Marketing describes a product line; a certificate of analysis (CoA) describes one manufactured batch — the one whose lot number matches the package you are holding. We have covered why independent batch testing matters; this is the companion skill: reading the document itself, the way a quality auditor does, in about ninety seconds.
The anatomy: five columns that do all the work
Every legitimate CoA is a table with the same skeleton. Test/Parameter — what was examined (Vitamin D3 content; Lead; Total aerobic count; Peroxide value). Method — the exact procedure used, cited by standard: “USP /“, “Ph. Eur. 2.4.27”, “AOAC 2011.” This column is the credibility hinge: a named compendial or validated method is reproducible by any competent laboratory on earth; a missing method column converts results into unverifiable assertions. Specification — the pass/fail limit the batch must meet (“90.0–110.0% of label claim”; “≤0.5 ppm”; “≤5.0 mEq/kg”). Result — the measured value. Units — and unit literacy prevents the classic misreadings: “ Test panels are matrix-specific, and their composition itself signals competence. For a vitamin D3 softgel, the rational panel is: identity/potency assay by HPLC against the label claim; the four heavy metals — lead, arsenic, cadmium, mercury — by ICP-MS (the modern instrumental method behind USP / limits); microbiological safety (total counts, absence of E. coli and Salmonella); and, because the active rides in an oil carrier, an oxidation marker. For marine and other PUFA oils, oxidation is the headline: peroxide value (PV ≤5 mEq/kg for fresh oil) plus ideally anisidine value and the combined TOTOX score — rancid omega-3 fails its purpose regardless of potency. For botanical extracts, add identity by chromatographic fingerprint (the adulteration frontier) and, where relevant, pesticide screens. A CoA showing only “potency: passes” for an oil product is not reassuring — it is conspicuously incomplete. The document is only as strong as its author, and two checks take one minute. First, who tested: a manufacturer’s in-house CoA reflects self-assessment; an independent laboratory’s CoA carries the credibility of an organisation whose entire business is being right. Second, accreditation: the certificate should display an ISO/IEC 17025 accreditation mark with a certificate number, issued by a national accreditation body — and those bodies run public online registries where the number can be checked against the lab’s name and accredited scope in seconds. Scope matters: accreditation is granted test-by-test, so a laboratory accredited for microbiology is not thereby accredited for ICP-MS metals. Global networks (the Eurofins model) and accredited national laboratories both pass this check; anonymous “lab reports” without an accreditation trail do not. Fraudulent and cosmetic certificates recur in predictable patterns. No lot number, or a lot that doesn’t match the package — certifying, at best, somebody else’s batch. No method column. Suspiciously round results (“100.0%” potency, “0.00” for every metal) — real instruments produce scatter. Results exactly equal to specifications across the board. No laboratory identity — missing name, address, accreditation number, analyst signature or report date. “Meets all specifications” summaries with no numbers at all. And the structural tell: a certificate that cannot be retrieved independently — quality-committed brands make batch CoAs downloadable by lot number, which is precisely the model behind lot-verification systems, because a document you can only obtain from the seller’s hand is a document the seller controls. Lot number matches package → laboratory named, independent, ISO 17025-verified in the registry → method column populated with compendial references → specifications printed beside every result → panel appropriate to the matrix (metals by ICP-MS always; PV for oils) → results plausibly scattered within limits. Six checks, no chemistry degree required — and any product that fails them has answered your question by other means. A real CoA is falsifiable: named batch, named lab, named methods, numeric results against printed limits, verifiable accreditation. Treat the document as the product’s laboratory passport — and treat its absence, for a daily-ingested product, as data. Was this article helpful? Disclaimer. This article is health journalism intended for general information and education. It is not medical advice and is not a substitute for professional diagnosis or treatment. Always consult a qualified healthcare provider about your individual circumstances. Full disclaimer →Which tests belong on which product
Verifying the laboratory itself
Red flags: the decorative CoA
The ninety-second routine
The clinical bottom line
Primary sources
Related Coverage
Editorial standards. This article was produced under the GMJ News editorial process, with oversight by the GMJ Editorial Board. Our editorial process. Spotted an error? Contact the editorial team.
