Two label systems (never bare “Evidence”)
Score Confidence (System A)
How confident we are in the quality and completeness of inputs used for the overall score — not how well the product “works.”
Score Confidence: HighScore Confidence: MediumScore Confidence: Low
Claim Evidence (System B)
Support for a specific claim on a page. Applied per claim, not to the overall score.
Claim Evidence: StrongClaim Evidence: ModerateClaim Evidence: LimitedClaim Evidence: Vendor-only
Missing data
Dimensions without enough data show Insufficient data (visible, not hidden) and use a conservative contribution cap so opacity is not rewarded. If too many dimensions are unscored, we withhold the total rather than invent one.
Conflicts of interest
Commission rates, cookie duration, and offer presence are structurally excluded from dimension scores. Affiliate relationships are disclosed on money pages and on theaffiliate disclosure page.
Updates
Reviews carry reviewed-on and updated-on dates. We re-audit on material price or formula changes, new evidence, or on a regular cadence — and we publish corrections rather than silently rewriting verdicts (corrections policy).