Groptim
All articles

How to rank in ChatGPT when there's no page one: who decides which business gets named

No one can honestly promise ChatGPT will name your business. Here's what OpenAI documents, what studies link to AI mentions, and a self-test you can run.

Groptim12 min read

The short answer

There is no page one in ChatGPT, so there is no fixed position to rank for. ChatGPT answers from what its model learned in training and, when it searches the web, from live pages it picks for that one question. Nobody outside OpenAI can promise you a place in either.

What you can do: make sure your website's settings do not shut ChatGPT's search reader out, be accurately present on the sites AI answers tend to draw from, and check the answers yourself, more than once. Each step says how well it is proven:

  • Documented guidance: the company that runs the AI says it in its own documentation.
  • Research-supported practice: large studies link it to being mentioned. That is a correlation: the two tend to go together, which does not prove one causes the other.
  • Recommended next step: sensible, but not proven.

There is no page one in ChatGPT: how are answers put together?

On Google you can picture a list of links and a top spot. ChatGPT gives one written answer, drawn from two places.

The first is training, the phase in which an AI model reads large amounts of text before anyone uses it. OpenAI's crawler documentation says its GPTBot crawler (an automated program that reads web pages) collects content "that may be used in training our generative AI foundation models". A business can be named from that memory, with no link.

The second is live search. The same page says another crawler, OAI-SearchBot, "is used to surface websites in search results in ChatGPT's search features". Searched answers can show pages as sources, with links.

Neither is a ranked list you can climb. Writing about AI search in general, a Microsoft Bing product manager said on 8 October 2025 that "ranking still happens, but it's less about ordering entire pages", with answers "often drawing from multiple sources".

Answers also change between asks. A small Ahrefs experiment on its own conference pages, published on 6 July 2026, tracking nearly 10,000 answers from ChatGPT, Gemini, Perplexity and Copilot, found that "between a page's first and last citation, it was cited on only about one in three eligible days."

So the useful questions are: how often am I named, next to whom, and where does the answer get its information?

Step 1: does your website shut AI search out?

Label: documented guidance.

Many websites have a robots.txt file: a short text file at yoursite.com/robots.txt that tells crawlers which pages they may read. Each crawler announces a name, called a user agent, and the file can allow or block each name separately.

OpenAI documents several crawler names. Three decide search, training and visits a person starts (OpenAI crawler documentation, undated, checked 9 October 2026):

  • OAI-SearchBot (search). "Sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers", though they "can still appear as navigational links", a term the page does not define. OpenAI recommends allowing it and says a change "can take ~24 hours" to take effect.
  • GPTBot (training). Blocking it "indicates a site's content should not be used in training generative AI foundation models." OpenAI says you can allow OAI-SearchBot while blocking GPTBot.
  • ChatGPT-User (a person's request). Because a person starts these visits, "robots.txt rules may not apply", and OpenAI says it "is not used to determine whether content may appear in Search."

Two things follow. If your file blocks OAI-SearchBot, removing that rule lifts the opt-out. OpenAI also asks sites to allow requests from its published addresses (OpenAI crawler documentation), so check that your host or firewall does not turn its reader away. Neither step creates a mention. Allowing the search reader makes you eligible, not chosen.

And you do not need to add a robots.txt file to be eligible. A missing file sets no restrictions. Google, for example, treats a missing file "as if a valid robots.txt file didn't exist" (Google's robots.txt specification, updated 31 August 2026).

Training is not search

At OpenAI, Google and Anthropic, the setting that opts you out of AI training is separate from the one their search uses, so opting out of training does not opt you out of their search. If you once blocked "AI bots", check which names you blocked:

CompanySearch readerTraining settingVisits a person triggers
OpenAIOAI-SearchBotGPTBotChatGPT-User
GoogleGooglebot (Google Search)Google-ExtendedNot covered here
Anthropic (Claude)Claude-SearchBotClaudeBotClaude-User
PerplexityPerplexityBotNone listedPerplexity-User
  • Google. Googlebot settings "affect Google Search" (Google's crawler list, updated 14 July 2026). To appear as a link in AI Overviews or AI Mode, "a page must be indexed and eligible to be shown in Google Search with a snippet" (Google's AI features page, updated 10 December 2025). Google-Extended covers Gemini training and also grounding, meaning handing Google Search content to Gemini as it answers, in Gemini Apps and some Google Cloud products. So blocking it can change some Gemini answers. Google says it "does not impact a site's inclusion in Google Search".
  • Anthropic. ClaudeBot is for training. Blocking Claude-SearchBot "may reduce your site's visibility and accuracy in user search results" (Anthropic's crawler article, dated 7 April 2026).
  • Perplexity. PerplexityBot surfaces sites in Perplexity's search and "is not used to crawl content for AI foundation models". Perplexity-User "generally ignores robots.txt rules" (Perplexity's crawler page, undated, checked 9 October 2026).

Our free check flags it, by name, when it can read your website's settings and they tell ChatGPT's, Perplexity's or Google's search reader to stay off your homepage. It does not report training-only settings or Claude's readers, so check those against the table above.

Step 2: are you present and accurate where AI answers draw from?

Label: research-supported practice. A correlation, not a promise.

AI answers lean on some sites more than others. Profound, a company that sells AI-visibility tracking, analysed 680 million citations (sources an AI answer points to) from ChatGPT, Google AI Overviews and Perplexity, collected from August 2024 to June 2025 (published 5 June 2025, updated August 2025). The top sites by share of all citations:

EngineMost-cited sites (share of all citations)
ChatGPTWikipedia 7.8%, Reddit 1.8%, Forbes 1.1%, G2 1.1%
Google AI OverviewsReddit 2.2%, YouTube 1.9%, Quora 1.5%, LinkedIn 1.3%
PerplexityReddit 6.6%, YouTube 2.0%, Gartner 1.0%, Yelp 0.8%

These are shares of the whole dataset, not of your kind of business. Even the top site gets fewer than 8 in every 100 citations, so most go to a long tail of other sites. And the snapshot ends in June 2025.

A second study looked at brands. SEO tool maker Ahrefs analysed 75,000 brands (26 May 2025) in Google's AI Overviews, the AI summaries shown above some Google results. It did not measure ChatGPT. How often a brand is mentioned across the web was the strongest signal it found, with a correlation of 0.664, against 0.218 for backlinks (links from other sites). On this scale, 0 means no relationship and 1 means the two always rise together. Ahrefs itself writes "correlation ≠ causation" and calls every factor it studied "moderate to very weak".

What that supports doing:

  1. Find your own list. Ask AI assistants your buyers' questions and note which sites the answers link to. That list beats any industry-wide table.
  2. Be accurate where you genuinely belong. On the review sites, forums and industry publications that appear for your questions, claim or correct your listing and keep it complete.
  3. Earn real mentions. Genuine reviews from real customers, useful community answers within each site's self-promotion rules, and coverage that names you because you are relevant.
  4. Industry directories are a recommended next step only. We have found nothing showing they lead to AI mentions.

Being listed somewhere is not shown to cause a mention. These studies show what tends to go together, not a switch.

Step 3: are your business facts consistent everywhere?

Label: recommended next step. Plausible, unproven.

Check that your name, what you do, who you serve, where you are, your prices and your opening hours read the same on your website, your profiles and your Step 2 listings. Let one page on your site, usually About, say plainly what you do and for whom.

Why bother? When sources disagree, an answer can repeat the wrong version, and fixing it costs little. We have not found a large study that isolates this effect, so treat it as housekeeping, not a growth lever.

Why doesn't ChatGPT recommend my business?

There is no single reason, and nobody can name yours without checking. These are the common possibilities, each with its label:

  • Your settings keep its search reader out. Documented guidance: blocking OAI-SearchBot keeps your pages out of ChatGPT search answers. You can check this today.
  • Few other sites mention you. Research-supported practice, from the Ahrefs AI Overviews study. It did not measure ChatGPT and does not show cause.
  • The sites answers draw on don't include you. Research-supported practice: see Step 2.
  • Answers vary between asks. Observed, not a step to take: in a small Ahrefs experiment, a cited page was cited on only about one in three eligible days. One answer is not the whole picture.
  • Your words don't match your buyers' words. Recommended next step: if buyers ask for "kitchen cabinet makers" and your site only says "bespoke joinery", use both.
  • Your facts are inconsistent or out of date. Recommended next step: see Step 3.

None of these is a diagnosis. The self-test below is how you find out which ones apply to you.

What should you not pay for?

  • Bought reviews. They mislead the very buyers you want to win. Ask real customers instead.
  • Link packages. Google's spam policies list "Exchanging money for links, or posts that contain links" as link spam (Google spam policies, updated 28 August 2026).
  • Mass submissions to generic directories. We have found no evidence they help.
  • Self-ranking "best of" lists, where a business ranks itself first among its competitors. Research finds they still get cited. In Ahrefs' study of 750 prompts (4 December 2025), "best X" lists made up 43.8% of all page types among the sources, and a self-promotional list "still showed up in more than a third of ChatGPT responses in the software category when its publisher was recommended". That study's author still saw a case for publishing them. But in Ahrefs' 2026 experiment, "among the answers that cited one of the conference-promoting pages, 43% never mentioned" the conference being promoted, and citations came and went. It is a manipulation pattern we do not build strategy on, and it can stop working without notice.
  • llms.txt or schema markup sold as citation fixes. llms.txt is a proposed text file that summarises a site for AI tools. Schema markup is code that labels facts on a page, such as prices or opening hours, for machines. Google says its AI features need no special files or structured data, and we found nothing in OpenAI's, Microsoft Bing's, Anthropic's or Perplexity's guidance for website owners saying their answers use llms.txt. Schema markup can help search engines understand a page (Microsoft Bing says the same for its AI answers), so it is ordinary site hygiene, not a citation fix we sell. Google: "You don't need to create new machine readable files, AI text files, or markup to appear in these features" (AI features page). Microsoft: "Schema is a type of code that helps search engines and AI systems understand your content" (Bing, 8 October 2025).
  • Promised mentions. Nobody outside the AI company decides which businesses its answers name, so nobody can sell you a mention. Where an AI product sells ads, an ad is not the same as being named in the answer.

How do you check it yourself?

Label: recommended next step.

  1. Write five to ten questions your buyers would really ask, such as "best [your kind of business] in [your city]", "[your business] or [a competitor]?" or "is [your business] any good?".
  2. Ask each one in a new chat, with memory or personalisation off where you can (for example, a temporary chat), so earlier conversations do not steer the answer.
  3. Ask more than one assistant: ChatGPT, Perplexity, Gemini, Claude and Copilot all answer differently.
  4. Ask each question at least three times, on different days.
  5. Log every answer:
QuestionAssistantDateNamed?Who else was named?Sites linkedAnything wrong about you?
[question][assistant][date][yes or no][names][sites][notes]
  1. Count how often you are named out of all tries, not where you appear. "Named in 4 of 15 answers" tells you more than one lucky screenshot.

Rerun the same questions after each change, so you compare like with like.

FAQ

Does ChatGPT use Google search?

OpenAI's crawler documentation for website owners does not mention Google or Bing, and does not say which search providers ChatGPT search uses. What it does document is OpenAI's own crawler, OAI-SearchBot, "used to surface websites in search results in ChatGPT's search features". Sites that block it "will not be shown in ChatGPT search answers", and OpenAI asks sites to allow requests from its published addresses (OpenAI crawler documentation). So the part OpenAI documents for you to control is whether OAI-SearchBot can read your site.

How long do changes take?

OpenAI says robots.txt changes "can take ~24 hours" to reach its search, and Perplexity says up to 24 hours. For mentions, nobody can honestly give you a timeline.

Should I block GPTBot?

That is a choice about training. It does not opt you out of ChatGPT search answers: OpenAI documents that you can allow OAI-SearchBot while blocking GPTBot. OpenAI does not document how it changes what future models know about you.

Where Groptim fits

Groptim's full audit report comes with a one-time pass that never renews. It shows measured mentions in AI answers to the questions buyers ask, next to your competitors, and, where usable evidence exists, where Google shows you and which sources to work on. These are answers we collect from AI models for your buyers' questions, not screenshots from inside any AI app. Every fix carries the same three labels used in this article. It cannot promise you a mention; nothing honest can.

Sources (checked 2026-10-09)