How do ChatGPT, Claude, Perplexity and Google read a product site?
They read the words people can see. Google crawls, renders and ranks, and uses structured data for rich results. AI assistants like ChatGPT, Claude and Perplexity fetch the page text and quote it, and most never see code hidden in scripts. So the answer has to live in the visible copy, say the same thing on every page, and be easy to find.
A product site now has 2 kinds of reader: people, and the machines that answer people's questions. Both reward the same thing, a clear answer in plain words.
Google. Google crawls the page, renders it, and ranks it. Structured data (JSON-LD) helps it show rich results like FAQs and prices. Page speed, headings and links between pages still matter.
AI assistants. ChatGPT, Claude and Perplexity send their own crawlers (GPTBot, ClaudeBot, PerplexityBot) or fetch pages live when someone asks. They read the page text and quote it. Code inside scripts, including most structured data, is usually dropped, so a fact that only lives there is invisible to them.
Answer first. Put the question as the heading and a complete answer in the first 2 or 3 sentences. That paragraph is what gets quoted. The detail follows for the people who stay.
One fact, one wording. If the pricing page says 1 review round and the FAQ says 2, an assistant repeats whichever it read last. Every price, day and number should read the same on every page, in the FAQ, and in the assistant's own notes.
Let them in, and give them a map. Allow the AI crawlers in robots.txt. Add an llms.txt at the root: a short, plain summary of who you are, what you sell and where the key pages are. It is the page an assistant can read in one pass.
This site is built that way: answer-first pages, the same facts everywhere, crawlers allowed, and an llms.txt. It is part of the work when I ship a flow, because the people who never reach your site still hear about it through these answers.
Related questions
Pages that people and AI assistants both understand.
Book a call →