Why your COA is lying to you — and how to read one that isn’t.

In 2014, my co-founder and I spent nine months stabilizing a serum around a "98% pure" botanical active. Under independent testing, the drum held 61%. The COA was beautiful — letterhead, stamp, signature. The drum was lying. The product died; the lesson didn't, and it is the reason Rawess exists.

The uncomfortable truth is that most COAs are not forged. They are decorated — technically true, structurally useless. Over a decade of buying, testing, and now issuing certificates, I have learned to read a COA the way a radiologist reads a film: the story is in what is missing. Here are the five tells.

Tell one: suspiciously round numbers

A real assay is a measurement, and measurements are messy. 99.2%, 98.7%, 40.3% — these look like an instrument talking. When a COA reports 99.0%, 98.0%, 95.0% across every lot, every time, you are not looking at analysis. You are looking at a specification sheet wearing a certificate's clothes. Ask for lot-to-lot history; a supplier who cannot produce variance is a supplier who is not measuring.

Tell two: values without methods

An assay without a method reference is a number without a definition. "Purity: 98%" — by what? HPLC at which wavelength? GC with what calibration? Titration against which standard? We print the method next to the number on every COA we issue, because a value that cannot be reproduced by your lab is a claim, not a fact. If the method column is empty, the certificate is asking for trust it has not earned.

We bought the HPLC before we bought the office furniture. It was not a joke — it was a promise about what kind of supplier we intended to be.

Tell three: no batch traceability

A COA should bind a specific quantity of a specific batch to a specific analysis on a specific date, with the analyst named. If the document references a "typical" lot, or the batch number field is suspiciously generic, you are holding marketing. The phrase to fear is "typical values" — it converts a measurement into a mood.

Tell four: the missing panels

Watch what the certificate does not test. Water content unlisted for a hygroscopic powder. Peroxide value missing on an oil. Heavy metals absent for a botanical. Microbiological panel skipped "per supplier policy." Omission is the quietest form of dishonesty — the certificate is true, and the drum is still wrong. Our rule: if it matters in your formulation, it prints on our release, whether the monograph demands it or not.

Tell five: one page for a complex molecule

A peptide COA that fits on one page is a confession. A serious certificate for a synthesized molecule carries purity by HPLC, identity by mass spectrometry, counter-ion quantification, residual solvents, water content, and the truncated-sequence map. A botanical extract owes you the marker profile, not just the lead compound. Length is not rigor — but brevity for a complex molecule is always an economy.

How to read one that isn't lying

• Every value sits next to a method, and the method is one your lab can reproduce.
• Lot numbers, dates, analyst signoff, and a second-reviewer signature (four-eye release).
• Assays show realistic lot-to-lot variance — ask for three consecutive COAs and compare.
• The uncomfortable parameters are present: residuals, moisture, heavies, micro.
• The supplier offers raw chromatograms on request, not just the summary page.

That last bullet is the whole game. A supplier who will hand you the raw instrument output — the chromatogram, the ICP-MS trace — has nothing to hide behind letterhead. A supplier who guards it has a filing cabinet, not a quality system.

Send us any COA from any supplier, and we will read it with you. It is the friendliest audit you will ever run, and occasionally the most expensive truth you will ever buy. Either way, you will know what is in the drum.