Almost every supplement buyer asks for a COA before the first order. Far fewer read one properly. A certificate of analysis is only worth the fields you actually check, and an official-looking document can still tell you nothing about the batch in your warehouse. Here is what to look at, in the order that matters.

What a COA is — and what it is not

A COA is a batch-specific record of test results compared against a specification. It is not a certificate of approval, not a safety guarantee, and not a marketing document. It answers one narrow question: for this lot, did these tests produce results inside these limits, using these methods?

If a supplier sends a COA that is not tied to a lot number, it is a specification sheet wearing a lab coat. Ask again.

The four blocks every COA needs

  1. Product identification — material or finished-product name, grade, and the specification it was tested against.
  2. Lot or batch number — plus the manufacturing and expiry (or retest) dates.
  3. Test, method, specification, result — one line per parameter, with all four columns filled.
  4. Authorisation — the testing laboratory, the date of analysis, and a signature or authorised signatory with contact details.

Missing any of the four, and the document cannot be used for a supplier qualification file or a customer audit. This is the first thing an auditor checks, and the first thing that quietly fails.

The "specification versus result" trap

Look at how the numbers sit next to each other. A specification is a limit — "not less than 95.0%", "not more than 3 ppm". A result is what the lab measured — "96.8%", "0.4 ppm".

The trap is a COA where the result column repeats the specification exactly, line after line. That pattern suggests the document was typed from a template rather than populated from a lab record. Real results wander inside the range; identical values across every parameter are a warning sign, not a sign of quality.

Related trap: a result with no units, or a result for an ingredient that is not in the formula. Both appear more often than they should.

Assay, identity and the parts buyers skip

Assay (how much active is present) gets all the attention. Identity matters just as much: does the material match what was ordered? For botanical extracts this usually means a marker compound or a fingerprint comparison, and for a blend it means each declared component is accounted for.

Also check the physical parameters, because they drive production: particle size and bulk density for powders, viscosity for liquids, and moisture or water activity for anything hygroscopic. A material can hit its assay target and still jam a filling line.

Microbiology and heavy metals

For finished supplements, microbiology is usually reported as total aerobic count, yeast and mould, and the absence tests for specified organisms. For raw botanicals, heavy metals are the decisive line item — lead, arsenic, cadmium and mercury, typically measured by ICP-MS.

Two questions to ask: is the limit set by your specification or copied from a generic template, and was the test run on the raw material, the finished product, or both? A raw-material COA does not describe the product you will ship.

Lot traceability is the field that actually protects you

Write the lot number down and keep it. If a customer complaint arrives eight months later, the lot number is the only path back to the batch record, the raw-material COAs behind it, and the retain sample. A supplier who cannot connect a finished lot to its input lots cannot support you through an audit or a recall.

In-house versus third-party testing

In-house laboratories are normal and fast, and most routine parameters are tested that way. Third-party testing matters most when a result is contested, when a customer specifies an accredited laboratory, or when the parameter is one the producer has an incentive to under-report. Ask which parameters are outsourced, to whom, and whether the report carries the laboratory's own reference number.

Seven red flags

  1. No lot number, or the same lot number across shipments months apart.
  2. Result column copied from the specification column.
  3. No method named, or a method that cannot produce that parameter.
  4. Missing units, or mixed units within one table.
  5. Testing date before the manufacturing date.
  6. Heavy metals reported as "pass" with no figure.
  7. A scanned document with a signature that is identical on every COA you have received.

What to ask before the first order

Ask for a sample COA for the exact product and grade you intend to buy, the specification document behind it, and confirmation that a COA will accompany every production lot rather than only the first one. Then compare two COAs from different lots: the layout should be identical, the numbers should not.

How Haiyue Shengda handles documentation

Every batch we release ships with a COA covering the parameters agreed in the specification, and we will discuss which tests are run in-house and which are sent out before you place the order rather than after. Trial batches start from 100 units per SKU, samples ship in 3–5 days, and production follows within 7 days. Review our quality and compliance page, or contact us with the specification you need to meet — we reply within one hour.