Signal AI checks how ready your site is to be crawled, understood and cited by AI search and large-language-model systems - ChatGPT, Claude, Perplexity, Copilot, Google AI Overviews and Gemini. One score, four dimensions, in under a minute.
People used to scan ten blue links and pick one. Now they ask ChatGPT, Perplexity, Copilot, Gemini or Google's AI Overviews a question and read a single, synthesised answer.
That answer is built from the sources a model can reach, parse and trust. If your site isn't one of them, you're not lower down the list - you're not in the room. And unlike a search ranking you can watch slip, there's no page two to tell you you've been left out.
Type a query, scan a page of links, click through a few, judge each site for yourself.
Ask a question, get one answer with a handful of cited sources, act on it without clicking.
Signal AI reads your site the way an AI crawler does and scores it across four dimensions - the same four you'll see in your report.
robots.txt, a current sitemap, structured data, an llms.txt file. A model can't read what doesn't exist - coverage is the floor everything else stands on.
An open crawl path, valid schema, no accidental blocks. Small misconfigurations quietly lock the models out while everything looks fine to a human visitor.
Plain answers, clean headings, facts a model can lift without guessing. Vague, padded copy gets skipped in favour of a competitor who said it straight.
Entity strength, author and organisation signals, content that matches the questions people actually ask. This is what turns "crawlable" into "cited".
AI answers compound. The first time a model cites a source for a question, that source becomes the default - and every later answer leans on it.
Answer engines are becoming the first stop for research, not the last - so the decision often happens before anyone reaches your site.
Models cite their sources inline. Being quotable is now what being rankable used to be, and the two don't always overlap.
Get cited early and you bank an advantage that's hard to unseat. Leave it and you spend next year dislodging whoever the model learned to trust.
None of this needs a replatform. Crawl access, structured data, an llms.txt file and clearer content are all fixable now. The hard part is knowing exactly what to change - which is what this Signal tells you.