How to read a cosmetic COA (and what a fake one looks like)
A certificate of analysis is the only document in skincare that tells you what was tested, on which batch, and what the result was. It is also the easiest document in the industry to fake, recycle, or produce in a form nobody can verify.
If you're going to ask brands for proof — and you should — here's what you're actually looking at.
What a real COA contains
A genuine batch certificate has six things. If any are missing, it isn't a certificate, it's a specification sheet with a header.
- Product name and identity — the exact material, with an identifier (CAS number, INCI name, or internal code)
- Lot or batch number — the same number printed on the jar or its carton. This is the thread that ties paper to product
- Test parameters with specification limits — not just "pass," but "assay 0.10%, limit 0.05–0.15%"
- Results — the measured value next to each limit
- Methods — how it was tested (HPLC for assay, plate count for microbiology, ICP-MS for heavy metals)
- Authorised signature or stamp, and a date
A certificate without a lot number cannot be tied to anything you own. A certificate without methods can't be checked by a second lab.
The three tests that matter for a peptide cream
- Assay / identity — is the peptide present at the stated concentration, and is it the right peptide? For GHK-Cu this is usually HPLC, sometimes with mass spectrometry confirmation.
- Microbiology — total aerobic count, yeast and mould, and absence of specified pathogens. This is where a water-based cream fails if it's going to.
- Heavy metals — lead, arsenic, mercury, cadmium. Cheap to test, rarely shown, and required of the raw material in any serious supply chain.
Preservative efficacy is a separate document — a challenge test, run to a standard like ISO 11930. If a brand has one, they'll show you. If they don't have one, they'll change the subject.
The four ways COAs get faked
1. The recycled certificate. One analysis from 2021, attached to every batch since. Tell: the issue date predates the product, or the lot number doesn't match the jar.
2. The borrowed certificate. A third-party lab analysed someone else's material, and the report gets circulated as though it were the seller's. Tell: the sample submitter named on the report isn't the company selling to you.
3. The spec sheet dressed as a certificate. Results column reads "conforms" or "typical" instead of measured values. Tell: no actual number appears anywhere.
4. The unverifiable report. A PDF with no lab name, no accreditation reference, no signature — often with a decorative seal. Tell: search the lab's name. If the lab doesn't exist, the report doesn't either.
What "third-party" actually means
An in-house COA from a manufacturer is a self-report. It's not worthless — the batch discipline is real — but it is the manufacturer grading their own homework.
A third-party report comes from an independent lab. In China, look for a lab with CMA or CNAS accreditation; internationally, ISO/IEC 17025. If the report doesn't name the lab or its accreditation, it's an in-house document with better formatting.
For a small brand, the honest position is: in-house COA per batch, third-party testing on the raw material and on the finished product periodically. Claiming more than that is a lie you'll eventually have to defend.
How to ask
One message, three questions:
What is the Copper Tripeptide-1 content of the finished product, per batch? Can I see the COA for the batch I'm buying, with the lot number? Was it tested in-house or by a third-party lab?
If the answers arrive as a document with a lot number and measured values, you're dealing with someone who has their paperwork in order. If they arrive as reassurance, that's your answer too.
A note on this page
Certificates of analysis describe materials, not outcomes. A perfect COA does not mean a product works; it means the product contains what it says it contains, which is a lower and more useful bar. This article is not medical or regulatory advice. Cosmetic products described on this site are not intended to diagnose, treat, cure or prevent any disease.
More on copper peptides and what the evidence actually supports: the journal index. Or see CU29.