GEOforge · How we measure AI search visibility
Ask an AI the same question twice and you'll often get two different answers. So a single check isn't a measurement, it's an anecdote. Here's what it actually takes to produce a visibility number you can put in a board deck and track over time. Move the sliders; the maths is live.
The core idea
Think of a coin. Flip it 10 times and you might see 7 heads — pure luck. Flip it 1,000 times and you land near the true 50%. A brand appearing in an AI answer is the same kind of coin flip, and because it's usually a rare event, you need even more flips to pin it down. The slider shows the margin of error (the ± wiggle room around a visibility number) shrinking as you ask each question more times.
Going deeper
Every AI answer also names other brands. The more times we ask, the more distinct rivals we discover. This curve shows two things at once: the rivals that matter are all found early, but the complete list never fully stops growing. Slide to see how the picture fills in.
Two different questions hide here. "Have we found the core competitors?" Yes, quickly. "Have we found every brand the AI might ever mention?" No, and we never will. Depth gets you the first; the second is a rabbit hole. Pattern derived from aggregate measurement data across real cycles — illustrative, names removed.
What it means for you
We don't check your AI visibility once — we measure it ~1,500 times per source, every cycle, and report it with a confidence range so you know exactly how much to trust it.
A one-off check is off by roughly ±9 points — noise swamps the signal. At our depth the number is tight enough to see a genuine 1–2 point shift week to week.
The rivals that matter all surface within ~25 runs. We map the real competitive set instead of whatever one answer happened to name.
We show the uncertainty instead of hiding it. Every figure is reproducible from the underlying data — a single screenshot can't say that.
Illustrative demonstration of measurement methodology. Figures are generated live from the statistics of measurement (the Wilson confidence interval for appearance rates and a species-accumulation calculation for competitor discovery); the competitor-discovery pattern is an anonymised aggregate of real measurement cycles with all names removed. No specific brand or competitor is shown.
I lead the PPC and Analytics teams. My professional goals are to make clients and colleagues happy. I also love numbers, charts, and data-driven decisions.
