Short answer
ChatGPT, Perplexity and Claude can only cite pages their search crawlers are allowed to fetch. Most sites that never get cited are blocking the wrong bot, usually by blocking AI training crawlers and catching the search crawlers with them. Fix access first. Content comes second.
Google’s AI opt-out toggle does not touch any of these. They are separate systems with separate crawlers, covered in should you opt out of Google’s AI features.
So a site can be fully visible in Google AI Overviews and completely absent from ChatGPT. The reverse happens too.
The crawlers, and which one matters
Each platform runs more than one bot, and they do different jobs. Confusing them is the most common mistake.
| Platform | Search bot (gets you cited) | Training bot (does not) |
|---|---|---|
| ChatGPT | OAI-SearchBot, ChatGPT-User | GPTBot |
| Claude | Claude-SearchBot, Claude-User | ClaudeBot |
| Perplexity | PerplexityBot, Perplexity-User | — |
| Copilot | Bingbot | — |
The middle column is what decides citation. Blocking GPTBot opts you out of training but leaves ChatGPT search untouched. Blocking OAI-SearchBot removes you from ChatGPT search entirely.
Crawler names and behaviour change, so check each platform’s current documentation before editing production rules.
The three ways sites block themselves
- A blanket AI block in robots.txt. One rule meant for training bots catches the search bots too.
- A CDN or firewall toggle. Many hosts now offer a one-click “block AI bots” setting. It often blocks the answer bots you wanted. This one is invisible in robots.txt.
- Content behind JavaScript. If the text only appears after a script runs, some crawlers see an empty page.
The CDN case is the one that catches experienced teams, because robots.txt looks perfectly clean while the firewall refuses the request.
How to check your own site
- Read your robots.txt at yoursite.com/robots.txt. Look for Disallow rules naming any search bot in the table above.
- Check your CDN or host settings for any AI-blocking toggle.
- Request a page as the bot. Fetch it with the crawler’s user-agent and confirm you get a full page, not a block or an empty shell.
- Check your server logs for the search bots actually visiting.
I ran exactly this on this site. Robots.txt names no AI bot, and requests made as OAI-SearchBot, PerplexityBot and Claude-SearchBot each returned the full page. Step three is the one that catches CDN blocks.
What none of them publish
No platform publishes its selection formula. Perplexity states it searches the web in real time and links citations to sources, but discloses no weights or thresholds.
Be wary of precise percentages about which sites each engine prefers. Most come from third-party studies with methods you cannot check, and they shift between samples.
What is consistently observed: ChatGPT tends to cite a small number of sources it treats as authoritative, while Perplexity cites more per answer and leans toward recent pages. A citation on one says little about the other.
What you can actually control
Once access is fixed, the same things that earn Google citations tend to help here.
- Answer first. The direct answer in the first sentence under each heading, then the explanation.
- Be specific. A figure nobody else publishes is quotable. A summary of what everyone says is not.
- Cite your own sources. Claims linked to where they came from read as verifiable.
- Keep it current. Dated, accurate information is safer to cite than stale information.
- Server-render your content. Text in the HTML, not loaded later by a script.
More on structure for citation is in generative AI impressions dropped in Search Console.
The honest limit
Unlike Google, these platforms give you almost no reporting. There is no equivalent of Search Console showing how often you were cited.
So the practical approach is fix access, improve the content, then check manually by asking each platform the questions your pages answer. It is imprecise, and anyone selling a guaranteed citation is selling something no platform lets them control.
One file often recommended for AI visibility, llms.txt, is not documented as read by any major AI search crawler. Details in llms.txt: does it do anything yet?.
A robots.txt example
If you want AI search visibility but not model training, the rules look like this. Check each platform’s current documentation before using them, because crawler names change.
# Let AI search crawlers fetch and cite pages
User-agent: OAI-SearchBot
Allow: /
User-agent: Claude-SearchBot
Allow: /
User-agent: PerplexityBot
Allow: /
# Opt out of model training only
User-agent: GPTBot
Disallow: /
User-agent: ClaudeBot
Disallow: /Allowing a bot in robots.txt does not help if your CDN blocks it. Test with a real request, as described above.
Blocking training is a business decision, not an SEO one. It has no documented effect on whether these platforms cite you in their search answers.
Check whether you are being cited
Test your key questions every month in ChatGPT search, Perplexity and Claude, and note which sources each one cites.
Then check your analytics for referral visits from chatgpt.com, perplexity.ai and claude.ai. Citations without visits tell you the answer satisfied people on the platform, which is still brand exposure, but a different result from traffic.
Record the results in one sheet over time. A single month tells you little; six months shows whether your citations are growing, holding or slipping.
Visible in Google, invisible in ChatGPT?
The cause is usually a blocked crawler rather than your content, and a CDN block will not show in robots.txt. Send your site and within 48 hours I will tell you which AI search bots can actually reach it.
Get free recovery audit