
A single Certificate of Analysis is supposed to tell you everything you need to know about a botanical extract: what's in it, how much, and whether it's safe to use.
In practice, that one page hides dozens of decisions — about testing methods, equipment, reference standards, and lab accreditation. Two COAs can both say "Pass" and mean very different things. As a B2B buyer, knowing what happens between the harvest and that piece of paper is the difference between a reliable supplier and a costly reformulation.
Here's a buyer's-eye view of what really goes on inside a plant extract QC lab — and how to use that knowledge to source smarter.
A modern botanical extract lab runs every incoming batch through four categories of tests, in roughly this order.
The workhorse here is HPLC fingerprinting. The sample is pushed through a chromatography column, and the resulting chromatogram is compared against a reference standard for that exact botanical. If the characteristic peaks don't line up, the batch is rejected before anything else happens. This is also where the lab catches the most common fraud: a cheaper look-alike plant being passed off as a premium one.
HPLC again, this time configured for quantification — measuring the exact percentage of marker compounds (steviol glycosides, total flavonoids, chlorogenic acid, ginsenosides, and so on). UV-Vis spectrophotometry is used for quick total-class readings (total polyphenols, total polysaccharides), but key specifications are always confirmed by HPLC.
Four checks in this category, any one of which can fail a batch outright:
Heavy metals (Pb, Cd, As, Hg) — measured by ICP-MS or AAS
Pesticide residues — scanned by GC-MS and LC-MS/MS, often 100+ compounds
Microbiology — total plate count, yeast & mold, coliforms, pathogens, run in a cleanroom over 48+ hours
Solvent residues — GC analysis, critical for any ethanol- or acetone-based extract
Particle size distribution, bulk and tapped density, solubility, pH, color, moisture, ash. None of these appear on a glossy marketing sheet, but they decide whether your tablet presses cleanly, your beverage stays clear, and your next batch behaves like the last one.
A COA is only as trustworthy as the lab that issued it. Watch for:
The header. Who issued the report — a CMA/CNAS-accredited third party, or the supplier's own QC team? Both are valid, but they are not equivalent.
The method and standard. Is the heavy metal limit quoted against USP, EP, CP, or an in-house spec? Standards are not interchangeable, and the same number can pass one and fail another.
The detection limits. If the report says "not detected," check the LOD. An LOD of 0.01 ppm and an LOD of 1 ppm tell very different stories.
The pharmacopeia reference for marker compounds. A "10:1 ratio extract" is not the same as a "standardized extract with ≥24% flavonoids." Make sure the COA matches the product type you ordered.
The batch number and date. A clean report from a different batch is worthless. Always match the batch on the COA to the batch on the invoice.
Validate by batch, not by SKU. A trusted SKU is a function of its last 5–10 batches, not its first one. Ask for a recent COA and an older one — if the numbers drift, that's your stability story.
Run a third-party re-test on critical orders. For new suppliers, large orders, or any shipment crossing a customs border, an independent lab re-test costs a few hundred dollars and a few days. It is the cheapest insurance policy in B2B sourcing.
Build a specification sheet, not just a quote. A well-written spec — with testing methods, acceptance criteria, packaging, shelf life, and a clear dispute mechanism — does more for quality than any number of "quality first" slogans on a homepage.
If you're sourcing medicinal-and-edible plant extracts and want a supplier who can talk specs instead of just prices, we'd rather start from your formulation needs than a sales pitch. [Request a free sample] or [download our QC whitepaper] to see our batch-level documentation firsthand.