Check who tested it, check that the batch code matches your package, then read the panel summary. Seven panels, all of them passing, on the batch you're actually holding. Everything else on the document is detail.
Reading the header: who tested what?
The top of the document answers three questions, and they're worth taking in order.
First, who ran the test. A licensed lab carries a state license number, and California publishes a register you can check it against. The finished-good documents behind our California product come from two licensed California laboratories, Infinite Chemical Analysis Labs in San Diego and Encore Labs in Pasadena, and either license can be looked up in a few seconds.
Second, what was tested. The header names the product, the matrix, which is the physical form the lab handled, and the sample identifier. Matrix matters more than it sounds: a concentrate and a flower product get different panel sets, because some tests only make sense on plant material.
Third, how much was tested. A batch size and a sample size appear together. One of our Live Rosin all-in-one documents records a batch of 861 units with 15 units sampled, collected on June 1, 2026. That ratio is the honest limit of what any certificate can tell you: a lab tested a sample and inferred the batch. Nobody tests every unit, and testing every unit would consume the product. Sampling is how the whole system works, and it's a reasonable compromise rather than a loophole, but it does mean a certificate describes a batch statistically rather than describing your specific cartridge.
There'll also be a track-and-trace tag, a long alphanumeric string in the state's METRC format. Every one of our documents carries batch and sample tags, and that's the thread tying paperwork to the physical package.
What are the seven panels?
Every batch of cannabis goods sold in California clears the same seven panels before sale. Here they are with what each one is actually looking for.
Worth saying plainly: the panel most people fixate on is potency, and it's the least useful one for judging quality. The four contaminant panels are where a document earns its keep, and they're the ones nobody photographs for a menu.
Cannabinoids
Potency, reported as totals and as individual compounds. You'll see Total THC and Total THCA listed separately, plus minor cannabinoids. This panel tells you strength and nothing about cleanliness. The two figures confuse people because they don't add up the way you'd expect. THCA loses mass when heat converts it to THC, so Total THC applies a conversion factor rather than a straight sum. On a concentrate you'll usually see results in milligrams per package as well as percentages, and the package figure is the one that describes what you actually bought.
Residual solvents
Traces of any solvent used in extraction, measured against action levels set per compound. This is where a solvent-based process gets verified, and it's the panel to read on any concentrate. Each solvent has its own action level, and they vary enormously: benzene sits at 1 microgram per gram while ethanol and acetone sit at 5000. That spread reflects how dangerous each compound is rather than how common. A result of 8 against a limit of 5000 and a result of 0.9 against a limit of 1 are very different situations, which is why reading the limit column matters as much as the amount.
Heavy metals
Lead, arsenic, cadmium and mercury. The plant takes up what's in its soil and water, so this panel is as much about where it was grown as how it was processed. Concentrating an extract concentrates anything the plant absorbed along with it, which is why this panel carries weight on oils even though the contamination happened in a field. A clean metals result on a concentrate is a statement about the farm as much as the lab.
Pesticides
Agricultural chemical residue, tested against a long list with individual action levels. The most consequential panel for flower and the one with the tightest limits. California tests against a long list of agricultural chemicals, each with its own action level, and those levels are set for a product that will be inhaled rather than eaten. This is the panel where a product most often fails, and failures here are the reason batch-level testing exists rather than annual spot checks.
Microbial impurities
Bacterial and fungal contamination, run by qPCR on most modern documents. A biological rather than chemical question, and one that matters more on plant material than on concentrates. The organisms tested for include specific pathogenic species rather than a general count, and modern documents run the analysis by qPCR, which detects genetic material. That's why a result can be positive for the presence of DNA from an organism that isn't viable, and why the thresholds are written the way they are.
Mycotoxins
Toxins produced by certain molds. Related to the microbial panel but reported separately, because the toxin can persist after the organism is gone. That persistence is the whole reason it's a separate line. A batch can clear the microbial panel because the mold is dead and still carry the toxin it produced while alive, so testing for one doesn't substitute for testing for the other.
Foreign material
Physical contamination, checked by visual inspection. The least glamorous line on the document and an entirely reasonable thing to test for. It covers hair, insect fragments, packaging debris and anything else that shouldn't have made it into a sealed product. Visual inspection sounds primitive next to mass spectrometry, and it catches a category of problem no instrument is looking for.
Where moisture and water activity show up
You'll see these on some documents and not others, and their absence isn't a gap.
Moisture content and water activity measure whether something can grow mold in storage, so California requires them on the product types where that risk exists. Which documents actually carry them comes down to the laboratory as well as the product. Across our own library of 502 certificates they appear on 312. Encore reports them on infused pre-rolls and leaves them off its cartridge and all-in-one documents, which is the rule working as written. Infinite Chemical reports them on every certificate it issues, cartridges and all-in-ones included.
This is worth knowing because a missing panel can look like an omission when it is a reporting convention. The question to ask is whether every panel that applies to this form of product appears.
Reading ND, LOD and action levels
Three pieces of notation account for most of the confusion, and none of them is complicated.
ND means not detected. It doesn't mean zero. It means the instrument looked and found nothing above its limit of detection, which is the smallest amount it can reliably see. A different lab with a more sensitive instrument might report a small number where this one reported ND, and both would be correct.
The action level is the legal threshold. Every analyte has one, and a result under it passes. Results are often reported in micrograms per gram for contaminants, and the gap between a detected trace and its action level is usually enormous. Seeing a number in that column isn't a warning sign on its own; seeing one near its limit would be.
Which leads to the point most brands avoid. Trace residual solvent appears on a great many certificates, including products made without any chemical solvent at all, because hardware, cleaning and handling contribute their own residue long after extraction. That's a statement about instrument sensitivity rather than about the product. Any brand promising a blanket zero is promising something the paperwork won't support. There's a related trap on the potency side. Two labs testing the same material can report figures a few percent apart, because instruments, preparation methods and calibration all differ slightly. Variance between batches is expected from an agricultural input, and it's why chasing the highest number on a menu is a poor way to choose between products that are otherwise similar.
The five-step read
Run these in order and you'll extract everything useful in under a minute.
One, check the lab and its license against the state register. Two, match the batch code on the document to the code on your package, because a certificate for a different batch tells you nothing about yours. Three, read the panel summary and confirm all seven say pass. Four, glance at the contaminant panels rather than the potency figure. Five, check the collection date, since a document from a year ago describes material from a year ago.
Step two is the one people skip and the one that matters most. It takes five seconds and it's the only step that connects the paperwork to the object. Everything before it establishes that a good document exists somewhere in the world. A brand can publish a genuine, clean certificate and still hand you a package from a different run. That isn't necessarily dishonest, and it's usually just a website that hasn't been updated. It still leaves you holding a document about material you don't have. Ask for the one that matches, and treat a shrug as information.
What can a certificate not tell you?
Quite a lot, and knowing the boundary keeps you from over-reading a good document.
It won't tell you about anything outside the panel set. A compliance certificate tests for a defined list, so a substance nobody thought to look for produces no line at all. Vitamin E acetate is the clearest example: it isn't a solvent, a pesticide or a metal, so a routine document has no place to report it. An absent line means nobody tested, and that's different from a clean result.
It won't tell you where the aroma came from. Terpenoid analysis is a test the licensee requests under 4 CCR §15725 rather than one of the panels every finished good must carry, so most batches don't have one. Across our own library it appears on 124 of 502 documents, concentrated on the Slow Cured and rosin lines, which are the batches where we commission a panel California does not require. For a typical cartridge the certificate is silent on whether the terpenes came from cannabis or from somewhere else.
It won't tell you how the product was made. Method lives in the brand's description, not the lab's report. A document can confirm that a solvent was removed and still say nothing about which one was used or why.
And it won't tell you whether you'll like it. Testing is a floor rather than a verdict on quality, which is the most common misreading of the whole exercise. A batch that passes every panel can still be dull, harsh or badly made, and a batch from a producer you love passes the same way a batch from a producer you've never heard of does. The document proves the floor was cleared. Everything above the floor is craft, and no lab reports on that.
