How to get cited by ChatGPT, Claude and Perplexity
Hephaestus Technology6 min read

To get cited by ChatGPT, Claude and Perplexity, first make sure their search crawlers can reach your site, then publish pages that answer your buyers' specific questions directly, with facts an assistant can check and quote. Nobody can guarantee a citation, but these steps remove the most common reasons a B2B website never appears in AI answers.
Key takeaways
- AI assistants can only cite pages their search crawlers can reach: OAI-SearchBot for ChatGPT, Claude-SearchBot for Claude and PerplexityBot for Perplexity.
- Blocking a training crawler (GPTBot, ClaudeBot) is a separate decision from blocking a search crawler; mixing them up is the most common mistake.
- Citations go to pages that answer a specific question directly, with facts the assistant can verify and quote.
- No one can guarantee citations, and AI answers vary between runs, so measure trends over months rather than single answers.
- A monthly routine of prompt testing, referral tracking and page updates is enough for most SMEs.
What makes an AI assistant cite a website?
An AI assistant cites a website when its search system retrieves a page that answers the user's question and judges it reliable enough to quote. ChatGPT, Claude and Perplexity all run web search behind the scenes: they look up pages, read them, write an answer and link to some of the sources.
That means two conditions must be met. Your page has to be in the assistant's reach (crawled and indexed), and it has to be the best available passage for that question. The original research behind generative engine optimisation, a 2023 paper accepted to KDD 2024, found that content changes could raise visibility in generative engine responses by up to 40% in test conditions, with results varying by domain. Content matters, but only after access is in place.
Step 1: Are the right crawlers allowed?
Check your robots.txt first, because a blocked search crawler makes every other step pointless. Each assistant separates search crawling from training crawling:
| Assistant | Search crawler (controls citations) | Training crawler | User-initiated fetcher |
|---|---|---|---|
| ChatGPT | OAI-SearchBot | GPTBot | ChatGPT-User |
| Claude | Claude-SearchBot | ClaudeBot | Claude-User |
| Perplexity | PerplexityBot | Not used for training, per Perplexity | Perplexity-User |
OpenAI states that sites opted out of OAI-SearchBot "will not be shown in ChatGPT search answers", that GPTBot is a separate control for training, and that robots.txt changes take about 24 hours to take effect. Anthropic says disabling Claude-SearchBot "prevents our system from indexing your content for search optimization", which may reduce your visibility. Perplexity recommends allowing PerplexityBot so your site can appear in its results.
Also check your CDN or hosting firewall. Cloudflare began blocking AI crawlers by default for new domains in July 2025, so a site set up since then may be blocking more than its owner realises. Our guide to AI crawlers and robots.txt choices walks through each decision.
Step 2: Which questions should you try to be cited for?
Pick questions where you have real expertise and where a buyer's answer could lead to your door. Good sources:
- Questions prospects ask on sales calls and in emails.
- Comparison questions ("X vs Y for a 50-person firm").
- "How much does it cost" and "how long does it take" questions in your field.
- Regulatory and practical questions specific to Ireland, the UK or the EU.
Write down 15 to 30 of these in the words a buyer would use. Generic questions such as "what is cloud computing" are dominated by large publishers; narrow, specific questions are where a smaller firm can win.
Step 3: How should each page be written?
Each page should answer one question in its opening two or three sentences, then support that answer with detail. Practical rules:
- Put the question, or something close to it, in the title.
- Answer it directly in the first paragraph, in sentences that make sense when quoted alone.
- Use headings that are the follow-up questions a buyer would ask.
- Include at least one thing worth quoting: a clear definition, a comparison table, a short numbered process or a sourced figure.
- Link every statistic to its original source.
- Give a date and a named author or company so the page can be judged for freshness and expertise.
Google's guidance applies to other assistants too: it recommends "unique, non-commodity content" in its advice on succeeding in AI search. Our 12-point checklist for content AI assistants cite turns this into a review you can run on any page.
Step 4: How do you give assistants evidence worth citing?
Publish facts that are yours. An assistant choosing between ten similar summaries has no reason to pick yours; one that finds the only page with real figures does. Options for a B2B SME:
- Anonymised data from your own projects (typical timelines, price ranges).
- A short annual survey of your customers or sector.
- Clear, public pricing or pricing ranges.
- Worked examples, clearly labelled as hypothetical where they are.
Keep facts consistent everywhere they appear: your website, LinkedIn page, directory listings and press mentions. Contradictory founding dates or service lists make it harder for any system to state facts about you confidently.
Step 5: How do you check whether it is working?
Test a fixed set of prompts monthly and track referral traffic. A simple routine:
- Run your 15 to 30 questions in ChatGPT, Claude and Perplexity with web search on, in a logged-out or fresh session where possible.
- Record whether your site is cited, which page, and which competitors appear.
- In your analytics, watch referrals from chatgpt.com, claude.ai and perplexity.ai, where the assistant passes referrer data.
- For Microsoft Copilot, Bing Webmaster Tools' AI Performance report shows citation counts and the grounding queries used.
- For Google, the generative AI performance report in Search Console shows AI impressions.
AI answers vary between runs, so look at trends over months, not single results.
Your checklist for this week
- [ ] robots.txt allows OAI-SearchBot, Claude-SearchBot and PerplexityBot (unless you have decided otherwise)
- [ ] CDN or firewall settings checked for AI crawler blocking
- [ ] 15 to 30 buyer questions listed in buyer language
- [ ] Top five questions each have a page that answers in the first paragraph
- [ ] Each page has one quotable element: definition, table, steps or sourced figure
- [ ] Company facts consistent across site and profiles
- [ ] Monthly prompt test scheduled, with a simple results sheet
- [ ] AI referrals segmented in analytics
Frequently asked questions
Can I pay to be cited by ChatGPT?
Not in the organic answers. Citations in ChatGPT search answers depend on its search systems finding your content relevant, and OpenAI's crawler documentation offers no paid route into them. Be wary of any service that promises guaranteed placement.
Does blocking GPTBot stop ChatGPT from citing my site?
No, not on its own. OpenAI states that GPTBot governs training use, while OAI-SearchBot governs whether your site appears in ChatGPT search answers. You can block GPTBot and still allow OAI-SearchBot.
How long does it take to start appearing in AI answers?
There is no published timeline. OpenAI says robots.txt changes for its crawlers take about 24 hours to be picked up, but being crawled does not mean being cited. Expect to judge progress over two or three months of consistent testing.
Why does ChatGPT cite my competitor and not me?
Usually because their page answers the specific question more directly, is easier to quote, or is referenced by other sites. Compare the cited page with yours: look at how quickly it answers, whether it gives numbers or examples, and whether its facts are sourced.
Do Claude and Perplexity follow robots.txt?
Their automated crawlers do. Anthropic says its bots honour robots.txt, and Perplexity says the same for PerplexityBot. Fetches made on behalf of a user, such as Perplexity-User and OpenAI's ChatGPT-User, may not follow robots.txt rules, according to their documentation.
If you would like help working through this list for your own site, book a free 30-minute call with our team.
Cover photo by Christina Morillo on Pexels.