Methodology
How PROOF Scores Products
PROOF turns permitted review and product data into a structured score. It does not personally test products, scrape unapproved sources, or let an AI invent a score.
A deterministic starting formula
30%
Overall review sentiment
20%
Reliability
20%
Performance
15%
Long-term ownership
15%
Value
Weights are stored as administrator-editable, versioned configuration. Live scores appear only when permitted aggregate evidence meets minimum coverage; otherwise PROOF shows Not Enough Data Yet.
What PROOF uses
Authorized APIs, affiliate feeds, licensed datasets, and other permitted review and product information may be imported for analysis. Review data remains subject to its permissions and source terms.
Structured analysis can identify recurring praise, recurring complaints, reliability language, long-term ownership experiences, value, performance, and ease of use. The final score remains formula-driven, not AI-generated.
Confidence reflects evidence coverage, source diversity and quality, recency, and category completeness. Affiliate relationships, retailer offers, and commissions are never inputs to PROOF scores. Scores may change as permitted data changes.