READING A COA WITHOUT A CHEMISTRY DEGREE

Certificates of analysis are written for regulators, not customers. They're four pages of instrument output with the useful parts scattered across them. This is the order we check them in when a batch arrives, and the specific ways a report can be technically accurate and still tell you nothing.

The four checks, in order

What the safety panels are actually looking for

Potency gets the attention. The safety screens are what make the product safe to put in your body.

• Pesticides — typically 60+ analytes. Hemp is a bioaccumulator: it pulls contaminants out of soil and concentrates them. Outdoor hemp grown next to conventional agriculture is the classic failure.
• Heavy metals — lead, cadmium, arsenic, mercury. Same bioaccumulation problem, plus vape hardware can leach metal into oil over time.
• Residual solvents — butane, ethanol, hexane left over from extraction. Matters most on concentrates, distillate and carts; irrelevant for untouched flower.
• Microbials — E. coli, salmonella, aspergillus. Aspergillus is the one that matters for anything you inhale, and it's what took out one of our own batches.
• Mycotoxins — the poisons mould leaves behind even after the mould is gone. Improper curing or storage.
• Foreign matter — stems, insect parts, dirt. Unglamorous and worth checking.

Three ways a COA can be true and useless

1. The report is real, the batch isn't yours

A retailer links a single COA from every product page. The report is genuine, the lab is accredited, and it describes a batch from eight months ago that has nothing to do with the pouch in your hand. Perfectly true, entirely useless.

The test: is there a different report per batch, reachable by the number on your package? If one link serves the whole catalogue, that's a decorative COA.

2. Potency only, safety omitted

Potency testing is cheap and fast. The full safety panel costs several times more. A report showing only cannabinoids is a producer who paid for the number that sells and skipped the ones that protect you. It'll be labelled something like "Potency Analysis" rather than "Certificate of Analysis" — and that distinction is doing a lot of quiet work.

3. Selective disclosure

A four-page report published as one page. The page you get is potency; the pages you don't include a marginal pesticide result. Check the page count in the footer — Page 1 of 4 with only one page linked means three pages you haven't seen.

Related tell: a batch tested three times with only the best result published. Sample and report dates that are months apart are worth a question.

The metadata that tells you who tested it

• Lab name and ISO/IEC 17025 accreditation. Unaccredited labs produce reports that look identical and mean less. If the accreditation isn't stated, ask.
• Independence. If the lab shares an address or ownership with the producer, that isn't third-party testing. We name our labs — Kaycha, Green Scientific, ACS — precisely so you can check they aren't us.
• Method. HPLC for cannabinoids, GC-MS for terpenes and solvents. Mostly reassurance that a real instrument was involved.
• Sampled date vs reported date. A week apart is normal. Two months apart on a "fresh batch" is a question.
• LOQ. The limit of quantitation. "ND" means below this threshold, not zero. A report that states its LOQ is being honest about its own resolution.

Why a failed batch is a good sign

Counterintuitive, and it's the most useful heuristic on this page.

Every producer eventually gets a bad lot. Contaminated soil, a mould event during cure, a filling line that wasn't properly sanitised. It's a statistical certainty across enough harvests. So a catalogue where nothing has ever failed isn't a catalogue with better inputs — it's a catalogue that doesn't publish failures.

Two of the batches on our lab results page failed. One for myclobutanil at 0.14 ppm against a 0.10 limit; one for aspergillus traced to a contaminated filling line. Both were destroyed with a witnessed disposal record and both reports are still published. That costs us nothing except the appearance of perfection, and it tells you the process is real.

When you're evaluating any hemp company, ask to see a failure. The answer is informative either way.

The ninety-second version

1. Batch number on the package = batch number on the report. If not, stop.
2. Δ9-THC under 0.3%. If not, it isn't hemp.
3. THCA × 0.877 = actual strength. Decide if that's what you wanted.
4. Six panels, all PASS. Not four. Not "potency verified."
5. Lab named, accredited, not owned by the producer.
6. Page count matches what you were given.

If a product can't survive that in ninety seconds, the problem isn't that you lack a chemistry degree.

Reviewed 31 July 2026. Labs referenced: Kaycha Labs, Green Scientific Labs, ACS Laboratory — all ISO/IEC 17025:2017 accredited and independent of us. Not medical advice.