Supplement sellers increasingly show you a certificate of analysis, and it is a good development. A COA is a laboratory’s report on a specific batch: what was measured, by what method, and what came out.
It is also easy to imitate, because most buyers will not open it. Here is what a real one contains, what an imitation looks like, and the single question that separates them.
This matters most for oral products, where you are buying a manufactured good rather than a prescription — the NMN league table notes which brands publish one at all.
The one question
Can you find a number that could have come out differently?
That is the whole test. A measurement is a number with a result in it. A tick, a checkmark, the word “Pass” and a green circle are all assertions, and an assertion cannot be wrong in an interesting way. If every field on the page is a symbol rather than a value, nothing has been reported to you.
What a real one looks like
We opened one for an NMN product and it reads like a laboratory document, because it is one. It names the lot and the raw material batch. It gives production and report dates and a best-by date. Identity was confirmed by infrared spectroscopy at a 99% match against a library standard, with a 95% specification. The assay reports 283 mg per capsule against a 250 mg floor. Heavy metals — arsenic, cadmium, mercury, lead — were each measured by mass spectrometry and reported in parts per million. There is a named signatory and an accreditation number.
Every one of those is checkable. You could take the lot number and ask about it.
A second real one, and why the number looks odd
Another publishes an assay of 130.51 mg per capsule against a 125 mg specification. That is over the label claim, which looks wrong until you know why it happens.
Actives degrade over shelf life. Manufacturers commonly build in a small overage so the product still meets its claim at the end of that life rather than the start. A 4% overage is ordinary. More usefully, a figure like 130.51 is a strong sign the number came off an instrument — nobody inventing a result writes 130.51.
A suspiciously round number is not proof of anything. But a precise one is mildly reassuring, and it costs nothing to notice.
What an imitation looks like
We opened one that is presented as a certificate and is not. It is a row of tick marks against test names, with no measured values, no lot number and no date, above a line explaining that the real documentation is available on request.
Nothing about that is illegal and it may well sit in front of a perfectly good product. It is simply not evidence. The same page also contradicts its own label elsewhere — the FAQ describes the certificate as verifying 250 mg per capsule while the label states 250 mg per two-capsule serving. Those cannot both be right, and a document with numbers in it would have settled which.
Compounded injections work differently: there is no retail batch to publish, and the relevant assurances are about the pharmacy instead — 503A versus 503B covers what those are worth, and you can only ask a pharmacy for a certificate once you know which pharmacy it is.
Batch numbers are the part that matters
A certificate is about one batch. A seller who publishes a certificate but not which batch you received has given you a document about somebody else’s bottle.
The better arrangement, which a few sellers run, ties the batch number printed on your product to a dated certificate you can open, with an archive of previous batches. One of them makes the argument explicitly: testing a raw ingredient is not the same as testing the finished capsule. That is exactly right, and it is the distinction most “third-party tested” claims quietly skip.
What third-party tested does not mean
It does not mean the product works. A certificate reports what is in the bottle, not what the contents do — that is a question for trials, and the answers there are narrower than the marketing.
It also does not mean anybody checked the dose against a useful one. A capsule can contain exactly what the label claims and still contain far less than the amounts trials used. Purity and adequacy are different questions and a COA only answers the first.
A short checklist
- Is there a lot number, and does it match the one on your bottle?
- Is there a date?
- Is there a measured value, not just a tick?
- Is the specification shown next to the result?
- Is the laboratory named, ideally with an accreditation number?
- Does the certificate cover the finished product or a raw ingredient?
Four or more and you are reading a document. One or two and you are reading a picture. Which sellers publish which is recorded in the provider reviews.