Visibility Score is the share of probe runs, per topic and per engine, where your brand was recommended or cited. Paraphrase variants of one question are treated as correlated samples, and every score ships with a Wilson 95% confidence interval. A change is only labelled significant when the two intervals do not overlap.
Share of Voice is your recommended-share divided by the sum of every tracked entity's recommended-share on the same topic set.
Per-engine scores are never blended silently. The cross-engine number is Σ(engine score × weight) / Σ(weights of engines with data), using the weights below. Engines without a row count as weight 1; if the table is empty, all engines weigh equally.
No weights configured. All engines currently weigh equally.