What a certificate of analysis actually tells you

August 13, 2026

A certificate of analysis is the only document that connects a claim on a label to a measurement in a laboratory. It is also easy to produce something that looks like one and isn't. This is what the real thing contains, and the four fields that separate the two.

The rule that created the paperwork

Until relatively recently no federal rule required a supplement manufacturer to verify the identity of the ingredients it used. The rule that changed this — 21 CFR Part 111, the good manufacturing practice regulation for dietary supplements — was published in 2007 and phased in by company size over three years.

Part 111 requires a manufacturer to set specifications and confirm the identity of incoming ingredients. The certificate of analysis is the paper trail that generates.

Worth saying what that does not amount to: the rule governs how a product is made, not whether it does anything.

Overhead close crop, a printed certificate page on a kitchen table with a pencil resting across it, daylight from one side, matte cream background

Figure 1 A manufacturing record, not a verdict Federal good manufacturing practice rules require manufacturers to set specifications and verify ingredient identity. A certificate of analysis is what that obligation produces on paper. 21 CFR Part 111, published 2007. Illustrative photograph.

What the document is, in one sentence

A laboratory's report of results for one specific batch, measured against a stated specification.

Specification and result are different things

A specification is what the material is meant to contain. A result is what a test found in a particular batch. A document carrying only the first is a product description with a laboratory typeface.

The five tests it usually reports

Identity

Confirming the material is what it claims to be. For a plant or a fungus this is not trivial: related species look alike as powder, and mislabelling in botanical supply chains is documented.

Potency, or assay

How much of the marker compound the material contains. This is the row that supports the number printed on the panel.

Heavy metals

Lead, arsenic, cadmium and mercury. Botanicals accumulate these from soil, so testing is a routine part of the process rather than a sign of suspicion.

Microbials

Total aerobic count, yeast and mould, and specified organisms — E. coli, Salmonella, Staphylococcus aureus.

Residual solvents

Extracts are made with solvents, commonly ethanol or water, and testing confirms how little of the solvent remains. The reference method is a pharmacopoeial one, USP General Chapter <467>.

Five-row schematic table diagram, cream ground, forest-green rules and type, five labelled bands, flat, no icons, wide margins

Figure 2 Five questions, five tests Identity, potency, heavy metals, microbials and residual solvents. A certificate covering fewer than these leaves the gap unmeasured rather than clean. Illustrative diagram of standard dietary-ingredient testing panels.

Identity is harder than it sounds

DNA barcoding works well on raw plant material and poorly on extracts, because extraction degrades the DNA it needs to read. For extracts the usual tools are chromatographic — a chemical fingerprint compared against a reference.

Concede the awkward part: a fingerprint match shows the chemistry is consistent with the named species. That is strong evidence rather than proof, and naming the method is what lets a reader judge which they have.

Potency, and the method underneath it

"Meets specification" is not a result. A result is a number with the method beside it.

Mushroom extracts are the clearest example. An enzymatic beta-glucan assay measures beta-glucans specifically, while a crude polysaccharide test will count starch from grain along with them — two numbers that can differ by a factor of several on the same material. That distinction is worked through in the one number to look for on a mushroom supplement.

Heavy metals, where actual limits exist

USP General Chapter <2232> sets daily-dose limits for elemental contaminants in dietary supplements: 5 µg for lead, 15 µg for arsenic, 5 µg for cadmium and 15 µg for mercury.

California's Proposition 65 is stricter, setting a maximum allowable dose level for lead of 0.5 µg a day — which is why so many products sold nationally carry a Proposition 65 warning while remaining comfortably inside the pharmacopoeial limits.

The trap in "not detected"

Not detected at what limit? A result reading "<0.1 ppm" is informative. A result reading "ND" with no limit of quantitation stated is a sentence rather than a measurement.

Horizontal comparison chart, four pairs of bars at very different scales, cream ground, forest-green bars, µg labels at bar ends, no gridlines

Figure 3 Two sets of limits, an order of magnitude apart Pharmacopoeial daily-dose limits allow 5 µg of lead. California's Proposition 65 level for lead is 0.5 µg a day, which is why the same product can be compliant and still carry a warning. USP General Chapter <2232> and California Proposition 65 maximum allowable dose level.

The four fields that decide whether it is real

A lot number that matches the pack. Without it the document describes some batch, not this one.
A test date. A certificate with no date cannot be placed against a manufacturing run.
The method, named. "Beta-glucan by enzymatic assay" is a method. "Beta-glucan: pass" is not.
The laboratory, named. With enough detail to look it up.

Miss any one of those and what has arrived is a specification sheet: a statement of what the product is supposed to be.

Macro crop, the header block of a laboratory report showing a lot number and date fields, matte paper, soft raking light, cream ground

Figure 4 The header decides it before the results do Lot number, test date, method and laboratory. A document missing any of the four describes an intention rather than a batch. Illustrative crop. No brand identified.

What ISO 17025 accreditation means

Competence, assessed by a third party

ISO/IEC 17025 is the international standard for the competence of testing laboratories. An accredited lab has been assessed on its methods, equipment, staff and quality system by an accreditation body.

It says nothing about the product

Accreditation is a statement about the laboratory. A perfectly accredited lab will report a bad result accurately, which is the point of it.

Accreditation has a scope

Laboratories are accredited for particular methods, not in general. A lab accredited for heavy metals by mass spectrometry may not be accredited for the assay that matters most on a given certificate. The scope document is public, and the accreditation number on the report is what looks it up.

Venn-style two-circle diagram with a small overlap, cream ground, forest-green outlines, two short labels, flat, no shading

Figure 5 Accredited, and accredited for this test Accreditation is granted method by method. A laboratory can be accredited and still report a result that falls outside the scope it was assessed for. Illustrative diagram. Scopes are published by the accreditation body.

What "third-party tested" does and does not mean

The phrase is not regulated

At minimum it means a laboratory other than the manufacturer performed a test. It does not say which tests, on which lots, against which specifications, or what the results were.

Certification is a different thing

Programmes such as USP Verified, NSF Certified for Sport and Informed Sport audit facilities, test product and licence a mark. A stronger statement than "third-party tested", and visible on the pack rather than on request.

Concede the cost. Certification runs into thousands of dollars per product and recurs, so plenty of small brands testing properly have no mark to show. Absence of a mark is not evidence of a problem; it is absence of evidence, which is why the lot certificate matters.

Why any of this is worth the trouble

Because label and content diverge often enough to have been measured. An analysis of 30 immune-health supplements found 17 inaccurately labelled, with some containing ingredients not on the label and others missing ingredients that were (Crawford et al., JAMA Network Open, 2022).

A separate analysis of 25 melatonin gummy products found 22 inaccurately labelled, with content ranging from 74% to 347% of the declared amount (Cohen et al., JAMA, 2023). Why that format is harder to hold to a number is covered in capsule, gummy or powder.

Two stacked proportion bars, each mostly filled forest-green with a small outlined remainder, cream ground, two short labels, no gridlines

Figure 6 Two analyses, the same direction Seventeen of 30 immune-health products and 22 of 25 melatonin gummy products were found inaccurately labelled in published analyses of US retail products. Crawford et al., JAMA Network Open, 2022 · Cohen et al., JAMA, 2023.

How to ask for the one that covers your pack

Could you send the certificate of analysis for lot [number printed on the pack]?

I am looking for the potency result for [ingredient], the method used, and the heavy-metal panel.

What a good reply contains

The lot number from the pack, a date, named methods, numeric results and the laboratory's identity. Ideally an independent laboratory, though a supplier's own certificate is not worthless.

What often arrives instead

A specification sheet. A certificate with the lot field blank. A different lot. Or a polite explanation that the document is confidential, which is itself an answer.

The five tests, and what each cannot settle

Test	What it establishes	What it cannot
Identity	The material matches the named species or compound	The quality of the growing or extraction
Potency	How much marker compound the batch contains	How much is absorbed, or whether it does anything
Heavy metals	Contaminant levels against a stated limit	Anything, if the detection limit is unstated
Microbials	Counts and named pathogens at time of test	What happens to the product in storage
Residual solvents	How much extraction solvent remains	Which solvent was used, unless named

Four things a certificate still cannot tell you

Whether the product works. That is a trial question, and no laboratory report speaks to it.

Whether the pack in a kitchen cupboard is from the tested lot. Only the lot number on the pack answers that.

Whether the sample was representative of the run. Sampling is its own discipline, and the certificate reports what arrived at the lab.

And whether the document is genuine. A certificate is a PDF. If a number really matters, the laboratory named on it can confirm it issued the report.

Where we stand

Our manufacturing specification requires a per-lot third-party certificate covering identity, the potency of each active, heavy metals and microbials, with the beta-glucan assay method named. The label figure for beta-glucans does not get printed ahead of the paper that supports it.

That is a process commitment rather than a result. Weal Focus should be judged on those documents when they exist, exactly as any other pack should be. The panel they sit behind is read in how to read a Supplement Facts panel in about sixty seconds.

Close crop, a lot number and best-by date printed on the back seal of a matte pouch, shallow depth of field, daylight, cream ground

Figure 7 The number that connects pack to paperwork The lot code stamped on a pack is what makes a certificate checkable. Without it, no document can be tied to the product in someone's hand. Illustrative photograph.

Simple decision flow, cream ground, forest-green rules, four questions descending to two outcomes, flat, generous white space

Figure 8 Four questions, asked in order Does the lot match, is there a date, is the method named, is the laboratory identified. Everything else on the document depends on those four answers. Illustrative diagram.

Sources for every figure, with the population studied in each, are on our citations page.

These statements have not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease.