How PRAGMA measures AI visibility.

What we measure.

One thing, precisely: when a buyer asks an AI engine who's best, does it name you? Not whether a keyword appears. Whether the engine recommends your business, by name, in the answer a real customer would act on.

The method.

• The engines. We query ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews — the surfaces where buying decisions now start — plus smaller on-device models.
• The questions. The real questions your buyers ask in your category and market ("who's the best {what you do} in {city}?"), not questions gamed to flatter you.
• The record. Per engine, per question: is your business named, and who else is. Named — not merely mentioned.
• The pillars. We separate answer-engine visibility, parametric (model-memory) visibility, and local presence — because one blended grade hides which lever is actually failing.
• The baseline. We freeze a Day-0 reading before we touch anything, so any change afterward is provable, not asserted.
• The reproducibility. Same inputs, same result. The score is just the disclosed tests applied — you (or a skeptic) can re-run it and land in the same place.
• The registry. A verified result lives at a public, dated, registry-style URL — not a screenshot in a deck.

What we don't claim.

We report what the AI engines actually say. We do not claim the score predicts your revenue. Whether visibility converts to inquiries is something we track honestly and separately — never something we assert to close a sale. A measurement that overpromises is the vanity metric we exist to replace.

What we track separately — the real inquiries.

There are two honest things to measure, and they sit on two levels. Against your market, we rank you on what the AI engines say — a public test anyone can re-run, on you or on your competitors. On your own site, we track the real inquiries those pages earn: the phone calls, the email taps, the contact forms — tied back to the page, and even to the AI engine, that sent them.

That second number is the one that matters most — a single call is worth more than a thousand views — but it is yours: nobody can see a competitor's private calls, and neither can we. So we rank you against your market on what's public, and on your own site we show you the real inquiries as they arrive, plainly, in data you own. What we never do is claim the score itself put the phone in someone's hand. We show you both, in the open, and let you judge the connection.

What we don't disclose.

We open the ruler — the tests and how the number is produced. We don't open the recipe — the specific craft of engineering a page so the engines start naming you. The measurement is transparent so you can trust it. The optimization is the work you hire. Full disclosure of the standard; no leash on the method.