Ranking in ChatGPT is different from ranking in Google because ChatGPT reads differently. It composes answers from training data and a live search layer, and when asked who provides a service it leans on directories, industry roundups and review profiles. Much of what it cites for a category does not rank in Google at all. Being named means being present, consistently, across the sources it reads.
What ChatGPT cites, the five-step method, which OpenAI crawlers to allow, and how to measure it.
To be named by ChatGPT for a category, a company has to be present in what ChatGPT reads, which is only partly its own website. ChatGPT composes answers from training data and a live search layer, and when asked who provides a service it draws heavily on directories, industry roundups and review profiles. Ranking in Google helps, but much of what ChatGPT cites for a category does not rank for it. The rest is presence in third-party sources, consistent entity data across the web, and citable specifics on your own pages.
This is the practical difference from Google. An AI Overview mostly cites what already ranks, so SEO is most of the work. ChatGPT names companies it has encountered across the web in consistent form, and it will name a company with a weak website if that company appears in three directories, two roundups and a review profile with the same name and category each time. It will skip a company with a strong website that exists nowhere else.
The measurement is direct. Ask ChatGPT the question a buyer would ask, with the category and the market, and record which companies it names and which sources it cites. Repeat for Gemini, AI Mode and AI Overview, because the four disagree. We measured 44 ChatGPT mentions and 2 AI Overview appearances for one company in one week in August 2026.
For a category-and-market question in one industrial category we measured in October 2026, ChatGPT's citations split roughly a third company websites, a third directories and roundups, and the remainder review profiles, trade press and other. The company's own site was a minority. That proportion moves by category and by month, but the pattern, third-party sources outweighing first-party, held in every B2B category we have looked at.
The implication is uncomfortable for companies that have only ever worked on their own website: most of the surface ChatGPT reads is not under their control. It is under the control of directory editors, trade publications and reviewers. Being present there, with the same name, the same address, the same category description, is the ChatGPT-specific work. It does not recur monthly. It is done once, checked quarterly, and it compounds.
Measured by asking ChatGPT the category question ten ways, recording every cited source, and classifying each. One category, one month. Treat the percentages as a shape rather than a constant.
Five steps: rank for the capability terms, because that is the raw material; be listed in the directories and roundups ChatGPT cites for your category, which you find by asking it; keep name, address, phone and category identical everywhere; state figures, certifications and named results on your own pages so there is something to quote; then ask ChatGPT the buyer's question monthly and record who it names and what it cites.
Yes, if you want to be found by ChatGPT. OpenAI operates separate crawlers: GPTBot for training, OAI-SearchBot for ChatGPT search results, and ChatGPT-User for pages a user asks it to open. Blocking GPTBot does not stop ChatGPT search from citing you; blocking OAI-SearchBot does. Many sites block all three without knowing the difference and then measure zero ChatGPT visibility.
| Crawler | Purpose | Block it and |
|---|---|---|
| GPTBot | Training data collection | Future models learn less about you; current search is unaffected |
| OAI-SearchBot | Indexing for ChatGPT search | ChatGPT search cannot cite your pages |
| ChatGPT-User | Fetching a page a user asked about | Users who paste your URL get an error |
The robots.txt on this site allows all three, along with ClaudeBot, Claude-SearchBot, PerplexityBot, Google-Extended and the others. A company that sells to buyers who use AI has no reason to refuse the crawlers those buyers' tools depend on. The AI visibility audit checks robots.txt and server responses for every major crawler as its first step.
Yes, by more than most vendors admit. For Google AI Overviews, Ahrefs found 76 percent of cited pages rank in the top ten; ChatGPT leans far more on directories and roundups. One company we measured in August 2026 had 44 ChatGPT mentions and 2 AI Overview appearances in the same week. A company can rank well in Google and be invisible in ChatGPT because it exists nowhere else on the web, or be named constantly by ChatGPT with a weak website because it is in every directory. The two are measured separately for that reason.
The practical consequence for a B2B company is that "SEO" and "GEO" are not two services to buy from two vendors. They are one body of work with a layer on top, and the layer is mostly entity consistency and third-party presence. The GEO vs SEO page sets out the overlap by engine. The AI Overviews page covers Google's side. The AI Mode page covers the engine that sits between them.
ChatGPT names the company it has met most consistently across the web, not the one with the best website.
The ChatGPT-specific ruleFigures described as measured are from Search Standing's own analyses using Semrush, Google Search Console and direct queries to each AI engine, on the dates stated. Published ranges move quarter to quarter; the patterns are what we rely on.
Send your company and your market. We will run the analysis and send back a summary naming your competitors, the terms they are winning, and where you sit against them.
No call required and no obligation. Read it, check it yourself, and decide. If the findings are worth discussing after that, we can talk then.
Prepared for your company against the competitors you name. Written by the person who would run the paid work.
Or email contact@searchstanding.com directly.