How to read a certificate of analysis

Every batch we ship goes out with a certificate of analysis. In our experience most brand owners file it without reading past the first line, and a meaningful number cannot say what it is for.

That is a shame, because the COA is the only document that tells you whether what is in the bottle is what you ordered. It is also your evidence if a customer, a regulator or a marketplace ever asks.

Here is how to read one.

What a COA is, and what it is not

It is: a record of tests performed on one specific batch, with the result for each and the specification it was tested against.

It is not:

• A certificate for your product in general. A COA covers one batch and nothing else.
• A licence or an approval. Those are separate documents.
• A guarantee of efficacy. It says what is in there, not what it does.
• Proof of quality on its own. A COA against a weak specification tells you a weak specification was met.

That last point is the one to internalise. The specification matters more than the result. A result of “complies” against a loose spec means very little.

The header, check this first

Before any test result, confirm the identification:

Field	What to check
Product name	Matches your order exactly
Batch / lot number	Matches the number printed on your bottles
Manufacturing date	Reasonable, and consistent with your order
Expiry / best before	Matches what your artwork says
Batch size	Consistent with your order quantity
Date of analysis	After manufacture; a date before manufacture is a serious red flag
Signature / authorised person	A named person with a designation, not a blank space

The single most common defect: a COA whose batch number does not match the bottles. That makes the document useless for traceability, which is the whole point of it, and it happens more often than it should.

Description, identification and physical tests

Description, appearance, colour, odour. Trivial-looking and genuinely useful: it is what you compare against if a later batch looks different.

Identification, confirms the material is what it says. For a herbal product this is the section that matters most and is most often thin.

What good looks like: TLC or HPLC fingerprinting against a reference, or a stated pharmacopoeial identification method. For botanicals, identity testing is not optional, species substitution is a documented, routine problem in Indian herbal supply chains, and a plant that fails identity is not a quality problem, it is a different plant.

What weak looks like: “Complies” with no method stated, or identification by organoleptic assessment alone.

Physical tests for tablets and capsules:

• Average weight and weight variation, consistency of the batch
• Disintegration time, how long before the dosage form breaks apart. If a tablet does not disintegrate, nothing in it is available regardless of what else the COA says. Worth a look.
• Hardness and friability, whether tablets survive transport
• Loss on drying / moisture, important for shelf life; high moisture invites microbial growth and degradation
• pH for liquids
• Bulk and tapped density for powders, relevant to whether your powder fills the pack you designed

Assay, the number that matters most

This is where the COA tells you how much of the active is actually present.

For vitamins and minerals: a quantified result per unit, against your label claim. If the label says 500 mg of calcium, the assay should show roughly that, with an overage allowance.

For standardised herbal extracts: the marker compound and its percentage. Curcuminoids in turmeric extract, withanolides in ashwagandha, silymarin in milk thistle, boswellic acids in shallaki.

Two things to look for:

First, is there an assay at all? Many herbal COAs skip it and report only identification and contaminants. That means nobody has measured how much active is in your product. If you are paying for a standardised extract, the assay is the proof you received one.

Second, what is the specification? “Not less than 95% of label claim” is a real specification. “Complies” is not. A range of 90–110% is normal for most actives.

On extract ratios: a specification of “10:1 extract” is not an assay and is weaker than it sounds, it describes a process, with no measured constituent. Where a marker exists, insist on a marker percentage.

Heavy metals, non-negotiable for herbals

Four are routinely tested: lead, arsenic, mercury and cadmium.

Why this is not a formality in India: heavy metal contamination in Indian herbal products is a documented, published problem. It arises from soil and water contamination, from processing, and in some traditional preparations from deliberately included mineral ingredients.

What to check:

• All four are present on the document, each with a numerical result
• Results are numbers, not “complies”, you want to see how close to the limit you are
• The limits stated match the applicable specification for your product category and market
• Testing is batch-wise, not a one-off from last year

Where it matters most: shilajit and other mineral-origin materials, bhasma-containing formulations, and any root or leaf material from unverified sourcing. If you sell shilajit without batch-wise heavy metal data, you are carrying a risk you have not priced.

And if you export: destination limits may be stricter than Indian ones. Check before you ship, not after.

Microbiology

Two groups of tests.

Counts:

• Total aerobic microbial count (TAMC)
• Total yeast and mould count (TYMC)

Specified pathogens, which must be absent:

• E. coli
• Salmonella
• Staphylococcus aureus
• Pseudomonas aeruginosa

What to check: that the pathogens are reported as absent, with the sample size stated, “absent in 10 g” is a real result, “absent” alone is less informative. And that the counts are given as numbers against a limit.

Why counts matter even when within limits: a result near the limit on a fresh batch may exceed it later in shelf life. A consistently high count across batches is a process signal worth raising.

Also relevant: aflatoxins, for materials prone to fungal contamination, particularly nuts, seeds and some roots. If your product contains those, ask whether aflatoxin is tested.

And pesticide residues, which matter for agricultural botanicals and are frequently omitted. Worth asking about, and essential if you are exporting or making organic claims.

What a weak COA looks like

Patterns that should make you ask questions:

“Complies” everywhere instead of numbers. The most common weakness. Numbers let you see trends and margins; “complies” hides both.

No method stated for any test. A result without a method cannot be reproduced or compared.

No assay for the active.

Heavy metals absent from a herbal COA.

No specification column, results with nothing to compare them against.

Batch number mismatched with your goods.

A date of analysis before the manufacturing date.

The same results, to the decimal, across several batches. Real analytical results vary. Identical numbers across batches suggest the document was copied rather than the batch tested.

No testing laboratory identified, in-house or external.

No authorised signatory.

In-house versus third-party testing

Both have a place, and the distinction is worth understanding.

In-house testing is normal, faster and cheaper, and a properly equipped manufacturer’s QC lab is a legitimate source of data.

Third-party testing at an accredited laboratory carries more weight with regulators, large buyers, marketplaces and export customers. NABL accreditation in India is the usual marker.

A sensible arrangement: in-house for routine batch release, with periodic third-party verification, and third-party for heavy metals on higher-risk materials, for export batches, and for any claim you are putting on the label.

And ask for the retained sample policy. A manufacturer should keep a sample of every batch through its shelf life. If a complaint arises two years from now, that sample is how the question gets settled.

What to actually do with the document

Keep every COA, matched to the batch, for the product’s shelf life plus a margin. This is a legal and practical record.

Compare across batches. The value compounds. Three batches where the curcuminoid assay reads 95.4%, 95.1% and 94.8% tell you the process is in control. A fourth at 91% is a conversation worth having before your customers have it.

Check the assay against what your label claims. If your label says 500 mg of standardised extract at 95% curcuminoids and the COA says 80%, your label is wrong, and that is your legal exposure, not the manufacturer’s.

Keep it accessible for complaints. When a customer, an FSSAI officer or a marketplace asks, producing the batch COA within the day is a different posture from spending a week looking for it.

Share it when asked. A brand that can produce batch documentation on request has an advantage in a category where most cannot.

Questions worth asking your manufacturer

1. Which tests are in-house and which are third-party?
2. Is the external lab NABL accredited?
3. Are heavy metals tested on every batch, or periodically?
4. Is the active assayed, and against which marker?
5. What identification method is used for each botanical?
6. Are pesticide residues tested?
7. How long are retained samples kept?
8. Can I have the COA at dispatch, rather than on request?
9. What is the procedure if a batch fails specification?
10. Can I see a COA from a previous batch of a similar product?

That last question is a good test of a manufacturer. One that hands over a sample COA without hesitation is used to being asked.

The short version

A COA covers one batch, not your product. Check the batch number against your bottles first, mismatches are common and make the document useless.

Look for numbers rather than “complies”, a stated method for each test, an assay for the active against a named marker, and all four heavy metals with numerical results on anything herbal. Confirm the assay matches what your label claims, because the label is your liability.

Then keep them and compare across batches. The trend tells you more than any single document.

More on supplier selection: choosing a contract manufacturer. On the licensing routes: FSSAI vs AYUSH.