How AI search picks its sources
ChatGPT, Perplexity and Google’s AI answers don’t rank pages the way classic search does. They pull facts from sources they can read and trust, then write an answer in their own words.
If your site is slow to crawl, vague about what you sell or contradicted elsewhere, the answer comes from someone else: a directory, an old article or a competitor.
Schema that AI tools actually read
Structured data tells machines who you are, what you sell and what it costs. Organization, Service, FAQ and Person markup are the ones that matter for a B2B service business.
The rule is simple: schema must match what’s visible on the page. Markup that promises more than the page shows gets ignored, or worse, flagged.
One story about your company, everywhere
AI tools cross-check facts. If your site says one price, LinkedIn another and a directory a third, the answer hedges or picks the wrong one.
Part of the work is cleanup outside your site: profiles, directories and partner pages, so every source tells the same story.
Pages that answer the question
Buyers ask direct questions. A service page that opens with a slogan and hides the price three scrolls down gives AI tools nothing to quote.
I restructure key pages so the first paragraph answers the question plainly: what it is, who it’s for, what it costs and how long it takes.
How to check what AI says about you
Ask the tools yourself, logged out, with the questions your buyers use. Write down which sources each answer cites.
Do it again a month after any change. AI answers shift often, so the trend across many questions matters more than any single reply.