A 2024 analysis published in Frontiers in Pharmacology tested 202 CBD products and found that 74% missed their stated potency by at least 10%, with the errors running in both directions. Another 26% of the sample failed to meet the basic definition of the product type printed on the package. Those figures describe a market where the label is a weak predictor of what a dose will actually deliver.
That sample spanned 100 tinctures, 48 gummies, 34 vape products and 20 topicals. Potency error behaved differently in each format, and so did the contamination profile.
Contaminant And Potency Findings
The same analysis screened for contaminants and found heavy metals in 44 of the 202 products. Residual solvents appeared in 181 samples. Pesticide screening identified 26 distinct compounds across 30 products, which points to cultivation inputs as the origin.
Hemp draws metals and other soil residues into plant tissue efficiently, so cultivation inputs largely determine finished-product purity. Extraction adds a second exposure point. Solvent-based methods leave traces when the purge step is rushed or the equipment is poorly maintained, and those traces concentrate in the distillate that later becomes tincture or gummy stock.
Finished-goods panels catch problems that raw-material testing misses, because contamination can enter at any stage between harvest and fill. Brands testing only incoming biomass have verified the cheapest input in the chain while leaving the actual product unverified.
None of this is new. Earlier testing of 84 over-the-counter CBD products found only 31% labeled accurately within 10% of the advertised content, and a National Institutes of Health study split the failures almost evenly, with 26% of products holding less CBD than the label promised and 43% holding more. The category has had the same measurement problem for most of a decade, which is why regulators treat a current certificate as the minimum evidence rather than a courtesy.
Format changes the size of the error. Tinctures depend on even suspension in the carrier oil, so a batch that separates in storage delivers a different amount at the top of the bottle than at the bottom. Gummies depend on distribution through the slurry before it sets, which means one unit can miss the batch average by a wide margin even when the batch itself tests correctly. Testing a composite of 10 gummies hides exactly the variation the customer will encounter.
Potency error matters because the compound is active. A randomized trial funded by the FDA gave healthy adults twice-daily CBD for 28 days and recorded elevated hepatic aminotransferase levels in part of the group. Reported adverse events included hepatic enzyme increase in 11% of participants, eosinophilia in 9%, somnolence in 8%, and diarrhea in 8%. The elevations reversed when dosing stopped and no serious adverse events occurred, though the signal was strong enough to publish.
CBD also inhibits several cytochrome P450 enzymes, which changes how quickly the liver clears a long list of common prescriptions. Someone taking a blood thinner or an anticonvulsant alongside a product containing 40% more CBD than the label states is receiving an unplanned dose change the prescribing physician cannot detect.
Accurate labels are how safety reaches a customer. Anyone following the printed serving size is trusting a number that only laboratory work can support.
Federal Enforcement And Label Language
The FDA has issued more than 200 warning letters to cannabinoid companies since 2019. In April 2025 the agency’s Center for Veterinary Medicine wrote to Bailey’s Wellness, Holista, and House of Alchemy over unapproved cannabis-derived products sold for uses the agency had never evaluated. The letters cite marketing language about as often as they cite manufacturing practice.
The pairing is deliberate. A therapeutic claim converts a supplement into an unapproved drug under the Food, Drug, and Cosmetic Act, and once a product falls into that classification the manufacturing standard the agency applies changes with it. Companies avoiding disease claims keep themselves inside a regulatory category their quality systems can realistically satisfy.
The Second Audience For Test Results
Certificates of analysis travel further than the factory. Underwriters reviewing a new account ask for them alongside licenses and product catalogs, and the firms that handle cbd payment volume treat those documents as a measure of operating discipline. Orderly lab records mean the reviewer finds what they need without a follow-up request.
Retailers and insurers run similar checks before taking on exposure. Each wants evidence that the contents match the label. Each accepts the paperwork a working lab program already produces.
How To Build A Testing Program For Regulatory Review
Every production lot needs its own certificate of analysis, tied to a lot number printed on the package and issued by a laboratory accredited to ISO 17025. Composite testing across several lots hides the variation the test exists to expose.
The panel should cover cannabinoid potency, residual solvents, heavy metals, pesticides, microbials and mycotoxins. Retention samples from every lot allow re-testing if a complaint arrives 8 months after the sale. Publishing the certificates with a lot-number index lets a customer confirm the contents of the unit in their hand.
Nothing in the program works without qualified suppliers. Buying distillate on the open market means inheriting whatever cultivation practices produced it. An independent test the brand commissions is stronger evidence than a certificate the seller hands over. Qualified suppliers submit to periodic audits and accept contractual specifications with rejection rights attached.
Logged complaints matched to lot numbers let a brand pull the retention sample and trace a problem to its source within days. That speed determines how much inventory has to be destroyed and how many customers are affected before the problem stops.
The Cost Of A Failed Batch
A contaminated or mislabeled lot rarely stays contained. Product recalls begin with destroyed inventory and the laboratory work needed to establish how far the problem spread. The larger cost arrives afterward, when the acquiring bank reviews the account, the retailer suspends the item pending re-qualification, and the insurer reprices the product liability policy. Each of those parties reaches the same conclusion from the same document set, and none reverse it quickly.
The FDA has been explicit that it has never reviewed the manufacturing of unapproved CBD drug products, and that the effects of consuming a product carrying unsafe levels of pesticides or heavy metals are unknown. That sentence appears in agency material aimed at consumers. It also appears in the risk memoranda written by the institutions deciding how long to keep a CBD merchant on their books.
Full-panel testing on every lot costs most operations a few hundred dollars per batch. The second cost is what closes companies.