How to get your business cited by ChatGPT and Perplexity
What OpenAI, Perplexity and Google say about appearing in AI answers, which AI crawlers to allow in robots.txt, and how to track AI visits and leads in GA4.
How to get cited by ChatGPT, Perplexity and Google’s AI answers comes down to two jobs: let their search crawlers reach your site, and give them pages with specific facts worth quoting.
In short:
- ChatGPT can only cite your site in search answers if OpenAI’s search crawler, OAI-SearchBot, can reach it. OpenAI says it ranks results “using multiple factors” and that “placement is not guaranteed”.
- Perplexity uses PerplexityBot to surface and link sites in its answers. Allow it in robots.txt.
- Google says there are “no additional requirements to appear in AI Overviews or AI Mode”. A page needs to be indexed and eligible for a snippet, and Google says you don’t need AI text files or special markup.
- Training crawlers (GPTBot, ClaudeBot, Google-Extended, Applebot-Extended) are separate from search crawlers. You can block them and still appear in ChatGPT, Perplexity, Claude and Google Search answers. Google-Extended also covers grounding in the Gemini app, so blocking it may affect answers there.
- A 2023 study by researchers at Princeton University and IIT Delhi found that adding quotations, statistics and cited sources raised a page’s visibility in AI answers by up to 40% in its tests. Keyword stuffing didn’t help. The engines it tested have changed a lot since.
- Since 13 May 2026, GA4 has an AI Assistant channel for visits from tools such as ChatGPT. It leaves out Google’s own AI Overviews and AI Mode. Add a hidden form field to carry the source into your CRM.
How to get cited by ChatGPT, Perplexity and Google
Their documentation covers what makes a site eligible, and a little about how answers are put together.
ChatGPT
OpenAI’s help article on searching the web with ChatGPT says ChatGPT “ranks search results using multiple factors intended to help users find relevant, reliable information. Placement is not guaranteed.” To be eligible, allow OAI-SearchBot and make sure your host or CDN lets through traffic from OpenAI’s published IP addresses. The same article says ChatGPT sometimes sends rewritten versions of a question to partner search providers, so being easy to find in ordinary search engines probably helps here too.
Perplexity
Perplexity’s crawler page says PerplexityBot “is designed to surface and link websites in search results on Perplexity” and is not used to collect training data.
Google’s guide to AI features and your website says the usual SEO basics apply and “there are no additional requirements to appear in AI Overviews or AI Mode”. To be shown as a supporting link, a page “must be indexed and eligible to be shown in Google Search with a snippet”. Both features may run several related searches behind the scenes (Google calls it query fan-out), and because they “may use different models and techniques”, the links they show vary. Google’s list of basics includes allowing crawling in robots.txt and at your CDN, keeping structured data in line with the visible text, and keeping your Business Profile up to date.
Which AI crawlers should you allow in robots.txt?
Each company runs one bot for search, often one for fetching a page when a user asks, and one for training. According to each company’s own documentation:
| Company | Bot | What it does | What blocking it does |
|---|---|---|---|
| OpenAI | OAI-SearchBot | Finds sites for ChatGPT search | Your pages stop appearing in ChatGPT search answers, though they can still show as plain navigation links |
| OpenAI | ChatGPT-User | Visits a page when a user asks | OpenAI says robots.txt rules “may not apply” |
| OpenAI | GPTBot | Collects training data | Signals your content shouldn’t be used for training |
| Perplexity | PerplexityBot | Finds sites for Perplexity search | Your pages drop out of Perplexity search |
| Perplexity | Perplexity-User | Visits a page when a user asks | Perplexity says it “generally ignores robots.txt rules” |
| Anthropic | Claude-SearchBot | Indexes pages for Claude’s search | Anthropic says it “may reduce your site’s visibility” |
| Anthropic | Claude-User | Visits a page when a Claude user asks | Claude can’t fetch your pages for users |
| Anthropic | ClaudeBot | Collects training data | Future content is excluded from training |
| Googlebot | Google Search, including AI Overviews and AI Mode | You leave Google Search | |
| Google-Extended | Gemini training, and grounding in Gemini Apps and Vertex AI | No effect on Google Search | |
| Apple | Applebot | Search in Spotlight, Siri and Safari | You drop out of those features |
| Apple | Applebot-Extended | Training Apple’s foundation models | Content isn’t used for training and stays in Apple search |
Sources: OpenAI, Perplexity, Anthropic, Google and Apple, all checked on 12 November 2026.
Allow the search bots. Training is a separate decision about your content. (Whether ChatGPT trains on what your staff type into it is another question, covered in is it safe to put confidential documents into ChatGPT.) Blocking the training bots keeps you in ChatGPT, Perplexity, Claude, Apple and Google Search answers. The one overlap is Google-Extended, which also controls grounding in the Gemini app. A robots.txt that allows search and blocks training looks like this:
# Search and user-requested visits
User-agent: OAI-SearchBot
User-agent: ChatGPT-User
User-agent: PerplexityBot
User-agent: Perplexity-User
User-agent: Claude-SearchBot
User-agent: Claude-User
User-agent: Googlebot
User-agent: Applebot
Allow: /
# Training. Delete this group if you're happy for your content to be used.
User-agent: GPTBot
User-agent: ClaudeBot
User-agent: Google-Extended
User-agent: Applebot-Extended
Disallow: /
User-agent: *
Disallow: /admin/Two details trip people up. Under the robots.txt standard, RFC 9309, a bot that finds a group naming it ignores the * group, so copy any Disallow lines you need (like /admin/ above) into the named groups too. And check your CDN or firewall. Bot protection that blocks unknown crawlers can block these as well, which is why OpenAI and Google both mention it.
A newer signal is on the way. Anthropic began recognising a machine-readable “no AI training” preference, based on a draft from the IETF’s aipref working group, on 30 October 2026, and is rolling it out across its services. Google says it intends to support the same signal in early 2027.
What did the 2023 GEO research find?
The paper is GEO: Generative Engine Optimization by researchers at Princeton University and IIT Delhi, first posted in November 2023 and accepted at KDD 2024. They built a benchmark of 10,000 queries and a test engine on GPT-3.5, then rewrote source pages in different ways to see which rewrites got more of the page into the answer.
- Their best methods raised visibility by up to 40%. Adding quotations, adding statistics and citing sources did best, with gains of 30 to 40% on their main measure.
- Keyword stuffing gave “little to no improvement”. On a smaller test with Perplexity, it did 10% worse than leaving the page alone.
The research is three years old and was built on GPT-3.5, and every engine it studied has changed since. The direction still matches what you’d want on a page anyway: specific numbers with sources, quotes from named people, and no padding.
How do you test whether AI assistants mention you?
Pick 5 to 10 prompts a buyer might type, and run the same set every week:
- Who are the best [category] in [city]?
- Recommend a [category] in [city] for [a specific need].
- Is [your company] a good [category]?
- [Your company] or [competitor]: which is better for [need]?
- How much does [service] cost in [city]?
Score each answer the same way every time:
| What to record | How |
|---|---|
| Mentioned | 1 if your company is named, otherwise 0 |
| Linked | 1 if your site is among the cited sources, otherwise 0 |
| Position | Where you appear among the companies named |
| Cited sources | Each cited URL, tagged as directory, review site, your site, competitor site, news or blog, forum, or other |
| Errors | Anything wrong about you, such as old prices or a service you don’t offer |
Run it in ChatGPT with search on, in Perplexity and in Google AI Mode. Use a clean account or a private window, keep the same location, and save the full answer text. After four weeks, look at two numbers: the share of runs that mention you, and the domains cited in at least half of the runs. The directories and review sites among them are the ones worth being listed on. Expect the answers to vary from week to week. Google says outright that its two AI features show different links, and ChatGPT rewrites questions before it searches.
To see how a whole market behaves, run the same set across 10 to 20 business categories in your area. You’ll see whether directories, review sites or company sites get cited most in each one.
How do you track AI visits and leads in GA4 and your CRM?
Google added an AI Assistant channel to GA4’s default channel group on 13 May 2026. Visits whose referrer matches Google’s list of AI assistants get the medium ai-assistant and the campaign (ai-assistant). Google’s channel definitions name ChatGPT, Gemini, Deepseek, Copilot and Grok, and the announcement names Claude. The channel excludes Google’s AI Overviews and AI Mode, which GA4 counts under Organic Search instead. Perplexity isn’t named on either page, so also check GA4’s Referral report for perplexity.ai.
GA4 tells you about visits. To know which leads came from AI answers, put the source on the lead itself. Add a hidden field to your contact form and this script to every page:
<input type="hidden" name="lead_source">
<script>
// Runs on every page. Keeps the first source seen in this tab.
if (!sessionStorage.getItem("lead_source")) {
const utm = new URLSearchParams(location.search).get("utm_source");
const ref = document.referrer ? new URL(document.referrer).hostname : "";
sessionStorage.setItem("lead_source", utm || ref || "direct");
}
const field = document.querySelector('input[name="lead_source"]');
if (field) field.value = sessionStorage.getItem("lead_source");
</script>It stores the first source a visitor arrives from, such as chatgpt.com or a utm_source value, and fills the field when they reach the form. Map lead_source to a field in your CRM. Also ask “How did you hear about us?” as a free-text question. Someone who reads your name in an AI answer and searches for it later arrives as organic or direct traffic, and only they can tell you where they first saw you.
Does llms.txt do anything?
llms.txt is a proposal, first published by Jeremy Howard in September 2024 and revised in August 2026, for a plain-text file that helps AI agents use a website. Its author notes that thousands of sites publish one and that Chrome’s Lighthouse now checks for it.
The AI platforms’ own documentation gives no reason to expect it to help with citations. Google says you “don’t need to create new machine readable files, AI text files, or markup” to appear in its AI features. None of the crawler documentation from OpenAI, Perplexity, Anthropic or Apple says their bots read it.
It costs little to add one. To see whether anything reads it, check your server’s access log. On a standard combined-format log, this counts requests for the file by user agent:
grep '"GET /llms.txt' access.log | awk -F'"' '{print $6}' | sort | uniq -c | sort -rnA monthly checklist
- Check robots.txt and your CDN’s bot settings still let OAI-SearchBot, PerplexityBot, Claude-SearchBot, Googlebot and Applebot through.
- Confirm your main pages are indexed in Google Search Console.
- Run your prompt set and log mentions, links and cited domains.
- Fix wrong facts where they come from: your own pages, your Google Business Profile and the directories that get cited.
- Add specifics to pages that should be cited: prices, sourced numbers and quotes from named people.
- Compare GA4’s AI Assistant channel and Perplexity referrals with the
lead_sourcevalues in your CRM.
Our AI search visibility add-on is work on how your company shows up when people ask AI assistants about your field. It costs $750 one-off, then $300 a month, and you can book a call to ask about it.