Which AI is your site built for?
ChatGPT, Claude, Gemini and Perplexity do not cite the same kind of page. Type an address: we read the site, write the questions its buyers ask an AI assistant, and judge every page against all of them while you watch. A couple of thousand judgments, a few seconds.
What this is, and what it isn’t
It’s a real run
Every page is fetched and judged as you watch. The seconds and the cost on the screen are the ones it took. There is no recording and no sample data.
It’s measured, not guessed
The engines’ tastes come from 1,920 pages they cited for 574 real buyer questions. ChatGPT goes to the vendor’s own page, Claude wants a straight answer, Gemini loves this year’s ranked list, Perplexity reads the comparisons.
It’s not a promise
A match says a page looks like what an engine cites. Whether the engine finds it, and trusts your site, is another matter. To see what the engines actually say about you, run a scan.
Questions
What does the live check do?+
You type a website address. CiteGraph reads the site's pages, writes five questions its buyers would ask an AI assistant, and asks 37 questions about every page: is it about one product, does it compare products, does it answer the buyer's question in its opening, and so on. A small model per engine turns those answers into a match with the kind of page ChatGPT, Claude, Gemini and Perplexity each cite. All of it happens while you watch, and the timings and cost on screen are the real ones.
Do ChatGPT, Claude, Gemini and Perplexity really cite different kinds of pages?+
In our data, yes. We looked at 1,920 pages the four engines cited for 574 real buyer questions, across 1,380 websites. 91% of the pages ChatGPT cited were about a single product, against 36% of the pages only the others cited. Gemini cited ranked lists 54% of the time, against 15%. Claude's pages gave step by step instructions 46% of the time, against 21%. Perplexity's compared named products 52% of the time, against 36%. Every one of those gaps holds up when whole websites are resampled.
Does a high match mean my page will be cited?+
No. A match says how much a page looks like what that engine cites. Whether an engine finds the page at all, and how much it trusts the site, are not in it. The scorers were tested on websites they never saw (ChatGPT 0.88, Claude 0.77, Gemini 0.77, Perplexity 0.65, where a coin scores 0.50). Perplexity's is the weakest because its taste overlaps with Gemini's. They are a useful guide to what to write, not a promise.
What is Jev?+
Jev is a model from TypeSafe that makes typed judgments instead of writing text: a yes or no with a probability, a choice from a list, a score on a described scale. It answers a few dozen questions about a page in well under a second for a fraction of a cent, which is what makes it possible to judge a whole site while you watch. CiteGraph uses it for judging only. The engines we measure, and everything that is written, are other models.
Is it free?+
Yes, with no account. Every run reads and judges a real site, so there is a daily limit per connection and a daily cap overall. What the engines actually say about you, question by question with the pages they cite, is the paid product.