Evidence Methodology

Each run records prompt text, provider, surface, locale, region, repetition, collector version, time, and raw evidence location. Production requires at least two configured providers; missing evidence is never replaced by a fixture.

Priority score combines business value, competitor gap, citation recurrence, source relevance, actionability, evidence quality, and freshness. Confidence is reported separately and falls when samples or providers are insufficient.

Citation Score is the single headline number shown on scan results and the workspace Today page. It is a confidence-weighted average of the Opportunity Scores in view: each opportunity contributes its score multiplied by a weight of 1.0 (high confidence), 0.6 (medium), or 0.3 (low), divided by the total weight, rounded to a whole number from 0 to 100.

Higher means more recoverable recommendation opportunity was observed — not that the brand is doing well. The inputs stay visible beside it: every contributing Opportunity Score, its Evidence Confidence, the repeated sample count, and the exact prompts, providers, locales, and regions behind them. When evidence is thin, confidence drops and the contribution shrinks instead of the score silently holding its value.

An Action Pack must reference stored response evidence. After a customer marks work complete and provides the target URL, rechecks repeat the saved project prompts and compare current gap score and target citations with the stored baseline.

Outcomes are Won, Improved, No change, or Insufficient evidence. They describe observed change and do not claim that the Action Pack caused the change.