A/B/C Classification
A measurement-event classification by how reliably a mention recurs across samples.
What is A/B/C classification?
A/B/C classification is a measurement-event scheme by stability across the N samples: A events are stable across all samples, B events appear in a majority of samples, and C events appear in a minority — typically noise.
A/B/C classification separates signal from luck: it tells you whether a brand’s appearance is a reliable fact or a one-off fluke.
Because each query is asked N=5 times (N-sampling), every detected mention can be graded by how often it recurred. An A event showed up in all samples — a stable, trustworthy presence. A B event showed up in a majority — present but not certain. A C event showed up in only a minority — usually noise from the engine’s randomness.
This grading is what lets the platform act on facts rather than artefacts. A brand counted once across five samples (a C event) should not drive a recommendation the way one counted in all five (an A event) does. The classification feeds the insight-extraction pipeline so the loop reasons on stable signals.
A/B/C is built directly on N-sampling and the variance it exposes, and it underpins the reliability layer of the SOV pyramid. Without it, a single lucky mention could masquerade as genuine visibility and send the closed loop chasing noise.