How to Read a Certificate of Analysis (COA): A Line-by-Line Buyer's Guide

Publish Date: 2026-10-13 11:45:51 Views: 86

A certificate of analysis is the single most-used and least-read document in supplement sourcing. It is also the only objective, lot-specific evidence you receive that a material meets what you paid for. A COA that is skimmed rather than read is a COA that protects no one.

What a Proper COA Contains

Before reading results, confirm the document is a document. A real COA identifies:

Supplier and material identity — trade name, botanical or chemical name, and plant part where relevant
Lot or batch number — the specific lot tested
Date of manufacture and date of analysis
Specification — the limits against which results are judged
Method — how each parameter was measured
Result and basis — the value, the units, and whether it is as-is or dry basis
Pass/fail or conformance statement
Authorised signatory

If any of these are missing, you do not have a COA — you have a claim. The testing obligations that make this document meaningful sit within the cGMP framework at 🔗[[21 CFR Part 111]].

The Lines That Matter, Interpreted

Identity

How the material is confirmed to be what it claims. Botanicals should show genus and species plus plant part; expect TLC, HPLC, or DNA-based identification. "Matches reference" without a method is not acceptable.

Assay / Potency

The headline number — but read it precisely:

Units: percentage, mg/g, or IU? They are not interchangeable on a spec.
Basis: as-is includes moisture; dry basis excludes it. A 5% assay as-is and a 5% assay dry basis can describe different amounts of active in the same bag.
What is being measured: total withanolides, total curcuminoids, elemental mineral — define the analyte, or the number is ambiguous.

This is where a zinc COA, for example, must state elemental zinc, not the weight of the zinc compound — a distinction covered in our zinc forms comparison.

Heavy Metals

Expect Pb, As, Cd, and Hg with values and a stated method — typically ICP-MS. Compare the value to your stated limit, not to a generic "pass." See our heavy metals and pesticide residue guide for the limits most buyers use.

Microbial Limits

Total aerobic count, yeast/mould, and absence of pathogens (e.g., Salmonella, E. coli). Pathogens should be "absent" with the sample size stated.

Loss on Drying / Moisture

High moisture shortens shelf life and changes the effective potency of everything else. For hygroscopic powders, this line predicts storage problems.

Ash (Total and Acid-Insoluble)

Total ash approximates mineral content; acid-insoluble ash approximates silica/earth contamination. Useful as a crude adulteration screen for botanicals.

Residual Solvents

Required for solvent-extracted materials — such as many botanical extracts. Absence or below-limit results here are non-negotiable for those products.

Pesticides and Mycotoxins

Not on every COA, but required for botanicals and at-risk materials. A multi-residue screen is the practical approach; see our contaminant guide above.

Physical Parameters

For powders: particle size, bulk density, tap density. These affect blend uniformity and capsule fill — a manufacturing concern, not just a quality one.

The Differences That Change the Number

Three quiet choices move a result without anyone lying:

As-is vs. dry basis — already covered; the single most common source of "why doesn't this match?"
Method — HPLC, UV, and titration can report different values for the same analyte. A result is only comparable to a spec if the method matches.
LOQ (limit of quantitation) — "not detected" is meaningless without the level at which it was not detected.

Always read the result and these three qualifiers together.

Red Flags

Red Flag	What It Usually Means
"Typical" values instead of lot-specific results	No real lot testing
No lot number or date	Not a COA at all
No method stated	Cannot be verified
No authorised signature	Not released by anyone accountable
Identical values across many lots	Template, not testing
"Conforms" with no numbers	No information
No LOQ on "not detected"	Unverifiable claim
Heavy metals without ICP-MS method	Lower-assurance testing

How to Verify a COA

Reading is not the same as verifying:

Reconcile against your specification — every line, every limit, every basis.
Request the chromatogram or raw data for disputed or critical parameters, especially standardised botanicals.
Confirm independence — a supplier's own COA is necessary but not sufficient; the model in our 🔗[[in-house vs third-party testing guide]] explains when to verify independently.
Retain the sample so a dispute can be re-tested.
Audit the process — the 🔗[[supplier audit checklist]] covers whether the COA is backed by a real quality system.

Specification vs. COA

The specification is what you require; the COA is what was measured. They are two documents that must agree. Write the specification first — with method, basis, and limit stated — then judge every COA against it. A COA cannot fill a gap you left in the spec.

COA Checklist

Lot number, manufacture date, analysis date present
Botanical/chemical name and plant part stated
Assay units and basis (as-is or dry) stated
Each contaminant limit matched to your spec with method named
Microbial pathogens "absent" with sample size
Residual solvents present for extracted materials
LOQ stated where "not detected" is reported
Authorised signature present
Chromatogram available on request

FAQ

What does "as-is" vs "dry basis" mean on a COA?

As-is includes moisture; dry basis excludes it. The same material can show different percentages depending on which is used, which is why the basis must be stated.

Why does my COA say "total heavy metals" instead of individual metals?

Colorimetric "total" tests are lower-assurance and not equivalent to ICP-MS individual-element results. For botanicals, require individual Pb, As, Cd, Hg by ICP-MS.

A result says "not detected" — is that enough?

Only with a stated limit of quantitation. "Not detected" without an LOQ tells you nothing about what was detectable.

Can I trust a supplier's own COA?

As evidence of intent, yes; as independent proof, no. Verify independently for identity and critical contaminants, especially for new suppliers.

What is the most common COA error?

Mismatched basis or method — a value that looks non-conforming (or conforming) only because it was measured differently from the spec.

Conclusion

A COA is decision-grade only when read line by line, with method, basis, and limit all considered. Specify first, reconcile every COA against that spec, watch for the red flags, and verify independently where it counts.

At JHD GLOBAL, every lot ships with a lot-specific COA — method, basis, and limits stated — plus chromatographic data and independent third-party verification on request.