Reading a COA: What the Label Really Tells You

Every legal jar of flower ships with a receipt from a laboratory. Most people never read it. They read the sticker on the front — the one that says 28.4% THC in a font two sizes too large — and make their decision there. That sticker is a marketing surface. The Certificate of Analysis, the COA, is the closest thing the cannabis aisle has to a nutrition label, and learning to read it is the single highest-leverage skill a shopper can develop. It is also the fastest way to stop overpaying for a number.

Here's the honest framing up front: a COA doesn't make you a chemist, and it can't tell you how a cultivar will make you feel. What it can do is let you grade the grow yourself instead of taking the front label's word for it. Nothing here is medical advice.

What's actually on the sheet

A full COA is a stack of tests, usually run by an ISO-accredited third-party lab and tied to a specific batch or lot number. Match that batch number to the one on your jar — a COA for a different lot is decoration, not documentation. The core sections:

Cannabinoid potency. This is the panel everyone skips to: THC, THCA, CBD, CBDA, CBG, and the minor cannabinoids, usually listed as both a percentage and mg/g. Note that raw flower is mostly THCA — the acidic precursor that only converts to THC when you apply heat. The honest headline figure is total THC, calculated with a decarboxylation factor (roughly THCA × 0.877 + THC). If a label is quoting a giant number, check whether it's quoting total THC or quietly stacking THCA and THC in a way that flatters the jar.

Terpene profile. Present on better COAs, absent on lazy ones. This is the aromatic fingerprint — myrcene, limonene, caryophyllene, pinene, linalool, terpinolene and the rest — and it tells you far more about the actual experience than the potency panel does. More on reading it below.

The safety screens. Pesticides, residual solvents, heavy metals, mycotoxins, microbials, and moisture/water activity. These are pass/fail against your state's action limits, and they are the part of the COA that exists to protect you rather than sell to you. A clean product will show "ND" (not detected) or "Pass" down the column. This is the section worth your attention even when the potency number is boring.

Reading terpenes: the natural spread vs. the spike

This is where a COA earns its keep. A plant that grew well tends to produce a spread of terpenes — a dominant one or two, then a long tail of supporting compounds in smaller concentrations. Total terpenes in genuinely aromatic flower often land somewhere in the 1–3.5% range, distributed across a dozen or more entries. It looks messy because biology is messy.

Now picture the opposite: a profile where a single terpene towers over everything, the total is high, and the supporting cast is nearly flat. That pattern is at least worth a raised eyebrow. In an unregulated corner of the market, one documented mechanism for a lopsided profile is post-production spraying — reintroducing terpenes (sometimes botanically derived, sometimes not) onto flower to boost aroma or mask a thin natural expression. We're naming a mechanism, not a brand: a single-terpene spike isn't proof of anything, but a natural grow rarely produces one flavor note screaming over silence. When the nose and the numbers disagree — the jar smells like a candle but the COA is flat, or vice versa — trust the disagreement enough to ask questions.

Why total cannabinoids beat isolated THC

The THC% arms race is the central distortion of this industry, and the data has finally caught up to what discerning shoppers suspected. A 2025 study out of the University of Colorado Boulder, published in Scientific Reports, tested 277 products from 52 Colorado dispensaries. Among flower, 44% failed the state's own accuracy standard — the requirement that the product land within 15% of its labeled THC — and the misses skewed high: far more products overstated potency than understated it. One jar labeled 24% actually measured 16%. Tellingly, concentrates fared far better, with roughly 96% of labels accurate, because homogenous oil is simply easier to sample and test than a heterogeneous flower canopy.

So the label number is best understood as label-reported — a claim, not a measurement you can bank to the decimal. Which is exactly why the whole cannabinoid picture matters more than the single hero figure. Total cannabinoids — the full column of THC plus CBG, CBN, the minors — plus the terpene spread is a better predictor of a satisfying session than two-tenths of a percentage point of THC. Chasing the highest number on the shelf optimizes for the one metric most prone to inflation and least correlated with quality. That's a tradeoff, not a moral failing on anyone's part, but it's a bad trade.

The part the COA can't show you: lab-shopping

Here's the uncomfortable structural reality. The producer chooses and pays the lab. When labs compete for that business, one axis of competition is who reports the friendliest number — a dynamic the industry calls lab-shopping or THC inflation. It's a known enough problem that 2026 has brought a regulatory response: more states are requiring ISO 17025 accreditation, running "potency audits" that pull products off shelves for retesting at independent reference labs, and attaching real penalties — fines, license revocation — to gaming the numbers. An Ontario pilot in 2025 flagged a substantial share of products as inaccurately labeled. The correction is underway, but it's uneven and state-by-state.

You can't see lab-shopping on a single COA. What you can do is notice patterns: which brands publish full panels including terpenes and safety screens rather than a potency-only stub, which use accredited labs, which post batch-matched COAs you can actually pull up, and whose numbers stay plausible year over year instead of drifting suspiciously north. A brand that shows its whole sheet is making a different bet than one that shows you only the trophy.

Self-grading, in practice

Reading a COA well comes down to a short discipline. Confirm the batch number matches your jar. Read total THC, not the inflated composite, and hold it loosely. Read the whole cannabinoid and terpene spread as the real signal. Scan the safety screens for passes. And treat a lonely, towering single terpene the way you'd treat any too-good-to-be-true detail — with curiosity, not alarm.

The number on the front of the jar is the industry's opinion of itself. The COA is the closest thing to a second opinion you'll get on the shelf. Learn to read it, and you stop being graded — you start doing the grading.

Sources: CU Boulder — Can weed labels be trusted? (2025); Accuracy of labeled THC potency, *Scientific Reports* (2025); Cannabis Lab Testing Standards 2026: Potency Audits & ISO 17025; The Rising Threat: THC Inflation and Laboratory Shopping — Cannabis Science & Technology; "Uncomfortably high": Inflated THC potency on retail labels, *PLOS One*

Keep reading

• The THC% Lie
• Total Cannabinoids vs THC
• Organic: The Word Cannabis Legally Can't Use