A direct claim
One sentence that answers the question without hedging, early in the section. Assistants extract sentences, and a sentence that says nothing definite cannot be extracted.
Guide
How to get cited by AI.
ChatGPT, Claude and Perplexity build answers from pages they could fetch and read. This is the order of work that makes yours one of them, starting with the part nobody checks.
The order
Five steps, in sequence
Each step is wasted effort until the one before it is true. Work them in order.
-
Let the crawlers in
Resolve your robots.txt for all 15 named AI crawler tokens, then confirm your CDN agrees with it. An allowed crawler blocked at the edge is the most common invisible failure.
-
Ship the words in the HTML
Measure Prose Parity. If your content arrives after hydration, no amount of writing will help, because none of it reaches the crawler.
-
Answer one question per page
Retrieval systems quote passages, not pages. A page that answers one question clearly is quotable; a page that covers six topics is a summary nobody can cite.
-
Put the answer near the top
State the answer in the first paragraph, then explain it. Assistants extract the passage that most directly answers the query, and burying it under context costs you the quote.
-
Be specific enough to be worth quoting
Numbers, dates, named methods and stated limits. Generic marketing copy is interchangeable, and an assistant has no reason to prefer yours.
By platform
The per-platform guides
Being cited by an AI assistant starts with being readable and allowed. These per-platform guides show how each one reaches the web, and what it takes to be a source it can use.
Guide
Get cited by ChatGPT
ChatGPT can only cite a page its crawlers are allowed to fetch and can read without JavaScript, so being permitted and readable comes first, and any citation follows from there.
GuideGet cited by Perplexity
To be cited by Perplexity, a page first has to be readable: it must fetch cleanly, carry its real text in the raw HTML before any JavaScript runs, and stay open to PerplexityBot in robots.txt.
GuideGet cited by Google AI Overviews
Google AI Overviews are built from Google's Search index, so getting cited starts with a page Googlebot can crawl, index, and read, and no site can promise the citation itself.
GuideGet cited by Claude
Claude reaches your page through named Anthropic crawlers that honor robots.txt and work from your raw HTML, so being cited starts with being fetchable, allowed, and readable without JavaScript.
The writing
What makes a passage quotable
Once the page is readable, these are what decide whether it gets used rather than another one.
Attributable detail
A figure, a date, a named source. Detail that can be checked is detail worth repeating, and it is what distinguishes a source from a paraphrase.
Clean structure
Headings that name the question, paragraphs that hold one idea, lists where the content is genuinely a list. This is how a retrieval system chunks your page.
Stated limits
Saying what your claim does not cover makes the rest more credible, and models reproduce hedged claims more faithfully than absolute ones.
Common mistakes
What does not work
Four things that get recommended regularly and do nothing measurable.
| The tactic | Why people try it | What actually happens |
|---|---|---|
| Keyword stuffing | Habit from classic SEO | Assistants extract meaning, not term frequency. Density has no effect on whether a passage is chosen. |
| Writing for the model | The belief that AI reads differently | The same clear prose that helps a reader helps a retrieval system. There is no separate machine dialect that outperforms plain writing. |
| Publishing more pages | Volume worked for search | Ten thin pages behind JavaScript reach a crawler as zero pages. One readable page beats all of them. |
| Adding llms.txt alone | It feels like a switch | No major vendor has committed to reading it. Serve one, but do not expect it to substitute for a readable page. |
Citation has a prerequisite, and almost everybody skips it. Before an assistant can quote you it has to reach the page, receive the words, and find the passage that answers the question. Most advice on this subject starts at step three and assumes the first two are fine. They usually are not.
Nothing here promises a citation. These are the conditions that make one possible, in the order they have to be met.
Common questions
Can I make ChatGPT cite me?
No, and nobody can. What you can do is remove every reason it cannot: let the crawler in, ship the words in the HTML, and write a passage that answers a real question specifically enough to be worth quoting. Anyone selling guaranteed placements is selling something they do not control.
How long does it take to see a change?
The readability change is immediate and measurable: re-scan after the deploy and the number moves. Whether assistants start using the page is outside our measurement and outside anyone's control, which is why this site reports the first thing and not the second.
Does being in training data matter?
Less than people think, and it is not something you can influence retroactively. AI search fetches pages and answers from what it reads, so a page that grounds a clear answer today matters more than what a model absorbed two years ago.
Should I block training crawlers?
That is a business decision, not a technical one. Blocking training while allowing search is a coherent position and the crawler directory has the exact robots.txt lines for it. Just be aware the two are separate tokens and blocking one does not block the other.
What is the single highest-impact change?
For most sites, moving prose out of client-side rendering. It is half the score, it is usually where the whole deficit is, and it is a change your existing framework already supports.
Start with the prerequisite.
Steps three to five are writing work. Steps one and two are measurable in under a minute, and until they are true the rest cannot help.