This page explains, in full, how PeptideProof computes evidence completeness and verification state. We publish this so the platform's conclusions are checkable, not a black box.
PeptideProof answers one question: can the evidence presented for a product or batch actually be verified? It does not diagnose, prescribe, recommend treatment or dosage, or declare a substance safe for human use.
The claim is corroborated by an independent source — typically the issuing laboratory's own verification record, or a database match with independently sourced accreditation evidence.
Some independent corroboration exists (e.g. the laboratory or vendor matches a known record), but not full corroboration of the specific value claimed.
The document itself states this value, with a named method where available, but no independent source has corroborated it yet.
A claim was made (e.g. a laboratory or vendor name) but PeptideProof could not match it to any known record.
Two sources disagree — for example, an external verification record reports a different value than the document, or the same report number appears differently on different pages.
This category does not apply to the report in question.
No evidence for this category was located in the document at all — distinct from a check that ran and failed.
A verification check was attempted but could not complete (e.g. a technical failure), distinct from a check that found nothing.
PeptideProof never collapses these into one another, because each proves a different thing:
Evidence completeness is a deterministic, rule-based percentage — never an AI-generated opinion. Each evidence category (identity, purity, quantity, contaminant panels, laboratory identity, report identity, batch and vendor association, and document integrity) has a fixed weight and a fixed point multiplier per verification state. The exact weights and multipliers are open in our source repository (packages/verification/src/rules.ts) and every report includes a "why this result?" breakdown showing the calculation for each category.
