Skip to content

COA evidence scoring methodology

HOW THE DRAFT SCORE WORKS

Evidence, with the context intact.

The score describes the evidence in a report. It is not a percentage purity, probability of authenticity or product safety rating. These are our draft weights, pending real-report calibration.

30 points

Issuer verification

Does the issuing lab’s record match this report?

20 points

Batch and sample traceability

Is the tested sample linked to the claimed product and batch?

25 points

Analytical coverage

Which identity, purity, quantity and method checks are documented?

15 points

Result interpretation

Are units, specifications and measurement limits clear?

10 points

Document completeness

Is the report complete, legible, dated and authorized?

Critical issuer, batch or identity mismatches remain visible and require review, even when a provisional score is displayed. Unknown evidence is not a failed laboratory test. Sterility, endotoxins and contaminants require their own reported methods; a purity result does not establish them.

Research references and limitations

Automatic scores describe documentation only and cannot exceed 70/100 while issuer verification is unchecked. Scanned pages use local OCR. A provisional number may appear before confirmation when OCR confidence is at least 80; that threshold is an implementation guard, not a validated accuracy guarantee. Confirm the extracted values against the original before relying on them. Choose powder, vial or solution format. Unresolved entries earn no points; the provisional number remains visible, with uncertainty flags. AI can assist with difficult pages, but does not verify the report. The PDF reader does not interpret signatures or prove chromatogram validity. Lab verification remains manual. Scores shown in examples use synthetic assertions only. A report cannot prove a supplied vial matches the tested sample.