Generative engines are non-deterministic. The whole category pretends otherwise. We treat the uncertainty as the problem to solve — and that honesty is the moat.
"Are we mentioned?" hides three different signals. We measure them apart, because each has a different lever: Answer Presence (are you named), Source Citation (is your own page the cited source), and Share of Voice (your standing against named rivals) — plus how favourably you're framed. The headline score is a transparent, weighted index you can take apart.
Every prompt runs many times, across engines and over time. Results carry a confidence interval, never a bare number. Before any work begins we measure the noise floor — so we can tell you the smallest lift we could honestly detect, and prove later movement is real rather than sampling drift.
Anyone can guess fifty prompts. We derive the real distribution your buyers use — from search data, sales calls and support tickets — classified by intent and weighted by volume. That per-sector corpus is proprietary, and it's what keeps the measurement representative.
Not "you're cited 12% of the time", but why. We log the pages each answer leans on and build a source graph per query cluster. Where a competitor appears and you don't is the gap — and the gap is your work plan. Measurement and delivery become the same thing.
We never claim deterministic ROI. We climb a ladder of evidence and report the strongest rung achievable for you:
A method that quantifies its own uncertainty turns AI visibility from a vibe into something auditable.
The audit applies all of the above to your firm and your competitors — before you commit to anything.
Request a free audit →