ChatGPT recommends brands it has clear, consistent evidence for, drawn from what its models learned in training and, when it searches the web, from the pages it retrieves at that moment. To get your brand recommended by ChatGPT, make your site crawlable by OpenAI’s search crawler, state plainly what you do and who you are best for, and earn accurate mentions on the third-party sites that answers in your category rely on. Because answers vary from one conversation to the next, judge progress across many prompts and repeated runs, not a single screenshot.
This guide explains the two ways ChatGPT forms an answer, what OpenAI has said publicly, what nobody outside OpenAI knows, and the steps that make your brand easier to recommend.
Key takeaways
- Two sources of knowledge: ChatGPT answers from what its model learned in training and, when it searches, from live web pages. You can influence both, on very different timescales.
- OAI-SearchBot is the access key: OpenAI says sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers. GPTBot, the training crawler, is a separate decision.
- Clarity wins recommendations: brands that say exactly what they do, for whom and at what price give the model something specific to repeat.
- Third-party evidence matters: reviews, roundups, communities and publications shape how ChatGPT describes a category and who belongs in it.
- Measure patterns, not screenshots: answers change with wording, context, memory and location, so track a fixed set of prompts over time.
How ChatGPT decides what to say about brands
When someone asks ChatGPT for a recommendation, such as “what is the best accounting software for a freelance designer”, the answer can come from two places.
From what the model learned
Language models are trained on large amounts of text, including public web pages, up to a cutoff date. If your brand was widely and consistently described before that cutoff, the model may already associate it with your category and use cases. If it was barely mentioned, or described inconsistently, the model has little to go on. This knowledge only changes when new models are trained and released, so it moves slowly.
From a live search
For many questions, especially ones about current products, prices or comparisons, ChatGPT can search the web and base its answer on the pages it retrieves. It decides on its own when to search, and users can also ask it to. Answers built on search usually show their sources as links. This layer moves much faster: a page that is crawlable and relevant can influence answers soon after it is discovered.
In practice, many recommendation answers blend the two. Background knowledge shapes which brands the model considers, and search results add current details and citations. That is why the steps below cover both your own site and your wider reputation. For the bigger picture, see how AI search engines choose sources.
What OpenAI has said publicly
- Separate crawlers: OpenAI documents GPTBot for collecting training data, OAI-SearchBot for surfacing websites in ChatGPT search, and ChatGPT-User for certain actions a user starts. Each can be controlled separately in robots.txt, and OpenAI publishes IP ranges for them.
- Search inclusion: OpenAI says sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers, though they can still appear as plain navigational links.
- Training is independent: blocking GPTBot tells OpenAI not to use your content for training foundation models. It is a separate choice from appearing in search answers.
- Search providers: OpenAI has said ChatGPT search draws on third-party search providers as well as its own technology and content from partners. Microsoft Bing is widely reported to be one of those providers, but OpenAI does not publish the full mix.
- Shopping: when OpenAI introduced shopping results, it said product results are chosen independently and are not ads, and that they draw on structured product information from third parties.
What is not known
OpenAI has not published how ChatGPT chooses which brands to name, in what order, or how it weighs one source against another. Nobody outside OpenAI can tell you the formula, and the behavior changes as models and search systems are updated.
Answers also vary for reasons that have nothing to do with your brand. The same question can produce different recommendations depending on its wording, earlier messages in the conversation, the user’s location, memory and custom instructions, and the model in use. That is why claims like “rank number one in ChatGPT” do not hold up. There is no single list to rank in, only a likelihood of being included across many conversations.
Step 1: Decide which recommendations you want to earn
“Get recommended by ChatGPT” is too vague to act on. Write down the specific situations where you want to be named. Good prompts combine a category, an audience and a need, for example “best project management tool for a ten-person agency”, “CRM that integrates with Shopify” or “alternatives to a named competitor for small teams”.
Aim for 25 to 50 prompts that reflect real buying decisions, and keep your brand name out of most of them, since branded prompts mostly test whether ChatGPT knows you exist. The free AI prompt ideas generator drafts a list you can trim. Ask each prompt in ChatGPT, note who is recommended and which sources are cited, and save the results as your baseline.
Tip: Be honest about fit. If ChatGPT consistently recommends you for small teams but not for enterprises, that may reflect your real positioning. Win the prompts where you are genuinely a strong answer before chasing the rest.
Step 2: Open the door to OpenAI’s search crawler
Check your live robots.txt for rules that affect OAI-SearchBot and ChatGPT-User, and make sure your CDN, firewall or security plugin is not blocking them. If you want to be recommended with a link to your site, OAI-SearchBot needs access to your public pages.
Decide on GPTBot separately. Allowing it lets your public content be considered for future training, which can help future models know your brand; blocking it does not affect search inclusion. Many brands that want visibility allow both, but it is a business decision, and our guide to which AI crawlers to allow walks through the trade-offs. The free AI robots.txt generator writes the rules for you.
Because ChatGPT search draws on third-party search providers, it is also sensible to make sure you are indexed in Bing as well as Google. Verify your site in Bing Webmaster Tools, submit your sitemap, and use IndexNow if your platform supports it.
Step 3: State what you are, plainly and consistently
A model can only repeat what it can find. Many brand sites describe themselves in slogans that make sense to insiders and say little to a machine. Fix that on the pages that matter most.
- One-sentence description: say what you are, who it is for and the main outcome, in plain words, on your home and about pages. Use the same description everywhere.
- Best-for statements: say which kinds of customer you serve best, and which you do not. “Built for agencies of 5 to 50 people” is far easier to repeat than “for teams of every size”.
- Pricing: publish prices or price ranges where you can. Recommendation questions often include a budget, and a missing price leaves the model to guess or skip you.
- Comparisons: publish fair comparison and alternatives pages for the competitors buyers weigh you against, including where the other option is the better fit.
- Facts in one place: put integrations, limits, regions, languages, security and support details on clear, crawlable pages, not only in PDFs or sales decks.
Write these pages answer-first, with descriptive headings and specific claims. The free answer readiness checker shows whether a page states its answer early and clearly enough to quote.
Step 4: Earn mentions where recommendations come from
Recommendations are rarely built from your website alone. Look at the sources ChatGPT cites for your prompts and the brands it names. You will typically find review platforms, “best tools” roundups, comparison articles, community discussions, documentation and industry publications. These are the places that shape which brands belong in the answer.
- List the recurring sources. From your baseline, note the domains that appear for several of your prompts.
- Fix your presence on them. Claim and complete your profiles on relevant review and directory sites, and make sure categories, pricing and descriptions match your own site.
- Ask happy customers for honest reviews. A steady flow of genuine, recent reviews helps people as well as machines. Never buy or fake reviews.
- Pitch the roundups. Contact the authors of articles that compare tools in your category, offer accurate information and a trial, and accept that the verdict is theirs.
- Contribute to communities. Answer questions where your audience asks them, under your real identity, and mention your product only where it genuinely fits.
- Publish citable assets. Original data, benchmarks and practical guides give writers a reason to reference you.
To see how your link profile compares with the brands ChatGPT keeps recommending, run your domain and theirs through the free domain authority checker. It is a proxy for authority, not a measure of AI visibility, but a large gap is worth knowing about.
Step 5: Check how ChatGPT describes you, and correct the source
Being mentioned is only half the result. Read how ChatGPT describes your brand: your category, features, pricing, audience and any weaknesses it lists. Wrong or outdated claims cost you recommendations even when your name appears.
You cannot edit ChatGPT’s answer directly, but you can fix the sources it is likely to draw on. Update your own pages first, then correct outdated profiles, old press releases and third-party articles that carry the wrong facts. If an answer cites a source, start there. Re-check the same prompts over the following weeks, remembering that knowledge from training updates only with new models, while search-based answers can change much sooner.
Step 6: Make product and page data machine-readable
If you sell products, ChatGPT can show product results with images, prices and links. OpenAI has said these results draw on structured product information, so make yours complete and consistent.
- Use Product markup with name, brand, price, availability and reviews on product pages, matching what the page shows.
- Keep product feeds accurate wherever you publish them, including marketplaces and comparison sites.
- Check OpenAI’s current merchant documentation for ways to share product data directly, since these programs are still evolving.
Mark up your other key pages too: Organization on your home or about page, Article on guides, and FAQPage where you answer questions on the page. The free FAQ schema generator creates valid FAQPage markup, and the llms.txt generator builds an optional file that points AI tools to your best pages. Neither is a confirmed ChatGPT ranking factor, but both reduce ambiguity. For a wider audit, the AI search optimization checklist lists 30 checks that apply to every engine.
Step 7: Measure recommendations over time
Because answers vary, measurement means sampling. Run your fixed prompt set on a schedule and record whether you are mentioned, in what position, whether you are cited and which competitors appear. Compare share of voice over weeks. The full setup is in how to track brand visibility in ChatGPT.
Add two more views. In analytics, look for visits from chatgpt.com; ChatGPT often adds a utm_source=chatgpt.com parameter to the links it shows, though not every visit carries it. In your server logs, check that OAI-SearchBot and ChatGPT-User fetch the pages you want recommended, and verify them against OpenAI’s published IP ranges. The AI crawler analytics docs show how aeoora reads this from your logs.
If you lose a recommendation or citation you used to have, use the diagnosis in how to win back lost citations before making broad changes.
Common mistakes
- Testing only branded prompts, which tells you ChatGPT knows your name but not whether it recommends you.
- Blocking OAI-SearchBot through a CDN default while trying to win ChatGPT search answers.
- Describing your product in slogans instead of plain facts a model can repeat.
- Planting fake reviews or promotional forum posts, which damages trust with people and platforms alike.
- Treating one screenshot, often from a personalized account, as the truth.
- Expecting changes to show up in training-based answers within days.
Frequently asked questions
Can I pay to be recommended by ChatGPT?
There is no public program that lets you pay for an organic recommendation in a ChatGPT answer. The route that works is the one in this guide: crawler access, clear positioning, third-party evidence and accurate facts.
Does blocking GPTBot stop ChatGPT from recommending me?
Not by itself. GPTBot is OpenAI’s training crawler, and OpenAI treats it separately from OAI-SearchBot, which is used for ChatGPT search. Blocking GPTBot can limit what future models learn from your site, but blocking OAI-SearchBot is what keeps your pages out of search answers.
How long does it take to start being recommended?
It depends on which layer you influence. Changes that affect search-based answers can show up within weeks once pages are crawled and indexed. Changes to what the model knows from training arrive only with new model releases, which you cannot schedule. Track a fixed prompt set and expect gradual movement.
Why does ChatGPT recommend a competitor and not me?
Look at the sources cited in that answer and at how the competitor describes itself. Common reasons are clearer positioning for the use case in the prompt, more reviews and roundup mentions, published pricing, or a comparison page that answers the question directly. Close the gaps that are real for your product.
Does ChatGPT give everyone the same recommendations?
No. Wording, conversation history, memory, custom instructions, location and the model in use all affect the answer. That is why you should measure across many prompts and repeated runs, ideally from a consistent, neutral setup, instead of relying on what one person sees.
Does SEO still matter for ChatGPT?
Yes. ChatGPT search relies on crawling and search indexes, so crawlable, well-structured, trustworthy pages help in both worlds. The difference is emphasis: plain facts, clear positioning and third-party corroboration matter even more when an assistant summarizes a category for the buyer. Our primer on answer engine optimization explains how the two fit together.
Track your ChatGPT recommendations with aeoora
aeoora runs your buyer prompts in ChatGPT every day, alongside Perplexity, Google AI Overviews, AI Mode, Gemini, Claude and Copilot. It shows where you are mentioned, your position in each answer against competitors, which pages are cited and how you are described, measures visits from ChatGPT with a cookie-free snippet, and suggests content and site fixes ranked by priority. Plans are Shine at $29, Signal at $69 and Scale at $299 a month, with 25% off yearly, and Scale adds ChatGPT Shopping for product visibility.
For a quick check first, get the free AI visibility report, which tests 25 buyer questions and emails you a PDF. To track your own prompts, start a 2-day free trial and follow the quickstart. Terms such as share of voice and citation are defined in the AI search glossary.


