A certificate of analysis (COA) is the document that turns a laboratory’s work into a decision you can act on: release a batch, accept a shipment, trust a supplement. But a COA is only as useful as your ability to read it, and a professional-looking layout can hide gaps that matter. This field guide breaks down every section of a COA, explains what each result actually means, and lists the red flags that should make you ask questions.
What a COA is, and isn’t
A COA is a formal record stating that specific tests were run on a specific batch of material, by specific methods, and reports the results against defined specifications. It is a snapshot of one batch, not a blanket guarantee of every batch a supplier makes. Understanding that scope is the first step to reading one correctly.
The anatomy of a COA
Nearly every COA shares the same skeleton. Learn the skeleton and you can read any lab’s format.
Header and sample identity
The top of the document identifies the material: product name, batch or lot number, sample description, the date received, and the date tested. These must match the material in front of you exactly. A COA for lot A does not certify lot B, and a mismatch here invalidates everything below it.
The results table
The core of the COA is a table where each row pairs four things: the test, the method used, the result, and the specification it is judged against. A result with no method and no specification is just a number floating in space; you cannot tell whether it is good, bad, or even measured the way you assume.
Pass / fail and specification limits
Each result is compared to a limit, and the row is marked pass or fail (or “conforms”). The limit is where the judgement lives. “Lead: 0.5 ppm” means nothing until you know the limit is, say, “≤ 1 ppm”, at which point 0.5 ppm is a pass with margin to spare.
Method references: why they matter
The method column is the part inexperienced readers skip and experienced readers read first. A method reference (a USP chapter, an AOAC method, or a validated in-house SOP) tells you how the number was produced, and therefore what it can and cannot claim.
Two labs can report “purity” for the same sample and mean different things because they used different methods at different wavelengths. The method reference is what makes results comparable and defensible. If a COA reports results with no methods named, you cannot verify or reproduce anything on it.
Understanding “not detected,” LOD, and LOQ
“Not detected” is one of the most misunderstood entries on a COA. It does not mean zero: it means below the method’s detection limit.
- Limit of detection (LOD): the lowest amount the method can reliably tell apart from nothing.
- Limit of quantification (LOQ): the lowest amount it can reliably put a number on.
A result of “not detected” is only as meaningful as the LOD behind it. A sensitive method with a low LOD gives a strong “not detected”; a crude method with a high LOD can report “not detected” while the contaminant is present just below its reach. Always read the detection limits alongside any “none found” result.
Specification limits: where they come from
Specifications are not arbitrary. They come from pharmacopeial standards (such as USP), regulatory limits, or a company’s own validated internal specifications built from stability and safety data. A good COA makes clear which standard a limit is drawn from, so you know whether “pass” means “meets USP” or “meets the supplier’s own house limit”, which are not the same claim.
Accreditation and signatures
The bottom of the COA carries the lab’s authorization: the signature of a responsible analyst or quality manager, and, for an accredited laboratory, a reference to its ISO/IEC 17025 accreditation and scope.
ISO/IEC 17025 accreditation is not a logo. It is a defined scope. A lab is accredited for specific tests on specific matrices. A result outside that scope is not covered by the accreditation, even if the lab holds it.
Checking that the specific test on your COA falls within the lab’s accredited scope is the difference between an accredited result and an accredited letterhead.
Common red flags
- Results reported with no method named.
- “Not detected” with no LOD/LOQ stated.
- Specifications missing, so results cannot be judged pass or fail.
- Batch or lot number that does not match your material.
- An accreditation logo with no scope reference, or tests outside the accredited scope.
- A test date long before or after the batch date, suggesting the COA may not represent the material you have.
Reading a COA in practice
Put it together into a habit. Confirm the identity matches your batch. For each row, read the method and the specification before the result. Treat “not detected” as “below this stated limit.” Confirm the accreditation covers the tests that matter. When any of those pieces is missing, the right response is not to distrust the number; it is to ask the lab or supplier to supply the missing context before you rely on it.
