What does it take to be cited by AI search such as ChatGPT and Perplexity?
First, AI crawlers must be able to read the site (robots.txt). Second, each document must give one answer to one question. Third, structured data must state what kind of document it is and who wrote it. Without these three, no matter how good the content is, it will not become a candidate for citation.
Step 1: allow AI crawlers in
Each AI service collects the web with its own crawler. Even if you allow ordinary search bots, AI crawlers are often blocked separately.
Check in robots.txt that the names below are not blocked. Conversely, if you do not want your content used for training, you can block them on purpose. Citation and training are different matters, so it is better to decide your policy first.
- GPTBot, OAI-SearchBot, ChatGPT-User: OpenAI
- ClaudeBot, Claude-Web: Anthropic
- PerplexityBot: Perplexity
- Google-Extended: Gemini and AI Overviews
Step 2: write sentences that are easy to quote
AI does not copy a whole document. It picks out the sentences it needs for the answer. So sentences that still make sense when lifted out of context are the ones that get chosen.
A sentence like "We provide the best service" carries no information when quoted, so it gets discarded. Sentences with conditions and values survive, such as "Weekday evening clinic hours run until 21:00 on Tuesdays and Thursdays."
Step 3: use structured data to state what the document is
Even for the same text, a machine has a hard time being sure from the body alone whether it is an FAQ, a product description or a company introduction. Structured data (JSON-LD) makes this explicit.
Stating who wrote it also matters. When the publishing organization is shown, it is used in judging trust.
Step 4: is there a document worth citing?
This is the step most often missed. A site with only a company introduction and a price list has nothing to cite. AI can answer "What does this company do?" but has no reason to list the company as the source of an answer.
The starting point is to build up documents whose titles are the actual questions your customers ask.
Frequently asked questions
How long does it take to be cited?
A document has to be indexed before it becomes a candidate, and for a new site indexing alone can take several weeks. There is no way to guarantee a timeline.
Does having llms.txt guarantee citations?
No. It is a supplementary way to tell AI crawlers about the structure of your site, and not every service has announced that it reads it. It is more accurate to think of it as something that does no harm to have.
References
- Google Search Central — AI features and your website (새 창에서 열림)
- The llms.txt proposal (새 창에서 열림)
Related reading
WiSite handles this for you automatically
The technical elements described here are applied automatically when your site is created.