Buyer and Vendor Checklist

As a buyer: what to demand human data

Work down this list in order, because the early items make the later ones meaningful. First, a certificate of analysis that matches the batch or lot code physically present on the product in hand — not a certificate for the product line, not last quarter's certificate, and not an undated PDF. A certificate that cannot be tied to the specific material is a document about some other material. Second, the laboratory: accredited to ISO/IEC 17025, with the specific analytes and the specific matrix inside its declared scope of accreditation. Accreditation is scope-limited by design, so a laboratory accredited for cannabinoid potency in flower is not thereby accredited for residual solvents in a vape liquid, and the scope document is public. Third, the panel: on any converted or novel cannabinoid product, a potency-only certificate is not adequate, and the full set is cannabinoid profile, residual solvents, heavy metals, pesticides and mycotoxins, with microbiological testing where the matrix warrants it and a broader elemental scan where a catalyst may have been involved. Fourth, the things the report should disclose rather than hide: the unassigned chromatographic fraction, a total-cannabinoid mass balance, and the reporting limits for every analyte, so that "not detected" can be read against the method floor rather than taken as absence. Fifth, verification: where the laboratory publishes a sample-identifier lookup, verify the certificate with the laboratory rather than with the seller. Sixth, the label: a per-serving and a per-container milligram figure, stated on the product, plus the compound named by its actual chemical name with isomer and stereochemistry specified where they matter. A marketing name is not an identity.

• Batch-matched COA tied to the lot code on the product in hand.
• ISO/IEC 17025 accreditation with these analytes and this matrix inside the declared scope.
• Full panel on any converted or novel product — never potency-only.
• Unassigned-peak reporting, total-cannabinoid mass balance, and the reporting limit for every analyte.
• Verify with the laboratory by sample identifier where a lookup exists.
• Per-serving and per-container milligrams on the label, and the real chemical identity of the compound.
• Dose-uniformity evidence for anything sold as a unit dose.

Sources: International Organization for Standardization / International Electrotechnical Commission 2017 · United States Pharmacopeia 2023 · United States Pharmacopeia 2023 · Meehan-Atrash J 2022 · Vandrey R 2015

As a vendor: what to provide human data

Everything in the buyer list, plus the four things that only the producer can supply. Homogeneity verification: increments sampled from multiple points in each batch, assayed individually rather than composited, with the relative standard deviation and the range reported, and a written specification that a batch must meet before release. The pharmacopoeial model is worth adopting even where no regulator requires it — ten units, an acceptance value combining the deviation of the mean from target with the observed spread, and no individual unit outside 75 to 125 percent of the mean — because it is the framework that exists precisely to stop a correct batch average from standing in for a correct dose. Lot traceability: every finished unit traceable to a lot, every lot to its input material and its certificates, so that a complaint or an adverse report can be resolved to actual material rather than to a guess. Retained samples: a physical retain from each lot, stored under conditions that preserve it, for long enough to cover the shelf life plus a margin, because without a retain a later question about a lot is unanswerable forever. And honest labelling of what is not established: where a compound's human pharmacology and toxicology are not established, the label and the product information should say so plainly rather than borrowing the familiarity of cannabis. That last one is a commercial decision that feels costly and is not, because the alternative is a market where nobody can distinguish a careful producer from a careless one, and in that market the careless producer wins on price until an outbreak takes the whole category down.

• Homogeneity verification by multi-point sampling, individually assayed, with RSD and range reported against a written release specification.
• Lot traceability from finished unit back to input material and certificates.
• Retained samples from every lot, stored properly, for shelf life plus a margin.
• Honest labelling where pharmacology and toxicology are not established — say it rather than implying familiarity.
• Keep the certificates accessible to buyers by lot code, not on request-and-wait.

Sources: United States Pharmacopeia 2023 · International Organization for Standardization / International Electrotechnical Commission 2017 · Johnson-Arbor K 2023 · Vandrey R 2015 · Lin K 2026

Red flags, in one list human data

Each item below is a documented failure pattern rather than a matter of taste, and any one of them is sufficient reason to treat a product as unverified. No certificate of analysis at all. A certificate whose batch or lot code does not match the product. Potency-only testing on a converted or novel cannabinoid product, which leaves residual solvent, catalyst, metals and side-products entirely unaddressed. A novel cannabinoid named on a label that no validated quantitative method and no commercially available reference standard could have measured, which means the number beside it was not obtained in the ordinary analytical sense. Not-for-human-consumption labelling on a product that is plainly intended and merchandised for consumption — the exact device used throughout the K2 era to stay outside both food and drug regulation while selling something for inhalation, and its reappearance is a direct signal that the seller expects the product not to survive scrutiny. Label potency far from assay in either direction. An unaccredited laboratory, or an accredited one whose declared scope does not cover these analytes or this matrix. And no per-unit milligram figure anywhere, which means the consumer cannot compute a dose even if everything else is honest. Two further flags are worth adding from the market as it stands: a certificate supplied only as an image or a PDF from the seller with no laboratory-side verification path, and a product line that changes its headline compound faster than any toxicology could follow, which reproduces the structural driver that made the analogue era unmanageable.

Sources: Meehan-Atrash J 2022 · International Organization for Standardization / International Electrotechnical Commission 2017 · Bonn-Miller MO 2017 · Johnson E 2022 · Castaneto MS 2015 · European Monitoring Centre for Drugs 2021*

The argument the industry should be making human data

The difference between the legal hemp and cannabinoid market and the K2 era is not that the compounds are safer, because some of them are barely characterised, and it is not that the intentions are better, because intentions were never the variable. The difference is three concrete practices: analytical testing against authentic reference standards, honest per-unit labelling, and lot traceability with retained samples. Those three things are what make a dose computable, a contaminant findable, an outbreak traceable to a lot instead of to a rumour, and a careful producer distinguishable from a careless one. Which means they are not compliance overhead sitting on top of the product. They are the product. What a legitimate processor is actually selling, over and above the molecules, is the assurance that the contents are known, uniform and stated — and that assurance is the entire reason a customer should prefer a regulated supplier to an anonymous one. The strategic implication follows directly and is worth stating without diplomacy: any regulatory or market pressure that strips out testing, labelling or traceability recreates the failure mode. A ban that pushes a compound out of the tested market and into an untested one does not remove the compound; it removes the certificate. A price war won by skipping panels does not make products cheaper; it makes them unverified. A rule written against structures rather than against practices selects for whichever molecule is least characterised. The industry's interest and the public-health interest point the same direction here, and it is one of the few places in this field where that is true — which is exactly why the argument should be made in those terms rather than as a plea for lighter regulation.

Sources: European Monitoring Centre for Drugs 2021* · Castaneto MS 2015 · Banister SD 2018 · Meehan-Atrash J 2022 · International Organization for Standardization / International Electrotechnical Commission 2017