How to get your brand cited by ChatGPT and Perplexity

To get cited by ChatGPT and Perplexity, let their search crawlers reach your pages, make those pages the clearest answer to the questions your buyers ask, and make sure the sites these engines already cite describe you correctly.

There is no form to submit and no tag to add. Both engines write answers from pages they can read at the time of the question, plus what their models learned in training. This guide covers the parts you control, in the order we work on them. It assumes you already have a site worth citing. If your product page does not say what the product does, fix that first.

Step 1. Make sure they can read your site

Both vendors publish the names of their crawlers, and both separate training from search.

Sources: OpenAI crawler documentation, Perplexity crawler documentation. OpenAI states that each of its settings is independent. You can allow OAI-SearchBot to appear in search results and still disallow GPTBot if you do not want your content used for training.

A robots.txt that allows the search crawlers explicitly looks like this:

User-agent: OAI-SearchBot
User-agent: ChatGPT-User
User-agent: PerplexityBot
User-agent: Perplexity-User
Allow: /

robots.txt is only one gate. Check two more:

• Your CDN and firewall. Bot-management features can block or challenge AI crawlers regardless of robots.txt. Ask whoever runs them to confirm that the user agents above are not blocked or served a challenge page.
• Rendering. Fetch a key page with curl -s https://yoursite.com/page and search the output for a sentence from the page. If the text is missing, it is being added by JavaScript after load. A crawler that does not run scripts sees an empty page.

Step 2. Find the questions that matter

Do not start with keywords. Start with the questions a buyer would type into ChatGPT or Perplexity before choosing a product like yours. Good sources:

• Questions from sales calls and demo requests, in the customer's own words.
• Support tickets that begin with “how do I” or “can your product”.
• Queries in Google Search Console that are phrased as questions.
• Threads in the forums and communities where your buyers talk.

Group them by intent: discovery (“best X for Y”), comparison (“A vs B”), problem (“how do I fix Z”) and brand (“is A any good”). Twenty to thirty questions is enough to start. Write them down exactly; you will ask the same wording again later.

Step 3. See who gets cited today

Ask each question on both engines and record the answer. Perplexity lists its sources with each answer. ChatGPT shows sources when it searches the web. For each answer, note:

• Which brands are named, and in what order.
• Which domains are cited as sources.
• Whether your brand is named, cited, recommended, or absent.
• Anything the answer says about you that is wrong.

Ask each question at least 3 times. The same prompt can name different brands on different runs, so one answer is a weak signal. After this step you have two lists that matter more than any keyword report: the competitors that keep appearing, and the domains that keep being cited.

Step 4. Make your pages the clearest answer

When an engine writes an answer, it tends to quote or paraphrase a short passage. Make that passage exist on your site, and make it easy to lift without changing its meaning.

Answer first

If a page exists to answer “What does X cost?”, the first sentence under that heading should give the price or the pricing model. Context comes after.

Use structure the engine can read

Definitions in one sentence. Comparison tables in HTML, not in images. Steps as numbered lists. Headings phrased as the question a buyer asks.

State facts with dates and sources

“Supports exports to CSV and XLSX since version 4.2 (March 2026)” is citable. “Industry-leading export options” is not. Where a claim depends on data, link the source.

Publish what buyers need to compare you

Engines cannot recommend you for a use case your site never mentions. If you are a good fit for small architecture firms, say so on a page that explains why, with specifics. If pricing is public elsewhere, make it public on your own site too, so the engine reads your version.

Keep facts consistent

Company name, product names, category, locations and founding details should match across your site, your structured data, your social profiles and your directory listings. Conflicting versions give the engine a reason to be vague, or wrong.

Step 5. Be described correctly where they already look

Go back to the list of domains from Step 3. For most categories it includes a few review platforms, some comparison articles, one or two communities and a handful of industry publications. This list is your outreach plan.

• Review and listing sites: claim your profile, complete it, and correct outdated details.
• Comparison articles: if you are missing, contact the author with the facts they would need to include you. Some will update the article.
• Communities: take part under your own name when a question is genuinely about your field. Disclose your affiliation.
• Errors: when an answer states something false about you, find which cited source says it and ask for a correction there.

Do not buy posts presented as independent, and do not post fake reviews. Apart from the ethical problem, platforms can remove them, and the brand is left with the damage.

Step 6. Measure again

Run the same questions, with the same wording, on the same engines, a month after your changes. Compare how often you are named and cited with the first run. Keep a note of what you changed and when, so that you can connect changes to results.

Set expectations by knowledge source. Changes to your pages and to third-party sources can appear in answers that use live search once those pages are recrawled. What a model learned in training changes only when the vendor releases a new model. If an engine answers a question without searching, your recent work may not show until then.

What does not work

• Hidden instructions to AI models in white text or HTML comments. It is manipulation, it can be detected, and it risks the brand.
• Mass-produced pages that restate the same answer for every city or keyword. They add nothing an engine needs to quote.
• Keyword stuffing. Engines paraphrase meaning; repeated phrases do not help.
• Relying on llms.txt alone. It is a proposed convention, not a mechanism that places you in answers.

A one-page checklist

1. OAI-SearchBot, ChatGPT-User, PerplexityBot and Perplexity-User are allowed in robots.txt and not blocked by your CDN or firewall.
2. Key pages return their main text in the HTML.
3. You have a written list of 20 to 30 buyer questions, grouped by intent.
4. You know which competitors and which domains appear in answers today.
5. Each important question has a page on your site that answers it in the first sentence.
6. Your core facts match across your site, structured data and profiles.
7. You have a list of third-party sources to correct or join.
8. You have a date to re-run the same questions.

We run this process as the Citation Loop. If you want to see where you stand first, request a free GEO audit.