The people behind the measurements.

Buyers stopped starting at Google. They ask ChatGPT, Gemini or Claude which tool to buy, and the answer names two or three vendors. If you are not one of them, nothing tells you that you lost — there is no rank to slip, no impression to miss. You are simply not in the sentence.

Most tools that appeared to measure this sell a percentage. One number, one run, one day. But citations in AI answers swing 40–60% month to month — re-run the same prompt tomorrow and the number moves. A single point-in-time percentage is not a measurement. It is a coin flip with a decimal point.

So CitedOS is built the harder way round. Every prompt is sampled across multiple runs and multiple engines. Every metric ships with a Wilson 95% confidence range, the sample size behind it, and the date it was measured. When a change is too small to separate from noise, we say so instead of drawing an arrow. That costs more to run and it makes our charts less exciting — and it is the only version of this product worth buying.

The full method is public: how the numbers are produced.