The COA is the one label on the jar that cannot lie, and almost nobody is taught to read it. Here is the order I read one in, and the four things I look at that a shopper does not.
A COA is the lab report for a specific batch. Not the strain name, not the indica or sativa label; the batch. It reports three things. Cannabinoid potency: the percentages of THC, THCA, CBD, CBDA and the minors, usually with a calculated total THC and total CBD. The terpene profile: the dominant terpenes and their percentages, though not every COA includes this. Contaminant and safety testing: pesticides, residual solvents, heavy metals, microbials and mycotoxins, each pass or fail against a limit.
One arithmetic note before anything else, because that line confuses more people than the rest of the page combined. Total THC accounts for the acid form converting on heating, which is why it reads higher than the raw THC figure above it.
I have said "over four hundred terpenes" from a stage. Our own database says 432. I am correcting it publicly, because I would rather find our own error than have a chemist find it. No published source reports 432. It is almost certainly a transposition of 423, Turner and ElSohly's 1980 count of total plant constituents: every compound in the plant, sugars and proteins and all. A total-compound number became a terpene number, and stayed there.
The defensible answer is about 120 terpenes with confirmed structures (Radwan et al., 2021, the only review with a stated inclusion criterion), or about 230 if you count oxygenated terpenoids separately (Hanuš and Hod, 2020). It matters here because of what it implies about the paper in your hand: your lab is quantifying a dozen or so of those, chosen by a regulator. The COA is not the plant. It is the handful of compounds somebody decided were worth measuring.
A patient I had worked with for months came in frustrated. She had found a profile that worked for her, the kind of result that lets a person stop thinking about their medicine and start thinking about their life. Then it ran out, she bought the "same strain" from another grower, and she reported that it did almost nothing. She had not done anything wrong. Two plants can carry the identical strain name and be chemically different products, especially in their terpenes, the part most sensitive to how a plant is grown, dried and stored.
So classify by what is in the plant, its cannabinoid and terpene profile, rather than by its name or its lineage story (Hazekamp and Fischedick, 2012). A strain name is a marketing label anyone can print. A chemovar is a measured fact about a batch. Indica and sativa do not survive this either: the largest analysis to date, roughly ninety thousand samples, found the labels do not map onto chemical profiles (Smith et al., 2022). REFUTED
Which is why the practical habit is boring and works: when a batch suits you, photograph its COA and write the date beside it, in the free WIZDOM Journal or in a notebook, so that next time you compare profiles instead of names.
Every jurisdiction regulates differently and there is no single international standard, so do not memorise thresholds. There is no world format for this document. On any COA, in any country, you are looking for the same four things, in this order.
Before a single percentage. Which laboratory tested this, on what date, and does the batch or lot number on the paper match the one on the jar. A COA is a snapshot of the batch on the day it was tested, and months on a shelf are months of monoterpene loss that the paper cannot show you. A gorgeous terpene panel from fourteen months ago is a historical document.
Now the potency block. Read the ratio rather than hunting the single biggest number. In our own database, across 26,102 flower lab tests, total THC came in between 21.6% and 22.2% whatever effect people reported. The cannabinoid knob is stuck: potency is pinned by the market, not by biology. Chasing the highest THC on the shelf is choosing between numbers that barely differ.
Read the terpene total first, then the top three, then the thing almost nobody reads: the balance between the light monoterpenes and the heavier sesquiterpenes. A panel that is nearly all caryophyllene and humulene, with the citrus and pine bands thin, tells you about age or drying as much as about genetics. Then read ratios rather than ranks. Myrcene relative to terpinolene tells you more than either number alone. And a product that reports no terpene data at all is telling you it does not consider that part worth measuring.
Two honesty notes while you are here, because the marketing around terpenes is worse than the marketing around THC. β-caryophyllene is a full, selective CB2 agonist, and the effect vanishes in CB2-knockout mice (Gertsch, 2008) ESTABLISHED as a mechanism, while human efficacy is UNKNOWN. Myrcene as a sedative is REFUTED in humans, and α-pinene clearing THC fog is REFUTED in a 2025 double-blind trial co-authored by Ethan Russo, who proposed the hypothesis. Read the profile as a whole recipe, never as a hero ingredient.
Pesticides, residual solvents, heavy metals, microbials, mycotoxins. You want the word pass against each, and a complete panel rather than a partial one. Then the part that gets misread constantly: a blank, a dash, or "not detected" means the result came in below the reporting limit for that lab and that method. Below the reporting limit is not the same as zero. Say "below the reporting limit" and never say "contains none", because the second sentence is a claim the paper does not make.
One named caution belongs here. Pulegone is the terpene to be careful about: it is hepatotoxic, with human fatalities on record. Watch for it in concentrated terpene products in particular.
Here is the machinery I owe that patient. The chemotype class is locked and the potency is pinned, so the terpenes are the variable, and they are volatile in the literal sense. The light ones leave first. Air-drying alone shifted one flower from 33.7% monoterpene and 61.0% sesquiterpene to 25.5% and 71.6%, with terpinolene dropping 82% of its share while caryophyllene oxide rose.
The jar drifts peppery.
Storage does not simply shrink the profile, it rotates it. So when a patient tells me their flower "used to be brighter" and now feels "heavier", they may not be building tolerance. They may be smoking an old jar. That explanation is chemical, defensible, and sayable to a physician without embarrassment.
Which gives you an age marker to read directly: caryophyllene oxide, alongside CBN if the lab reports it. Do not confuse it with its parent. β-caryophyllene binds CB2 at 155 nM; the oxide binds nothing there and needs around 560 µM. One oxygen atom, four orders of magnitude, and the oxide is what rises as flower ages.
Print this, or keep it on your phone. It is the whole method in seven lines.
Then the question that turns a document into a decision: does this profile resemble the one that worked for me, or differ from it? That is the only comparison that has ever helped a patient of mine, and you cannot make it without the earlier COA.
This is close to what I say out loud, and coaches in training get it as a script. "Let's not assume you did anything wrong; the name on the jar is the least reliable part. Do you have the COA for the one that worked, and this one? We will compare profiles, not names: the ratio, the minors, the top terpenes, the test date. Then we set an expectation and track it, so next time you choose by chemistry, not by a label."
That would have saved my patient a wasted purchase and a week of self-doubt. Notice what she would be reading: never a single hero terpene, always the full profile. The whole recipe. Where that argument is thinner than people claim, I set out in the evidence for the entourage effect, and medical versus recreational cannabis is the other half of this.
A COA, or certificate of analysis, is the laboratory report for one specific batch of a cannabis product. It reports cannabinoid potency, often a terpene profile, and a safety panel covering pesticides, residual solvents, heavy metals, microbials and mycotoxins. It describes the batch in front of you, not the strain name and not the indica or sativa label.
The laboratory name, the batch number and the test date, before any percentage. A COA is a snapshot of that batch on the day it was tested, and months on a shelf are months of monoterpene loss that the paper cannot show you. Then read the cannabinoid ratio, then the terpene profile, then the safety panel.
No. A blank, a dash or "not detected" means the result came in below the reporting limit for that laboratory and that method. Below the reporting limit is not the same as zero, so read it as below the reporting limit and never as contains none. Every jurisdiction sets its own limits and its own required panels.
Because two plants can carry the identical strain name and be chemically different products, especially in their terpenes, which are the part most sensitive to how a plant is grown, dried and stored. A strain name is a marketing label anyone can print. A chemovar, the measured cannabinoid and terpene profile of a batch, is a measured fact, and that is what the COA gives you.
It can give you markers, not a verdict. A thin terpene total, a profile weighted towards the heavier sesquiterpenes, a raised caryophyllene oxide reading, and CBN where the laboratory reports it, all point towards age or hard drying. The test date matters as much as the numbers, because the paper describes the batch on the day it was sampled.
No. Every jurisdiction regulates differently and there is no single international standard, so the required tests, the action limits and the layout all vary. Do not memorise thresholds from one country's COA and apply them to another's. Read for the same things instead: the lab and date, the cannabinoid ratio, the dominant terpenes, and a clean contaminant panel.
Please read this part. This is patient education, not medical advice, and I am not a physician and not a prescriber. Nothing here is a recommendation to buy, avoid or change any product. Anything involving a lab value or a prescribed medicine is routed to your prescriber, and if a COA worries you, that is a conversation to have with your prescribing doctor and your pharmacist rather than with a shop assistant.
The free book gives you the map: the biphasic curve with its real numbers, the honest terpene count, and this COA method in full. The twelve-week Masterclass is where you practise it in front of people who will tell you the truth.
See the WIZDOM Masterclass