How to read a botanical extract COA without being fooled

1. The header tells you whether the document is even about your batch

Before reading a single result, check that the COA identifies the product, the botanical species, the plant part, the internal lot number, the manufacturing date, the expiry or retest date, and the specification it was tested against. A COA without a lot number cannot be tied to the material in the drum, and a COA without a specification reference has no pass/fail basis.

The botanical name matters more than the trade name. "Ginseng extract" is not a specification; Panax ginseng root extract standardized to ginsenosides is. The same applies to plant part: root, leaf, aerial parts, and fruit are different ingredients with different marker profiles.

• Product name, botanical name, and plant part
• Manufacturer name and manufacturing site
• Batch or lot number that matches the drum label
• Manufacturing date, expiry or retest date
• Specification number and issue date
• Test methods named next to each result

2. Identity is not the same as assay

Identity answers "is this the right plant?" Assay answers "how much of the marker is present?" A COA can show a perfect 95% assay and still describe the wrong species, the wrong plant part, or a material cut with a cheaper source of the same marker.

Look for an identity test — macroscopic description, thin-layer chromatography, or a botanical fingerprint — in addition to the assay. For high-value botanicals such as saffron, resveratrol, and elderberry, an authenticity check is worth requesting even when it is not on the standard panel.

3. The assay method changes the number

HPLC and UV-VIS measure different things. UV often measures a family of compounds against a single reference standard, which is why a UV silymarin result of 80% and an HPLC silybin result are not comparable. Neither is wrong; they answer different questions.

Check the calculation basis too. A result "on a dry basis" will be higher than the same material "as is" when moisture is present. If two suppliers quote 50% EGCG, ask both whether the figure is as-is or dry basis before comparing price.

• Is the method stated for every assay result?
• Is the marker a single compound or a family total?
• Is the result as-is or on a dry basis?
• Is a chromatogram or spectrum available on request?

4. Physical parameters predict production problems

Moisture, ash, particle size, bulk density, and flowability rarely excite a buyer, but they decide whether the extract runs on a capsule line or cakes in a drum. A 95% extract with 8% moisture and poor flow can stop a filling machine; a 92% extract with tight moisture and good density may run perfectly.

For beverage and cosmetic applications, add solubility, color, and odor to the release specification. These are legitimate quality parameters, not subjective preferences.

5. The contaminant panel is where the real risk sits

Heavy metals, pesticide residues, residual solvents, and microbiology are the four panels that most often fail on arrival or in a customer audit. The limits should be written into the specification and matched to the destination market, not left to a generic statement.

Residual solvents should reflect the extraction system. An ethanol-water extract and a hexane extract do not carry the same solvent risk, and the COA should test what the process could plausibly leave behind.

• Heavy metals by ICP-MS, with individual results rather than a total
• Multi-residue pesticide panel appropriate to the herb and destination
• Residual solvents matched to the extraction and carrier system
• Total plate count, yeast and mold, E. coli, Salmonella
• Ethylene oxide where the destination market requires it

6. Red flags worth a follow-up question

A COA that is identical across batches, a document with no method column, a result that says "complies" without a value, or an expiry date that predates the manufacturing date are all reasons to pause. So is a supplier who cannot explain why a number is what it is.

None of these automatically means the material is bad. They mean the document is not yet doing its job, and the burden of proof has shifted back to you.