How to Get ChatGPT to Recommend You or Your Brand

You open ChatGPT and type the exact question a client would ask: "best freelance video editors for YouTubers" or "who should I follow to learn guitar online". The answer comes back confident and friendly. It names four people. One of them started the same month you did. You are nowhere, and the worst part is that the AI sounds so sure, like the list it gave is simply the truth.

That sting is the whole point of this guide. More people now ask an assistant before they ask a friend, and the assistant can only recommend people it can find and understand. Here is LitFame's honest thesis: if you don't get famous, how can AI discover you? Famous in a small, specific sense: written about, reviewed, linked and described the same way in enough places that a machine reading the web can say your name with confidence.

This guide shows how to get recommended by ChatGPT using what OpenAI actually documents: how it answers with and without web search, the crawler setting that decides whether you're even eligible, where its sources come from, and the steps that put you into them. Then Perplexity and Gemini, a repeatable test, and the myths to ignore.

How ChatGPT decides what to say about you

ChatGPT has two ways of answering a question about a person, brand or category, and the work you do to show up is different for each.

Answering from what the model already learned

OpenAI's help article "How ChatGPT and our foundation models are developed" says its models are built from three sources: information publicly available on the internet, information accessed through partnerships with third parties, and information that users, human trainers and researchers provide or generate. Public pages, forums, blogs and posts can all end up in training data, filtered to remove things like spam and sites that aggregate personal information.

Without search, ChatGPT works from patterns learned in training, and the same article notes an element of randomness, so one question can produce different answers. Two consequences matter. The model learned from a snapshot of the web, so anything after it is invisible to this mode. And it is most confident about names that appeared often and consistently in what it read: a creator mentioned in dozens of articles, threads and reviews has a far stronger footprint than one with a single bio page.

Answering with web search

OpenAI's "Searching the web with ChatGPT" help page says ChatGPT may search the web automatically when your question would benefit from current information, and users can also start a search manually from the tools menu. When it searches, ChatGPT typically rewrites the question into one or more targeted queries and sends them to third-party search providers, then may send further, more specific queries after reviewing the first results. If memory is on, saved memories can shape those rewritten queries. Responses that use search may include citations and a Sources panel.

This mode moves faster. A crawlable page about you that answers the question well can be read and cited this week.

You don't choose which mode a stranger triggers, so aim to be strong in both.

OAI-SearchBot: the setting that makes you eligible

OpenAI publishes a page called "Overview of OpenAI crawlers" that lists its user agents. Three matter here, and people mix them up constantly.

The help page on web search puts it plainly: to make a website eligible for inclusion, allow OAI-SearchBot to crawl the site and confirm your host or content delivery network allows traffic from OpenAI's published searchbot IP addresses. The crawler page adds that it can take about 24 hours after a robots.txt change for search results to adjust.

The mistakes that quietly lock you out

• A blanket "block AI bots" rule. Some site builders and CDNs offer a one-click toggle that blocks AI crawlers. If it blocks OAI-SearchBot, you've opted out of ChatGPT search without meaning to. Check what your toggle actually covers.
• A firewall that challenges unknown bots. Your robots.txt can say "allow" while your CDN serves a challenge page. OpenAI publishes its IP ranges at openai.com/searchbot.json so you can allowlist them.
• Nothing public to crawl. If everything about you sits behind a login, crawlers have nothing to read. A simple personal site fixes that.

Should you also allow GPTBot?

That's your call. Blocking GPTBot keeps your content out of future training, which some creators prefer for their writing or art. Allowing it lets your public pages become part of what future models learn, which helps the no-search mode recognise you. Either way, it doesn't affect ChatGPT search eligibility.

One more line from OpenAI is worth keeping on a sticky note: "Placement is not guaranteed." The help page says ChatGPT ranks search results using multiple factors intended to help users find relevant, reliable information. Eligibility gets you into the pool. Everything after that depends on how useful and well-supported the pages about you are.

Where ChatGPT gets the sources it cites

OpenAI doesn't publish a list of preferred sites or a ranking formula for ChatGPT search. What it does document gives you enough to work with.

• Search providers. ChatGPT sends rewritten queries to third-party search providers, so ordinary search visibility still matters. Being indexed by the major search engines, including Bing, is cheap insurance.
• Pages it reads to answer. Once results come back, ChatGPT reads pages and cites the ones it uses. Pages that clearly state facts, names and context are easier to use.
• Structured product data. For shopping questions, OpenAI's "Shopping with ChatGPT Search" page says it considers structured metadata from first-party and third-party providers, such as price and description, plus other third-party content. Product review summaries are generated from reviews on public websites. Shopify merchants are integrated through Shopify Catalog, and other merchants can apply to provide a direct product feed.

Notice the pattern. For "best X" questions, much of the useful material is written about you by someone else: roundups, reviews, forum threads, interviews. Your site explains who you are, third parties supply the proof, and most people only build the first.

For a hint at the open web's heavyweights, Pew Research Center's 2025 analysis found Wikipedia, YouTube and Reddit were the most cited sources in Google's AI summaries. That study covered Google, so it describes ChatGPT only loosely, but community and video platforms are clearly on AI's reading list, and you can show up on them.

Step 1: Get mentioned where recommendation lists get built

This is the part that feels most like "getting famous", and it's the part that moves the needle for category questions. Your goal is to be named, accurately, on pages that people (and AI) consult when they want options.

Lists and roundups

Search the questions your future clients ask, like "best wedding photographers in Kuala Lumpur" or "top Notion template creators", and open the articles that rank. Many of those authors update their lists and accept suggestions. Send a short, specific pitch: who you are in one line, one piece of proof (a result, a number, a notable client or project), and a link. Don't pay for placement in lists that hide the payment. Undisclosed paid placements can break advertising disclosure rules where you live and they wreck trust when spotted.

Reviews and directories

For local businesses, a Google Business Profile with real customer reviews is basic infrastructure. Freelancers can use whichever review platforms clients in their field check. Ask happy clients for honest reviews right after delivery. Never buy fake reviews: Google, Trustpilot and most review platforms prohibit them, and Google's own guide to AI features says seeking inauthentic mentions isn't as helpful as it might seem because its systems block spam.

Press, podcasts and collaborations

A local news feature, a guest spot on a niche podcast with a written show-notes page, a collab video with a creator in your space, an interview on an industry blog. Each creates a page that describes you in someone else's words, usually with a link. Small outlets count.

Reddit and forums, done properly

People ask Reddit for recommendations all day, and those threads are public. The way in is to be useful: answer questions in your field in detail, share what you know, and let your profile explain who you are. Read each subreddit's rules on self-promotion and disclose when you're recommending your own thing. Fake accounts praising yourself get banned and screenshotted. Our Reddit growth guide covers how to build karma and reputation the slow, legit way.

Step 2: Make your identity consistent everywhere

AI systems have to work out that "Jess Tan, video editor" on Instagram, "jesstanedits" on YouTube and "Jessica Tan" in a podcast transcript are the same person. When the signals line up, all your proof attaches to one identity.

• Pick one public name and stick to it. Use the same spelling across platforms, bylines and invoices.
• Write one bio line and reuse it. Something like "Jess Tan is a Singapore-based video editor for education YouTubers." Put the same core sentence on every profile, adjusting length only.
• Link everything to one home base. A personal site or link page that lists all your official profiles. Then link back to it from each profile.
• Clean up old identities. Update or close abandoned accounts with an old name or niche.

If you have a personal site, schema.org's Person or Organization type includes a sameAs property for listing your official profile URLs. It's optional and no engine has promised it will boost you, but it states your identity in a machine-readable way at almost no cost. For the full playbook on owning the search results for your own name, read personal SEO: own your name.

Step 3: Build a clear "who I am" page

When ChatGPT searches for you by name, you want at least one result that answers "who is this?" in plain text, written by you. Most creators have a link-in-bio page with six buttons and no sentences. That gives a reader, human or machine, almost nothing to quote.

A strong about page includes:

1. A one-sentence definition at the top. Name, what you do, who you do it for, and where if location matters. This is the line you'd love an AI to repeat.
2. Specifics. Services, topics, formats, platforms, and the kind of clients or audience you serve.
3. Proof. Notable projects, collaborations, press mentions with links, results you can back up.
4. Your official profiles. Linked, so the page connects your identity together.
5. How to contact or hire you. A clear next step for a person who arrives from an AI answer.
6. A visible "last updated" note. Stale facts are a common source of wrong AI answers. Keep prices, locations and offerings current.

Write it as real text on the page, since crawlers can't reliably read words baked into an image. If you sell products, give each its own page with a clear name, price, description and honest reviews, the fields OpenAI says its shopping results draw on.

Structured data and the proposed llms.txt file get a lot of hype. Our guide to generative engine optimization for creators explains where they fit. Short version: a clear page with good text matters far more than any file.

Step 4: Publish citable facts and stats you own

An AI answer that names you usually needs a reason: a fact, a number, a distinctive claim. If you're the original source of something useful, every article that cites it creates another mention pointing back to you.

The research backs this up in a limited way. The 2023 paper "GEO: Generative Engine Optimization" by Aggarwal and colleagues, a team including researchers from Princeton, Georgia Tech and the Allen Institute for AI, built a benchmark of queries and tested how rewriting source pages changed their visibility in generative engine answers. They reported that their methods could boost visibility by up to 40 percent, and that adding statistics, quotations and citations to credible sources were among the strongest methods, while keyword stuffing did little. Caveats: it was a research setup, mostly their own test engine plus a check on Perplexity, and results varied by domain. It doesn't reveal how ChatGPT ranks today.

Ways to create facts you own:

• Your own results. "Edited 140 videos for 12 channels in 2025" or "average client turnaround 48 hours". Only publish what's true and you can back up.
• Small surveys. Poll your audience or customers and publish the results with the sample size and date. Fifty honest responses in a niche can be more citable than a vague claim.
• Case studies with numbers. Before and after, with what you did and how long it took.
• Clear pricing and specs. Prices, packages and turnaround times on a public page. People ask AI "how much does X charge" all the time.

Put each fact on a stable, dated URL and update it in place. Never invent numbers: a fake stat repeated by an AI and then debunked is the kind of fame nobody wants.

Perplexity and Gemini work differently

ChatGPT isn't the only assistant your audience uses. The basics carry over, but the plumbing differs.

Perplexity

Perplexity's help article "How does Perplexity work?" says it searches the internet in real time for each question and includes numbered citations linking to the original sources. That makes it closer to a search engine with a summary on top, so fresh, crawlable pages matter even more. Its crawler documentation lists PerplexityBot, used to surface and link websites in search results and not used to crawl content for AI foundation models, and Perplexity-User, which fetches pages for user requests and generally ignores robots.txt because a user asked. To appear, allow PerplexityBot and its published IP ranges. Perplexity also marks some cited domains with Government, Academic or Trusted labels, and its help center says partnerships and payments do not affect those labels.

Gemini

Gemini lives inside Google's ecosystem. Google's crawler documentation says the Google-Extended token lets publishers manage whether content Google crawls may be used to train future Gemini models and for grounding in Gemini Apps, meaning content from the Google Search index provided to the model at prompt time. Google-Extended does not affect inclusion or ranking in Google Search. Gemini Apps help says some responses include a Sources button. Being well indexed and well described in Google Search is the foundation here too. Google's AI Overviews and AI Mode are a separate surface with their own rules, covered in how to get cited in Google AI Overviews and AI Mode.

How to test whether ChatGPT recommends you

One check on your own account tells you little: memory shapes answers and one run is a coin flip. Use a repeatable protocol.

1. Check access first. Open yoursite.com/robots.txt and confirm OAI-SearchBot and PerplexityBot aren't disallowed. Ask your host whether bot protection challenges them.
2. Use a clean session. Test logged out, or use Temporary Chat, so memory doesn't nudge the results toward you.
3. Write three kinds of prompts. See the table below. Use the words your audience would actually type.
4. Run each prompt several times, with and without Search. Note whether you appear, how you're described, and which sources are cited.
5. Save everything. Date, prompt, full answer, cited URLs. A simple spreadsheet works.
6. Check your analytics. Links from ChatGPT carry utm_source=chatgpt.com, so you can filter for that in your analytics tool to see real visits.
7. Repeat monthly. AI answers shift. Look for trends across many runs.

The cited sources are the gold. When a competitor appears and you don't, open the pages cited for them: those are the roundups, reviews and threads you should be on. A painful screenshot becomes a to-do list.

If ChatGPT gets a fact about you wrong, fix the source first: your about page, your profiles, and any third-party page carrying the error.

Myths to ignore

"You can pay ChatGPT to recommend you"

You can buy ads, and that's a separate thing. OpenAI began testing ads in ChatGPT in the United States on February 9, 2026, for Free and Go users. Its "Ads in ChatGPT" help page says ads do not influence ChatGPT's answers, run on separate systems from the chat model, appear below responses, and are clearly labeled. Advertisers have no ability to shape, rank or alter responses. The shopping help page says product results are selected independently and are not ads. So an organic mention can't be bought from OpenAI, and anyone selling "guaranteed ChatGPT placement" is selling something OpenAI says doesn't exist.

"An agency can guarantee you'll be recommended"

OpenAI's own help page says placement is not guaranteed. A consultant can improve your pages and coverage, and nobody outside OpenAI controls what the model says.

"Follower count is what AI looks at"

No AI company documents follower counts as a ranking or recommendation signal. What gets you into the sources these systems read is being mentioned, linked, reviewed and consistently described across the web, plus useful pages of your own. A big audience helps indirectly because it tends to produce more of those mentions, and that's the real reason growing your audience matters. Our piece on why you need to be famous makes the full case.

"You need an llms.txt file to be found"

llms.txt is a proposal from llmstxt.org. Google's optimization guide says Google Search doesn't use it, and OpenAI's crawler documentation doesn't mention it. It's harmless to add, and it won't replace any of the steps above.

Frequently Asked Questions

How do I get ChatGPT to recommend my business?

Make sure OAI-SearchBot can crawl your site, publish a clear page explaining what you do and for whom, and earn mentions on the pages ChatGPT reads for recommendation questions: roundups, review platforms, press, podcasts and forum threads. Keep your name and description consistent everywhere. OpenAI says placement is not guaranteed, so treat this as improving your odds over time.

Does ChatGPT use Google or Bing to search?

OpenAI's help page says ChatGPT search sometimes partners with third-party search providers, sending them rewritten versions of your question, and it doesn't name a single provider on that page. Being well indexed in the major search engines is a sensible baseline because those results can feed what ChatGPT reads.

How long does it take for ChatGPT to start mentioning me?

For search-based answers, OpenAI says robots.txt changes take about 24 hours to be reflected, and new pages can be found as soon as they're discoverable through search. Appearing in answers that don't use search depends on future model training, which OpenAI doesn't schedule publicly. Most people should expect months of steady work before category recommendations change.

Can I pay to be featured in ChatGPT answers?

No. OpenAI sells ads that appear below some responses for Free and Go users, labeled as ads. Its help center says ads do not influence ChatGPT's answers and advertisers cannot shape, rank or alter responses. Product results in shopping answers are also described as selected independently and not ads.

Does blocking AI crawlers stop ChatGPT from recommending me?

Blocking OAI-SearchBot removes your site from ChatGPT search answers, apart from possible navigational links. Blocking GPTBot only signals that your content shouldn't be used for training. ChatGPT can still mention you based on other sites that talk about you, which is another reason third-party mentions matter.

Do more followers make ChatGPT more likely to recommend me?

Follower count is not a documented signal for ChatGPT, Perplexity or Gemini. What helps is being mentioned, reviewed, linked and consistently described across the public web. A larger audience often leads to more of those mentions, so growth helps indirectly.

Why does ChatGPT give different answers when I ask about myself?

OpenAI explains that its models include an element of randomness, so the same question can produce different answers. Memory, whether search was used, and how the question is phrased also change results. Test in a clean session, run each prompt several times, and track trends rather than single answers.

How is Perplexity different from ChatGPT for getting recommended?

Perplexity searches the web in real time and attaches numbered citations to its answers, so fresh, crawlable pages matter a lot. Allow PerplexityBot, which its docs say is used for search results and not for training foundation models. The same off-site work, such as reviews, lists and press, helps on both.