Markets measured
20
When someone asks ChatGPT, Claude, or Gemini "who should I call?", a handful of businesses get named and everyone else is invisible. We built the instrument that measures it, we publish the results, and we fix the two things that usually cause it: a site the models can't read, or no site at all.
A single measurement of a place and a trade produces a public ranking — and tells us, for every business in it, which of three problems they have. The evidence is the same either way: their row on a public page, what their robots.txt blocks, whether they have a website at all.
People are being pointed at them and landing on nothing. A 3–5 page site in about 72 hours — $490 for the year or $59/month, domain and content theirs to keep.
They have a site, but the assistants don't recommend them. The audit shows which models name them, who's ranked above them, and the technical reasons why — $249.
They're already recommended first. Positions move as models re-index, so we track the market monthly and send the changes — $99/month.
A public record of what frontier AI models believe — a fixed canon of questions asked weekly, tracking consensus, disagreement, and drift across model releases.
Enter the observatory →Public leaderboards of which local businesses AI assistants actually recommend, per place and trade — plus the full audit report for any business on them.
Browse the Index →For businesses with no working website. A simple, fast, mobile-first site in about 72 hours — convenience, not bespoke design, and priced like it. You own the domain.
See what's included →Cryptographic proof that a backup happened and is intact — independently verifiable, so you don't have to trust the vendor. A separate brand and site, same owner.
Visit dokaz.net ↗