A Certificate of Analysis is only worth what its weakest field is worth. Nine specific defects turn a report from evidence into decoration: no batch number, no lab address, no signature, a boilerplate panel reused across unrelated brands, missing analytes, no detection limits, no dates, unaccredited methods, and no sample identity. Any one of them is a reason to ask a question before you buy.
This is the follow-up to our walkthrough on how to read a COA section by section. That post covers what each part of the document means. This one covers what a bad document looks like — including examples pulled from supplier certificates for products we carry, because the point of publishing lab paperwork is that you get to check our work too.
The nine red flags at a glance
| # | Red flag | Why it matters |
|---|---|---|
| 1 | No batch or lot number | The report cannot be tied to the package in your hand |
| 2 | No laboratory address | An unlocatable lab cannot be checked or held responsible |
| 3 | No signature or approving analyst | Nobody has put their name to the result |
| 4 | Boilerplate panel reused across brands | The analyte list was chosen by convenience, not by product |
| 5 | Missing analytes | Silence on a compound is not clearance of that compound |
| 6 | No LOD or LOQ | "Not detected" has no floor, so it has no meaning |
| 7 | No dates | An undated result could describe any batch, ever |
| 8 | Unaccredited methods | The result sits outside the lab's assessed scope |
| 9 | No sample identity | Nothing establishes what material was actually tested |
1. There is no batch or lot number
A COA describes one lot of material produced at one time. Without a batch identifier, the document describes nothing you can hold. It becomes a general statement about a brand, which is precisely what a certificate is not for.
Check the batch on the report against the batch printed on your package, and check that they are identical rather than similar. We have read certificates where the lot field simply says "N/A" — a brand telling you, in the space reserved for the most important field on the page, that it did not record which material it sent to the lab.
This flag outranks the other eight. A perfect report that cannot be matched to your product is a report about someone else's product.
2. The laboratory has a name but no address
A real testing facility occupies a real building. Accredited reports carry a street address, usually a phone number, and a certificate number issued by an accrediting body. A logo with nothing behind it is unverifiable by design.
The check takes thirty seconds: search the lab name together with the address. You should find a facility, a published scope of accreditation, and a business that exists independently of the brand paying for the test. If the only trace of that laboratory anywhere is the PDF you are currently reading, you have your answer.
3. Nobody signed it
Accredited laboratories release results under the signature of a named lab director or approving analyst, usually as a digital signature with a release timestamp. That signature is the accountability mechanism. An unsigned results table is a spreadsheet.
The related failure is more common: the cropped screenshot. A results table posted as an image with the header and footer removed has had every accountable field stripped out while keeping the reassuring numbers. Ask for the complete PDF. A brand that has one will send it without friction.
4. The same panel appears across unrelated brands
This is the flag almost nobody checks, and it is the one that taught us the most about our own shelf.
Two certificates covering different brands we carry use an alkaloid panel that is identical byte for byte: the same seventeen analytes in the same order, with the same limits of detection to two decimal places. That list includes psilocybin, psilocin, baeocystin, norbaeocystin, aeruginascin, muscimol and ibotenic acid — compounds associated with mushrooms, not with Uncaria tomentosa. It is a general-purpose screening group the laboratory runs for many clients, not a panel built around what these particular products are.
A reused panel is not fraud. Labs sell standard method groups and brands buy them; it is how the industry works. But it tells you the analyte list was chosen from the lab's catalogue rather than from the product's actual risk profile, and that is exactly how the next flag happens.
5. The compounds you care about are not on the list
A non-detect result covers only the compounds the method looked for. Anything absent from the analyte list was not cleared, not measured, and not addressed.
We will be specific about the paperwork we hold. The supplier certificates we hold for products we carry report non-detect for 7-hydroxymitragynine, mitragynine, speciociliatine, speciogynine, paynantheine, corynantheidine and mitraphylline. Those same certificates do not include mitragynine pseudoindoxyl, MGM-15 or MGM-16 as analytes at all — three compounds named in the temporary scheduling action the DEA filed in July 2026, which we cover in is 7-OH banned. Corynoxine B is likewise absent from the panel.
So we say what the panel supports — no kratom, no mitragynine, no 7-OH — and we do not say the products were cleared of compounds nobody looked for. Where we hold no certificate for a brand at all, we say that too rather than letting one brand's paperwork stand in for another's. Any seller telling you a product is "free of all scheduled compounds" on the strength of a panel like this is describing a document they have not read to the bottom.
6. There are no detection limits
"ND" means the analyte was not found at or above the method's limit of detection. Without the LOD printed beside it, you cannot tell whether the floor was 0.01 mg/g or 10 mg/g, and the result is uninterpretable.
Detection limits also deserve arithmetic rather than alarm. One panel we reviewed lists a 7-OH detection limit of 120.50 ppm, which reads as enormous until you convert it: 0.1205 mg per gram, so on a roughly three-gram chew the largest quantity that could sit beneath the floor is about 0.36 mg — around a hundredth of a percent of the unit's mass. Our objection to that certificate is the analyte coverage described above, not its sensitivity. That distinction is worth getting right, and the same unit-conversion habit is what makes published regulatory thresholds readable — we work through those in the 7-OH threshold explained.
7. The report has no dates
A complete COA carries three: date received, date analysed, date reported. Undated results float free of any production run, which makes them permanently reusable and permanently unfalsifiable.
Compare the analysis date against the manufacture or expiry date on your package. A certificate issued two years before a product was made is not evidence about that product. Botanical raw material varies between harvests and lots, which is the entire reason batch testing exists. One report cannot cover a year of production.
8. The method sits outside the lab's accredited scope
Accreditation is granted per method, not per laboratory. A lab can genuinely hold ISO/IEC 17025 accreditation and still run methods outside the scope an assessor reviewed. Honest labs disclose this in a footnote, in small type, at the bottom of the page.
One certificate we read while evaluating a prospective brand carried an asterisk on exactly the analytes we most wanted to see, with a footnote explaining that the method had been validated internally by the lab but not by the accrediting body, and was therefore not an accredited work item. Every analyte on that method group carried the mark. The result may well be perfectly sound — but it is not accredited, and a retailer repeating it as accredited is overstating the evidence. Read the footnotes; this always lives in the footnotes.
9. Nothing establishes what was tested
Sample identity is the chain connecting a jar on a lab bench to a package on a shelf: the lab's internal sample ID, the product name as submitted, the format, the strength, and ideally the measured unit weight.
Two failures show up here. The first is a report with no identity testing, so nothing confirms the material was the labelled species — which matters for a plant whose common name is shared by several unrelated species. The second is a product name that does not match your package. A report for the 50 mg tablet is not a report for the 100 mg tablet, and a report for one flavour is not a report for another when the flavours were produced as separate lots.
What we do with the flags we find
We publish the certificates we receive on our COA page, gaps included, and we describe the gaps instead of rounding them away. We would rather tell you that a panel omits four compounds than let a "lab tested" badge imply a completeness the document does not have. The same standard decides which brands we agree to carry, and the buyer-side version of it is set out in how to buy botanical products safely online.
Most of this category says "lab tested" and publishes nothing. Reading nine fields puts you ahead of almost every buyer in it.
Frequently Asked Questions
Is a COA with one red flag automatically worthless?
No — severity varies. A missing sample ID is a documentation weakness. A missing batch number means the report cannot be connected to your product at all. Flags 1, 5 and 6 are the ones that make a certificate unusable as evidence. The others are reasons to ask the seller a direct question and then judge the answer you get.
Can a certificate be forged?
Yes, and altered PDFs are far more common than wholly invented ones. Look for mismatched fonts inside a single table, misaligned rows, inconsistent decimal places, or a header that does not sit square with the body. Where a laboratory offers direct verification of a report number, use it.
Why would a mushroom panel appear on a Cat's Claw product?
Because it is a standard method group the laboratory already runs. The analytes were bundled by the lab, not selected for the product. It is not evidence that anything is in the product; it is evidence that the panel was not designed around the product.
What should I ask a seller who publishes no certificates?
Ask for the full PDF for your specific batch, with the lab name, address, accreditation number, analyte list and detection limits intact. The response tells you what you need to know. "COA available on request" that never actually arrives is itself a complete answer.
Where can I see the products these certificates cover?
The lineup sits in the Cat's Claw collection, and each listing links to the certificate we hold for it. If you cannot match a report to the package in front of you, send us the batch number and we will find it.
These statements have not been evaluated by the U.S. Food and Drug Administration. These products are not intended to diagnose, treat, cure, or prevent any disease. For adults 21+. Legality depends on the specific formulation and applicable state law — check the rules where you live. Nothing here is legal advice.