How to Actually Read a COA: LOD vs LOQ, CNAS Accreditation, and Red Flags to Watch For

A certificate of analysis (COA) is supposed to be the document that lets you trust a batch without personally testing it. In practice, a lot of procurement and QA staff sign off on COAs without actually being able to evaluate what the numbers mean — which defeats the purpose. Here's what the key terms actually say, and what to check before you treat a COA as verification rather than paperwork.

Why "Not Detected" Doesn't Mean "Zero"

A COA line that reads "Not Detected" for lead or arsenic looks reassuring, but it doesn't mean the contaminant is literally absent. It means the measured concentration fell below the method's limit of detection (LOD) — the lowest concentration the analytical method can reliably distinguish from background noise. Below the LOD, the analyte could be present anywhere from zero up to just under that threshold; the test simply can't tell you which.

Limit of quantification (LOQ) is a related but distinct number, and it's always higher than the LOD. LOD answers "is something there at all?" LOQ answers "how much, with a number I can actually trust?" A result between the LOD and LOQ is sometimes reported as "detected, not quantifiable" — present, but not reliably measurable at that level. If a COA collapses this distinction into a bare "ND" or "Pass" without stating the LOD/LOQ actually used, you're trusting a claim you have no way to evaluate.

It's also worth asking which analytical method produced the number. A heavy-metals test by ICP-MS typically reaches a much lower LOD than a cheaper colorimetric method — both can honestly report "Not Detected" while describing very different levels of sensitivity.

What CNAS Accreditation Actually Signifies

CNAS (China National Accreditation Service for Conformity Assessment) accredits testing and calibration laboratories against the international ISO/IEC 17025 standard for laboratory competence. A CNAS-accredited lab has had its equipment, method validation, staff competency, and internal quality system independently assessed, with periodic re-assessment to keep the accreditation current. Accreditation doesn't guarantee any single test result is correct — it means the lab's process for producing results has been vetted by a recognized third party, rather than taken purely on the lab's own word.

When a COA names a CNAS-accredited (or equivalent internationally recognized) lab and gives its accreditation number, you can independently confirm the lab's scope of accreditation actually covers the test in question. That verification step is exactly what a generic lab logo, with no accreditation reference, doesn't give you.

In-House COA vs. Independent Third-Party Report

A supplier's in-house COA reflects testing performed in the supplier's own lab, on the supplier's own equipment, by the supplier's own staff. That's not inherently unreliable — competent manufacturers run rigorous internal QC — but it is inherently self-reported, with no independent party confirming the numbers.

An independent third-party lab report comes from a separate accredited laboratory with no commercial stake in the outcome. For larger purchase decisions, or for any claim you plan to rely on downstream (label claims, regulatory filings, assurances to your own customers), a third-party report carries more weight precisely because the lab gains nothing from a favorable result. Many buyers reasonably request both: a batch-specific in-house COA for routine shipments, plus periodic independent third-party verification to spot-check consistency over time.

Red Flags in a Sketchy COA

• No batch or lot number. A COA that isn't tied to a specific batch can't be traced back to the physical material in front of you — it could be a template reused across shipments.
• No signature, seal, or lab identification. A legitimate report identifies who performed the test and typically carries a signature or official seal, not just a company logo.
• No reference to an accredited lab or test method. If the document doesn't name the testing lab, its accreditation status, or the analytical method used (HPLC, ICP-MS, a pharmacopeial method, etc.), there's nothing external to verify against.
• Results suspiciously centered exactly at the spec limit. Real analytical variation produces scatter — 99.6%, 98.9%, 99.2% across different batches. A COA where every batch reads exactly at the spec ceiling suggests the numbers were written to match the spec, not measured from the material.

What a Trustworthy COA Should Include

At minimum: a batch/lot number, test date, the specific analytical method used per parameter, LOD/LOQ where relevant, a named testing lab (in-house or third-party) with an accreditation reference where applicable, and a signature or seal. If any of these is missing, it's reasonable to ask for it before treating the document as verification rather than paperwork.

How PathGene Approaches This

Every batch we ship carries a batch-specific COA, and our outgoing quality control stage includes routine third-party laboratory verification before release, so the figures on the document are tied to the physical lot rather than a reused template. If your own QA team wants to arrange independent third-party testing on a received batch, that's a normal part of buyer diligence and something we support.