How to get cited by ChatGPT, Perplexity and Gemini

Google’s AI Overviews get most of the attention, but a Delhi business owner asking “which digital marketing institute in Rohini” is as likely to ask ChatGPT or Perplexity as Google. The three systems choose sources differently. Here is what they have in common, where they differ, and what to do about it.

What all three have in common

They fetch and read real pages. None of them answers purely from training data any more for a question like this. They run a search, retrieve pages, and synthesise. So a page that is blocked in robots.txt, or that renders its content only in JavaScript, is invisible to them regardless of how good it is.

They prefer pages that answer directly. A passage that can be lifted whole — question answered, fact stated, no context needed — beats a better page where the answer is spread across six paragraphs.

They weight recency and dates. A page that says “as of September 2026” tends to be preferred over one with no date, especially for anything involving prices, fees or “the best X”.

They cross-check. If your page states a fact that no other source agrees with, it is less likely to be quoted. Consistency across your site, your Google Business Profile and third-party listings is a citation signal, not just a trust signal.

Where they differ

	Google AI Overviews	Perplexity	ChatGPT (search)
Source pool	Google’s index, weighted by normal ranking	Its own crawl plus live search	Bing-backed search plus its own fetching
Typical citations	3–6, often from pages already ranking top 10	5–10, shows them prominently	Fewer, less prominent
Crawler to allow	Googlebot (and Google-Extended for training)	PerplexityBot	GPTBot, OAI-SearchBot
Favours	Established ranking + structured facts	Fresh, specific, well-sourced pages	Clear direct answers

The practical consequence: you cannot block AI crawlers and expect to be cited by AI assistants. Plenty of Indian sites currently disallow GPTBot in robots.txt on principle and then wonder why ChatGPT never mentions them.

A checklist for a page you want cited

1. One question per heading, answered in the first two sentences under it.
2. Facts in tables. Fees, durations, comparisons, steps.
3. A date on every claim that can change. Prices, “best”, “latest”.
4. Something original. A number you measured, a test you ran, a screenshot you took. Summaries of other people’s pages do not get cited over those pages.
5. Consistent facts everywhere. Site, GBP, LinkedIn, directories — same name, address, phone, fees.
6. All relevant crawlers allowed. Check robots.txt for GPTBot, ClaudeBot, PerplexityBot, Google-Extended.
7. A text/plain /llms.txt with your key facts. Google ignores it; others may fetch it. It costs nothing.
8. Schema that matches the page. Course with a real Offer.price, FAQPage whose answers are the same words as on the page. Contradictions between schema and visible text are a reason to be skipped.

What does not work

• Naming the assistants in your headings. “Best course for ChatGPT” cites nobody.
• Fake “as seen on” badges or made-up statistics. Cross-checking is exactly what these systems do.
• A hundred thin pages. Scaled content enforcement in 2026 hits the whole domain, not just the thin pages.

How we know

Every student on the AI SEO course runs this test on a real site in week one, and again in the final project.