How AI decides who you should hire.
Every week, we ask OpenAI, Anthropic, Google Gemini, and Perplexity the same
questions a real customer would type — "Who are the best roofing companies
in Dallas, Texas?" — across 100 U.S. markets, using each provider's
official API with active web search.
This is what they actually say, measured directly and published in full —
not simulated, not sampled from reviews. We measure. We don't optimize.
Why not just ask ChatGPT yourself?
A single question gives you a single answer, once — not a measurement.
We ask the same frozen set of questions to four AI models, every week,
across every market we track. That repetition is the only way to tell
a real shift in how AI recommends companies from one model's randomness
on one day. No one else runs this at this scale, on a fixed weekly
cadence, with every prompt and every response published in full.