Reasonable inference |
| Specific facts in text | Vague claims give an answer nothing to repeat | Reasonable inference |
| Good position in regular search | OpenAI documents use of providers such as Bing in some cases | Partly documented |
| Accurate mentions elsewhere | Answers can draw on sources other than your own site | Reasonable inference |
For the longer version, read how to become a brand AI can recommend.
How do you know whether anything changes?
Use two measures. Ask a fixed list of questions every month, as described in how to check whether ChatGPT mentions you. And watch your analytics: OpenAI's publisher FAQ says ChatGPT adds utm_source=chatgpt.com to referral links, so you can see visits that arrive from it. Together they show presence and traffic. Neither gives you a rank, and a change can't always be traced to something you did.
What are the warning signs in an offer?
- A guarantee of mentions, positions or a share of answers
- A claim of a special relationship with OpenAI, unless they can show it
- A fixed trick, such as one file or one type of markup, sold as the solution
- Reports with a single score and no list of the questions behind it
- Bulk articles or planted mentions. Google warns that its spam policies apply to attempts to manipulate its AI answers, and we'd stay away from this everywhere
A good partner says what they'll change, how they'll measure it and that the outcome is not in their hands. That's also how we work: on our Visibility & Growth plan (€299 a month, excluding VAT) we monitor, report and make one defined improvement a month, without promising a mention.
