GEO & AI search

Get found in ChatGPT: What OpenAI officially says

SERP TOWN Editorial · 27 October 2026 · 12 min read

For ChatGPT to show your website in answers that use web search and link to it as a source, OpenAI's crawler OAI-SearchBot must be allowed to read your pages. According to OpenAI, this is a requirement for being found in ChatGPT, but not a guarantee. OpenAI itself writes that ChatGPT ranks search results using multiple factors and that a place in the results is not guaranteed. OpenAI does not disclose what those factors are.

In this guide you will learn how, according to OpenAI, ChatGPT finds websites and shows sources, what is known about this and what is not, and which steps you can take. What GEO is in general, and what a study on it found, is covered in our guide GEO: How your business shows up in ChatGPT, Gemini and co.. This guide is only about ChatGPT, and only as far as OpenAI describes it itself. Where we recommend something that OpenAI does not say, we tell you so.

How ChatGPT searches the web and shows sources

According to OpenAI's help page "Searching the web with ChatGPT" (as of October 2026), "ChatGPT can search the web to answer questions with current information and links to relevant sources". The search can start automatically when a question would benefit from current information. You can also turn it on yourself.

Sources in the answer: According to the same page, "Responses that use web search may include citations." Selecting a citation opens its source. With the Sources button you can see, when available, the cited sources and other relevant links. When ChatGPT search launched in October 2024, OpenAI already wrote in the announcement "Introducing ChatGPT search" that answers include links to sources such as news articles and blog posts. But OpenAI also warns: "Search results and citations can be incomplete, outdated, or incorrect."

Where the results come from: On the help page, OpenAI writes that "ChatGPT search sometimes partners with other search providers." ChatGPT then typically rewrites the question into one or more targeted queries and sends them to those providers. For details on how this is processed, OpenAI refers to the privacy statements of Microsoft and Shopify. The 2024 announcement mentions "third-party search providers" and content provided directly by partners. Which search index ChatGPT uses exactly, which provider is used when, and how the results are weighted is not stated on any of the OpenAI pages we checked.

How ChatGPT ranks: OpenAI only writes that ChatGPT ranks search results using multiple factors intended to help users find relevant, reliable information, and that "Placement is not guaranteed."

Local questions: According to OpenAI, ChatGPT may use an approximate location based on your IP address, for example for nearby restaurants, local news and weather. For this, ChatGPT may share the location with third-party providers that supply, for example, listings of nearby businesses. OpenAI does not name these providers on that page. We do not know whether your Google Business Profile is one of them. How to set up and maintain your listing on Google is explained in our guide Create and optimize your Google Business Profile.

OpenAI's crawlers and what they control

A crawler is a program that visits and reads websites automatically. Whether it is allowed to do so is something you set in your robots.txt. According to the internet standard RFC 9309, this text file sits in the top-level directory of your website, that is, at /robots.txt. OpenAI describes its crawlers in the Overview of OpenAI Crawlers (no date on the page). According to OpenAI, each setting is "independent of the others".

• OAI-SearchBot is for search. It is used so that websites can appear in ChatGPT's search results. Sites that opt out of OAI-SearchBot are not shown in ChatGPT search answers, though they can still appear as a navigational link, that is, a simple link to the website. OpenAI recommends allowing OAI-SearchBot in your robots.txt and allowing requests from OpenAI's published IP ranges.
• GPTBot collects content that may be used to train OpenAI's AI models. If you block GPTBot, you signal that your content should not be used for training. According to OpenAI, you can allow OAI-SearchBot and still block GPTBot.
• ChatGPT-User visits a page when users ask ChatGPT something and ChatGPT opens a page for it. It does not crawl the web automatically. Because a person triggers the request, robots.txt rules may not apply to it, according to OpenAI. ChatGPT-User does not control whether your content may appear in search. That is what OAI-SearchBot is for.
• OAI-AdsBot checks the landing pages of ads in ChatGPT. According to OpenAI, it only visits pages that were submitted as ads.

According to OpenAI, after a change to your robots.txt it can take about 24 hours for OpenAI's search systems to take it into account.

What is known if you want to be mentioned in ChatGPT

In our view, many people who search for ChatGPT SEO hope for a list of tricks. OpenAI does not provide one. This is what the sources we checked say:

Known, because OpenAI says so:

• For a website to be eligible for the search results, it should allow OAI-SearchBot. In addition, the host or the network that delivers the site (CDN) should allow requests from OpenAI's published IP addresses. This is stated on the help page about web search.
• A place in the results is not guaranteed. ChatGPT ranks using multiple factors.
• In the 2024 announcement, OpenAI wrote that any website or publisher can choose to appear in ChatGPT search.
• You control training (GPTBot) and search (OAI-SearchBot) separately.

Unknown, because OpenAI does not disclose it:

• which factors determine the order and how strongly they count,
• which search index ChatGPT uses exactly,
• when ChatGPT mentions a business in an answer without web search, for example from the model's knowledge,
• whether allowing GPTBot increases the chance of being mentioned. OpenAI describes GPTBot only for training.

What research says: The well-known study "GEO: Generative Engine Optimization" did not examine ChatGPT itself. The research team built its own test system: for each question it fetched the top five sources from Google Search and had the GPT-3.5 Turbo model write an answer from them. In addition, they tested their methods on Perplexity. What worked there is summarized in our GEO guide. Whether the results apply to today's search in ChatGPT has not been shown.

For online shops: On the page "Power product discovery in ChatGPT", OpenAI describes that merchants can apply to share their product data (a product feed) with ChatGPT. According to OpenAI, if you sell through Shopify or Etsy, you are already integrated. According to that page, shopping in ChatGPT is currently live for users in the U.S.; OpenAI plans to add more regions over time. OpenAI does not say there what this means for shops in German-speaking countries. More about shops on Google and in AI answers is on the page for the e-commerce district: SEO and GEO for online shops.

Steps for your business

Steps 1 to 3 are based on OpenAI; steps 4 and 5 are our recommendation.

1. Allow OAI-SearchBot in your robots.txt

Open your-domain.com/robots.txt in your browser and check whether it contains Disallow: / under User-agent: OAI-SearchBot. Also look for a group with User-agent: *: according to RFC 9309, a crawler follows this group if there is no group of its own for it. If your robots.txt blocks OAI-SearchBot in either of these ways, your website will not appear in ChatGPT's search answers, according to OpenAI. A rule that explicitly allows OAI-SearchBot everything can look like this, for example:

User-agent: OAI-SearchBot
Allow: /

This example is ours; OpenAI does not give it. OpenAI only recommends allowing OAI-SearchBot. And allow about 24 hours for a change to reach OpenAI.

2. Check your firewall, host and CDN

The robots.txt is only one part. OpenAI also lists as a requirement that your host or CDN allow requests from the published IP addresses of OAI-SearchBot. In its guidance for advertisers, OpenAI also writes: "A robots.txt rule alone may not be enough." According to OpenAI, this is because a firewall (a protective program that filters access), a CDN or a bot-protection service may apply restrictions of their own. This guidance applies to the landing pages of ads. Our recommendation (OpenAI does not say this): have someone check in the same way whether OAI-SearchBot can reach your normal pages without signing in, without a CAPTCHA and without a browser challenge.

3. Decide deliberately about GPTBot

Whether you allow GPTBot is a separate decision. According to OpenAI, it concerns the training of the models, not search. So you can appear in search answers and still refuse training. Which choice suits you depends on how you feel about your texts being used for AI training. OpenAI does not describe any disadvantage in search from blocking GPTBot.

4. Write the most important things clearly as text on your pages

Our recommendation (OpenAI does not say this): write on your website what you offer, for whom and where, and answer the questions customers really ask you. In our view, ChatGPT can only use from your page what is actually written there. What Google says about text on the page, and what the GEO study measured, is in the GEO guide. What this means for each business type is shown, for example, on the page for the Local Business district for businesses with a fixed address.

5. Back up your statements and keep them up to date

Our recommendation (OpenAI does not say this): only give numbers if you can back them up, link your sources, and keep details such as services and opening hours up to date. OpenAI itself advises its users to open a cited source and check that it supports the answer and when it was published or updated. In our view, a page with clear, supported and current information is therefore a good source, for people as well as for AI answers. That ChatGPT prefers such pages has not been shown.

How Perplexity, Claude and Gemini handle access to your website

Perplexity and Anthropic also describe separate crawlers, and Google has its own token for Gemini that you can use in your robots.txt. If you also want to be visible in Perplexity and other AI assistants, it is worth taking a look.

• Perplexity: According to Perplexity's crawler documentation, PerplexityBot is designed to surface and link websites in Perplexity's search results. According to Perplexity, it is not used to train AI models. Perplexity recommends allowing PerplexityBot and permitting its published IP ranges. Perplexity-User visits pages when users ask a question and, according to Perplexity, generally ignores robots.txt rules because a user requested the fetch. For firewalls such as Cloudflare's, Perplexity describes how to allow its crawlers. According to Perplexity, changes take up to 24 hours.
• Claude (Anthropic): In its help article on crawling the web (as of April 7, 2026), Anthropic describes three crawlers: ClaudeBot collects content that could contribute to training. Claude-User visits pages when users ask Claude something. Claude-SearchBot improves search results. According to Anthropic, blocking Claude-SearchBot or Claude-User may reduce your visibility in search results. According to Anthropic, you opt out through your robots.txt, and you should do this for every subdomain (such as shop.your-domain.com).
• Gemini (Google): According to Google's common crawlers (last updated October 15, 2026), the Google-Extended token controls whether content may be used for training future Gemini models and for so-called grounding in Gemini Apps. According to Google, grounding means the model is given content from the Google Search index when it answers. According to Google, it has no effect on Google Search.

Our recommendation (no provider says this): for each assistant, write down which crawler affects search, and check your robots.txt against this list once. That way you do not accidentally block search when you really only wanted to refuse training.

How to check whether ChatGPT mentions you

OpenAI does not name an official metric for your visibility in ChatGPT on the pages we checked. There are still a few clues. Our recommendation (OpenAI does not say this):

• Spot check with web search: Ask ChatGPT the questions your customers would ask, turn on search and look under Sources to see whether your website is included. This is not a measurement, only a snapshot. According to OpenAI, location, saved memories (if memory is turned on) and a VPN can influence the results.
• Server logs: OpenAI publishes OAI-SearchBot's identifier and its IP ranges. If you have access to your server's logs, you can see there whether OAI-SearchBot visits your pages. A visit only shows that it can read them, not that ChatGPT mentions you.
• Visits in your analytics: Whether and how visits from ChatGPT are labeled in tools such as Google Analytics is something we could not confirm in any OpenAI source for this guide. Look at your referral sources, but do not treat the result as a complete figure.

Be careful with tools that promise an exact "AI visibility": since OpenAI does not disclose the factors and answers change, in our view no tool can measure it completely.

ChatGPT SEO and GEO: How the two fit together

ChatGPT SEO is part of what is called GEO or SEO for AI: making sure that AI assistants can use your website as a source and link to it. For ChatGPT, the technical basis is clearly described; everything beyond that is not. What is known about Google's AI Overviews, the results of the GEO study and seven general steps are in the guide GEO: How your business shows up in ChatGPT, Gemini and co.. More guides are in the SERP TOWN Library.

Your next step

Want to know where your business stands on Google and in AI answers? The 3-question check from SERP TOWN shows you which business type you are and where your biggest levers on Google and in AI answers are. If you like, you can then send us a request: we will look at your website and send you a short teaser with a few tips.