You can't rank in ChatGPT the way you rank in Google. The answer changes with the wording, the run, the location and the person asking. What you can do is raise the odds of being named: get onto the pages ChatGPT reads when it searches (best-of lists, review sites, directories), keep your facts the same everywhere, let OpenAI's search crawler into your site, earn reviews, and publish data worth quoting. Then test it honestly, logged out, across ChatGPT, Gemini and Perplexity.
On Monday the CMO forwards a screenshot with three exclamation marks: "We're number one in ChatGPT!" On Wednesday a sales rep says a prospect asked ChatGPT for options in your category and you weren't on the list. Both screenshots are real. Neither one tells you much.
If you work in-house or at an agency, you are getting some version of "how do we rank in ChatGPT?" every week now. People also call it ChatGPT SEO, LLM SEO, or getting recommended by ChatGPT. The honest answer starts with a correction: there is no rank in the classic sense. This guide covers how ChatGPT actually builds an answer, the things that move it, how to test without fooling yourself, and what OpenAI's brand-new Business Hub does and doesn't show you.
There is no "rank" in ChatGPT, only odds
A Google result page for a query is roughly the same for everyone in the same place on the same day. A ChatGPT answer is generated fresh each time. Four things shift it:
- The prompt. "Best accounting software for a small firm" and "cheapest accounting software for accountants" are different questions with different answers, even though a human would see them as close.
- The run. Ask the same thing twice and the list can change in membership, order and length.
- The location. OpenAI says ChatGPT estimates a general location from your IP address and can pass it to search providers, so "near me" and city-specific questions get local results.
- The person. If memory is on, ChatGPT may use saved memories when it rewrites a search query. Your account knows things about you that a buyer's account doesn't.
How much does the run alone matter? In a study published in January 2026, SparkToro and Gumshoe had volunteers run the same brand-recommendation prompts many times on ChatGPT, Claude and Google's AI. ChatGPT and Google's AI returned the same list of brands less than 1% of the time. The more stable signal, they found, was how often a brand showed up across many runs, not where it sat in any single answer.
So stop thinking about position one. Think about odds: across the questions your buyers really ask, how often are you named, how high, and on which engines? That is what "ranking in ChatGPT" means in practice, and it is what you can actually move.
How ChatGPT builds an answer when it searches the web
ChatGPT answers in one of two ways. Without search, it answers from what the model learned in training, and you can't change that quickly. We explain that side in how AI engines choose which brands to recommend. With search, which is where buying questions increasingly land, the process is documented, and it is where your work pays off fastest.
Here is what OpenAI's own help page on ChatGPT search says, as checked on 13 November 2026:
- It decides to search. ChatGPT "may search the web automatically when your question would benefit from current information". Users can also turn search on by hand.
- It rewrites the question. ChatGPT "typically rewrites your query into one or more targeted queries" and sends them to search partners. After looking at the first results it may send more specific follow-up queries. Location and, if memory is on, saved memories can shape those queries.
- It reads what comes back. The answer is written from the pages those searches return, and "responses that use web search may include citations". OpenAI adds that it "ranks search results using multiple factors intended to help users find relevant, reliable information. Placement is not guaranteed."
- It needs to be allowed in. To be eligible, a site has to let OAI-SearchBot crawl it and let traffic from OpenAI's published IP ranges through.

The practical reading: a buying question turns into a handful of ordinary web searches, and the pages that win those searches become the raw material for the answer. If a best-of list, a review platform and a directory all name the same three tools, those three tools are very likely to be in the answer. If you are on none of those pages, a beautiful homepage won't save you.
What actually moves ChatGPT recommendations
None of this is a trick. It is the slow, ordinary work of being visible on the web in the places that answer buying questions.
1. Be on the pages the answers cite
Start with the questions buyers ask, run them, and write down every page ChatGPT cites. You will see the same kinds of source over and over: "best X for Y" roundups, comparison posts, review platforms, industry directories, sometimes a forum thread. That list is your target list. Pitch the authors of the roundups with something genuinely useful, claim and complete your review-platform profiles, fix or add your directory entries. If a competitor keeps getting named, check which of those pages it is on and you are not. We walk through that gap analysis in why ChatGPT recommends your competitor, and through the kinds of sources AI leans on in what AI cites about your brand.
2. Keep your facts the same everywhere
When the pages ChatGPT reads disagree about you, the safe move for the model is to say less, or nothing. Use one brand name, one category description, one clear line on who you serve and where. Make your website, LinkedIn page, review profiles, directory listings and Google Business Profile all say it. If you rebranded two years ago and half the web still uses the old name, fix that before you write another blog post.
3. Make your site easy for OpenAI to crawl
OpenAI runs separate crawlers for separate jobs, and its crawler documentation spells out the difference. OAI-SearchBot is "used to surface websites in search results in ChatGPT's search features". GPTBot collects content that may be used to train models. They are independent: you can allow OAI-SearchBot and block GPTBot. The line that matters most: "Sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though can still appear as navigational links."
So check three things. Your robots.txt doesn't disallow OAI-SearchBot. Your CDN or firewall isn't blocking OpenAI's published IP ranges as "bot traffic", which happens more often than people think. And the facts you care about (what you do, who for, pricing if you publish it) are in plain HTML text, not buried in images or behind a login. OpenAI says robots.txt changes take about 24 hours to be picked up.
4. Earn reviews on the platforms that show up
Review platforms do double duty. They are cited directly, and their ratings feed the roundups that are cited. Ask real customers, steadily, on the two or three platforms that appear in your answers, and answer the reviews you get, including the bad ones. Don't buy them or fake them; it is against platform rules and, as we cover in our post on how engines choose brands, it is a brand risk you don't need.
5. Publish something worth quoting
Original data is the closest thing to a shortcut. A survey of your customers, a benchmark from your own product data, a pricing study in your niche: other writers cite it, those articles rank, and the numbers get pulled into answers with your name attached. Opinion pieces rarely get this treatment. Numbers with a clear source and date do.
How to test whether ChatGPT recommends you, honestly
A confession first. On 12 November 2026 I asked ChatGPT, from my own logged-in account, which platform to use for managing many business locations. It put BizLoc8 first. Then it added, helpfully, "Since you're building Bizloc8.ai". That wasn't a ranking. That was memory. ChatGPT knew who was asking, and no buyer would have seen that answer.
Most "we're number one in ChatGPT" screenshots are made this way. Here is the test I'd trust instead.

- Log out, or use a clean session. OpenAI says people who aren't signed in can also use web search, so a logged-out or incognito window is the closest you'll get to a stranger's view. If you must stay logged in, turn memory off.
- Use buyer phrasing, without your brand. "Best analytics tool for a 20-store retailer", not "Is Rosewood Analytics good?". Put your brand in the question and of course it gets mentioned.
- Cover the intents. Ten to twenty prompts: "best", "cheapest", "for [industry]", "alternatives to [competitor]", "[category] in [city]".
- Repeat each prompt. Three to five runs each, because one run is a coin flip. Count how often you appear.
- Check Gemini and Perplexity too. They often name different brands for the same question. A client who only checks ChatGPT will be surprised by a prospect who uses Perplexity.
- Hold location steady. Run city-specific prompts from, or about, each city that matters.
- Write it down. For each answer: mentioned or not, position, competitors named, sources cited. Re-run the same sheet in a month.
For the full manual method and how to turn it into ongoing tracking, see how to track your brand in ChatGPT, Gemini and Perplexity. Fair warning: 20 prompts, five runs, three engines and three cities is 900 answers. That is where people start looking for tooling.
What about ChatGPT Business Hub?
On 10 November 2026 OpenAI introduced Businesses in ChatGPT, in beta, pitched at the 1.2 billion people OpenAI says use ChatGPT each week. The part marketers care about is Business Hub. Here is what OpenAI's announcement and its setup help page confirm, as checked on 13 November 2026:
- Claim your business. Verify by business phone, business email, a DNS record on your domain, or Shopify, then complete identity verification with Persona before you can publish changes or see insights.
- Update your details. Website, description, category, logo, delivery and return policies, and product feeds with pricing and availability. OpenAI reviews changes before they go live.
- See insights for the past 28 days. Mentions, website clicks and "discovery topics", which are common themes in answers that mentioned or cited you.
- Give an agency access. The admin who claimed the business can add teammates and grant an agency access.
And what it doesn't do. Insights are aggregated and only appear above a minimum activity threshold; a blank number can mean "not enough data", not zero. You don't see prompts or conversations. For an online business, a mention counts when ChatGPT references your website; OpenAI says "the brand name alone doesn't count". We could not find any competitor view in OpenAI's documentation, and both Digiday and Trending Topics report there is no comparison against competitors. It covers ChatGPT only. Local stores and restaurants can, for now, only claim an existing US listing. And OpenAI is direct that claiming "doesn't guarantee increased visibility or give you direct control over ChatGPT's independent recommendations".
My take: claim it. It is free, it is first-party data straight from OpenAI, and keeping your official details correct in ChatGPT is exactly the "same facts everywhere" work from earlier. Mention counts and discovery topics will tell you things nobody else can. Then fill the gaps it leaves on purpose: which buyer questions you win and lose, who is named instead of you, which pages the answers lean on, and what Gemini and Perplexity say. That is the job of a tracking tool.
Running the whole test in BizLoc8
Everything above can be done in a spreadsheet. This is how the same method runs in BizLoc8 AI Visibility when you have more brands, prompts or cities than you have patience.
- Set up the brand. The setup wizard asks for the website, brand, country, languages and the cities you care about. It can generate an optional buyer persona from the website.
- Choose the prompts. Prompts are the questions buyers ask AI. BizLoc8 suggests them, or you write your own, and each is tagged with its intent and search volume. This is step 2 and 3 of the honest test, done up front.
- Add competitors. They are discovered for you, or you add the names your sales team keeps hearing.
- Run the engines. Prompts run on ChatGPT, Gemini and Perplexity, with one answer per engine, per city, per prompt. The first run happens automatically after purchase. After that you run on demand: everything, one engine, or selected prompts.
- Read where you stand. The dashboard shows whether you are mentioned and at what position in each answer, engine-by-engine visibility, average rank by intent, a prompt-by-prompt heatmap with an answer summary, sentiment for you and each competitor, and a headline verdict on why you are losing.
- Find the pages behind the answers. Cited sources are broken down by engine and by prompt, with citation trends, plus the search fan-out queries Gemini used to ground its answer.
- Work the Action Items. Each cited domain gets an entry: its role, its authority, whether you are on it, which competitors are, which engines cite it, and a citation strategy for getting on it. This is section one of "what moves it", as a to-do list.
- Fix your own site. Generate page copy, schema JSON-LD and meta tags, and run the Website Audit for technical, schema, Core Web Vitals and content gaps.
- Re-run and compare. Run the prompts again when the work has had time to land, and watch rank over time. Export the AI Visibility PDF or the AI Strategy PDF for the client or the boss.
Agencies can keep each brand as a separate project and give teammates access. If you want to see the problem before committing, the free AI Visibility audit runs on ChatGPT with no signup to start and shows your mention rate, the competitors ahead of you and the domains the answers cite.
See how often ChatGPT names you
Run your buyers' real questions on ChatGPT, Gemini and Perplexity and see who gets named, and which pages put them there.
Mistakes that waste months
- Promising "number one in ChatGPT". Nobody controls the answer. You control how often the brand appears on the pages the answer is built from. Sell that, and report frequency, not a single position.
- Reporting from a logged-in account. My own BizLoc8 result above is the cautionary tale. Memory and chat history bend the answer toward you.
- Blocking OpenAI by accident. A blanket "block AI bots" setting in a CDN or a copied robots.txt can shut out OAI-SearchBot. Blocking GPTBot for training is a fair choice; blocking search is usually not what you meant.
- Only working on your own site. Schema, FAQs and clean copy help, but most of what ChatGPT reads about you sits on other people's pages.
- Trying to game it. Prompt-spamming services, fake reviews and astroturfed threads don't work and can backfire.
- Ignoring Business Hub, or treating it as the whole picture. Claim it for the free first-party data, then measure the prompts, competitors and engines it doesn't cover.
- Testing once and moving on. Answers change. A monthly re-run of the same prompts is the only way to know whether the work did anything.
Frequently asked questions
Can you rank number one in ChatGPT?
Not in a fixed way. ChatGPT writes a new answer each time, and the brands named change with the wording, the run, the location and the user's memory. A January 2026 SparkToro and Gumshoe study found ChatGPT returned the same brand list less than 1% of the time across repeated runs. You can raise how often you are named, and how high, but nobody can promise a permanent first place.
What is ChatGPT SEO, or LLM SEO?
It is the work of getting a brand named in answers from ChatGPT and other large language model tools. It overlaps with classic SEO, because ChatGPT search reads web pages that rank for the queries it sends, but most of the effort goes into being listed on third-party pages such as best-of lists, review sites and directories, keeping your facts consistent, and letting OpenAI's search crawler reach your site.
How do I get my website to show up in ChatGPT search?
Allow OAI-SearchBot in your robots.txt and make sure your host or CDN lets through traffic from OpenAI's published IP ranges. OpenAI says sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers, and that robots.txt changes take about 24 hours to take effect. Being eligible does not guarantee placement; your pages still have to be relevant to the searches ChatGPT runs.
If I block GPTBot, will ChatGPT stop recommending me?
Not from search. OpenAI documents GPTBot as the crawler for model training and OAI-SearchBot as the crawler for ChatGPT search, and says each setting is independent. You can block GPTBot and still allow OAI-SearchBot so your site can appear in ChatGPT search answers.
Does claiming my business in ChatGPT Business Hub make ChatGPT recommend it more?
No, not directly. OpenAI says claiming lets you manage the information you provide but does not guarantee increased visibility or give you control over ChatGPT's independent recommendations. It is still worth claiming for accurate details and its free insights on mentions, website clicks and discovery topics.
Why does ChatGPT recommend my brand to me but not to my customers?
Usually memory and context. If memory is on, ChatGPT may use what it knows about you when it rewrites a search, and your past chats tell it which brand you care about. Test from a logged-out or clean session, with buyer phrasing that does not include your brand name, and repeat each question several times.
Sources
All checked 13 November 2026.
- OpenAI Help Center: Searching the web with ChatGPT (query rewriting, location, memory, citations, OAI-SearchBot eligibility).
- OpenAI: Overview of OpenAI crawlers (OAI-SearchBot, GPTBot, about 24 hours for robots.txt changes).
- OpenAI: Introducing Businesses in ChatGPT, 10 November 2026, and Help Center: Setting up your business on ChatGPT.
- Digiday: ChatGPT is about to show brands their mention count, 9 November 2026. Trending Topics: ChatGPT now tells companies how often they're mentioned, 10 November 2026.
- SparkToro: AIs are highly inconsistent when recommending brands or products, as reported by Search Engine Journal on 30 January 2026.
- BizLoc8 AI Visibility features as of 13 November 2026.