A certificate of analysis shows numbers; a certificate of conformity shows confidence. Which one your shipment actually needs depends on the decision someone downstream is going to make with it.
Here is the whole difference in two sentences. A certificate of analysis reports measured results for one specific batch — assay, impurities, water content, each value sitting next to the limit it was judged against.
A certificate of conformity states that the batch meets its specification, and shows no numbers at all. One document lets the reader judge. The other asks the reader to believe.
That gap sounds academic until it costs a week. Then it costs a week.
A buyer in Chennai learned this during a customer audit, not at receiving. His team had filed certificates for every incoming lot — neat folders, one per batch, all signed. The auditor opened one, scanned two lines, and asked a question nobody on the buyer’s side could answer: where are the results? The paper in his hand was a conformity certificate. It said the material met the specification. It never said what the specification numbers had actually been. Two years of receiving records, and not one page in the folder contained a measured value.
The audit did not fail over quality. It failed over a document nobody had looked at closely enough to notice what it was.
Two Documents, One Word Apart
Both certificates arrive on the same letterhead, signed by the same quality department, covering the same drums. That is exactly why they get mixed up.
The certificate of analysis — CoA — is a test report. It names the product, the batch, the specification each test was judged against, and the value that testing actually produced. Where the how each line of a CoA maps to a limit matters most is this pairing: result beside criterion, on every row, for the lot you received. Remove any one of those pieces and the document stops doing its job.
The certificate of conformity — CoC — is a declaration. Someone in the supplier’s quality unit signs a statement that the goods conform to the agreed specification or standard. On some versions you will find the specification listed by name. What you will not find is a single measured value. There is nothing to check the statement against. The document is a promise wearing a signature.
ICH Q7 is blunt about this when it comes to intermediates and APIs: incoming material should be tested on receipt or accepted against a certificate of analysis, with at least an identity test confirmed against the label. A conformity statement does not appear anywhere in that expectation. The regulation wants data, not assurance.
Side By Side, Without The Jargon
The fastest way to see which document you are holding:
| What you are looking at | Certificate of Analysis (CoA) | Certificate of Conformity (CoC) |
|---|
| What the page contains | Measured results, one per test, with limits | A signed statement of conformance |
| Numbers on the page | Yes — beside the limit they were judged against | None |
| Tied to | One batch and its identifier | An order, a shipment, or a product line |
| What a reviewer can do | Compare, trend across lots, verify against incoming tests | File it |
| What it supports | Release decisions, investigations, audits | Low-risk items where no measurement applies |
| What it cannot do | — | Replace data in a release decision |
One line in that table does most of the damage when it gets ignored: trending. Ten CoAs from ten lots of the same intermediate tell you whether a supplier runs comfortably inside the limits or hugs the edge. A stack of conformity certificates tells you nothing on that axis, because there is nothing on the page to stack.
Where The Two Certificates Get Swapped
Three moments in a normal purchase generate most of the confusion.
The first is customs. Some destination markets ask for a conformity certificate at import — a document attesting the goods match the declared description and standards. Clearing agents, who handle paperwork for a living, sometimes request a “certificate” without specifying which. The supplier attaches whichever is handy. The shipment clears. Then the buyer’s QC asks for the CoA and finds out the folder only holds the declaration. Nobody flagged the swap, because both documents looked official and neither agent knew the difference was load-bearing.
The second is receiving. A CoC attached to a drum looks like a release document, and a rushed warehouse team treats it as one — the lot goes straight to stock. That is precisely the step where your receiving checklist should catch it: no measured values on the paper, no release. The checklist line costs ten seconds. Skipping it costs the traceability an auditor will ask for two years later.
The third is the first order from a new supplier. A conformity certificate is cheaper to produce, faster to issue, and — this is the part that should raise an eyebrow — sometimes offered in place of the analytical certificate by suppliers who never ran the tests. A buyer who accepts it has waived the only evidence that testing happened at all. We say this as a manufacturer: when a new buyer’s RFQ arrives, the forged certificates that survive a casual glance are almost never the conformity kind, because nobody bothers faking a document with no data on it. The analytical certificate is where the fakes live. Treat a supplier’s reluctance to issue one as information.
When A CoC Is Actually Enough
None of this makes the conformity certificate useless. It has honest jobs.
Packing components are the obvious one. Drums, liners, desiccants, labels — for these, the relevant claim is usually categorical: food-grade liner, UN-rated drum, no direct product contact. A signed conformance statement fits, because the property being attested is not a measured chemical value. Spare parts and non-product-contact items land in the same bucket.
Some regulatory attestations also ride on conformity certificates — statements that a shipment complies with a transport rule or a trade requirement. Again: the claim is categorical, not analytical. Nothing on the page pretends to be data.
The test that separates honest use from shortcut is one question. Could the claim on this document have been produced by a measurement? Assay can. Water content can. If yes, the claim needs a CoA, full stop. If no — if the claim is about category, rating, or compliance with a rule — a CoC carries it fine.
Where buyers go wrong is applying that test to the wrong document. A conformity certificate covering the drum is fine. A conformity certificate standing in for the intermediate’s analytical data is not, and no clause in a purchase order fixes that after the fact.
Whose Signature Should Be On It
There is a quieter version of this problem, and it starts before the document even reaches you: who issued it?
An analytical certificate should come from the quality unit at the site that made the material. That sounds obvious until you see how the paperwork actually flows. A trader in Mumbai buys from a plant, resells to you, and forwards “the certificate” with the shipment. Sometimes the forward is faithful — factory letterhead, QC signature, batch results intact. Sometimes the document has been reprinted on the trader’s own letterhead with the same numbers retyped, and now nobody can say whether the values were copied or invented. One of those versions is evidence. The other is a story about evidence.
How do you tell them apart at a glance? Look at two things. The issuing address should point at a production site, not an office floor in a commercial district. And the signature block should name a person and a quality function — “QC Manager” with a surname, not a sales director’s stamp over a blank line. Neither check is chemistry. Both take under a minute.
The conformity certificate has the opposite exposure. Because it contains no data, anyone can produce one — a trader, an agent, a laptop. That is not an accusation; it is the reason a CoC can never carry more weight than the issuer deserves. A promise is only as good as the promise-maker, and a page of results is only as good as the lab that produced them. Which one you are reading determines which question you should be asking about the signature.
We watch buyers make this call differently once they have been burned once. The good ones stop asking “did a certificate arrive?” and start asking “whose certificate arrived, from where, signed by whom?” Same folder, same paper — completely different evidence.
Five Checks At Receiving, Five Minutes Total
None of this needs a laboratory. It needs a habit.
First: which certificate is in the folder? Read the title line. Conformity documents usually carry the word “conform” or “comply”; analytical ones say “results” or “analysis”. If the folder for a production intermediate holds only a conformity statement, stop there and ask for the CoA before the material moves.
Second: is there a number on the page anywhere? Measured values with units. A document listing “typical values” is neither a CoA nor a CoC — it is a sales sheet wearing a signature, and what a complete analytical certificate carries includes lot-specific results, not averages across a campaign.
Third: does the batch identifier on the certificate match the drum? This is the check that catches copy-forward certificates, and it connects directly to the identifiers stamped on your drums — three systems, three strings, one lot. The paper and the steel have to agree.
Fourth: is every result sitting next to its own limit? Results without criteria are unreadable, and criteria without the limits each result gets judged against are just decoration. The pairing is the whole mechanism.
Fifth: are the dates coherent? Analysis date before manufacture date, or a retest window that expired in transit — both show up more often than anyone expects. The certificate’s date fields are what the date field that starts the clock is built on, and an incoherent one poisons the retest schedule downstream.
Five checks, five minutes, one clerk. That is the entire cost of never having the Chennai problem.
The Clause That Ends The Argument Before It Starts
Everything above can be compressed into one sentence in the purchase order or the quality agreement, and it saves more time than any checklist: each production lot ships with a lot-specific certificate of analysis reporting measured results for every specification parameter; certificates of conformity do not satisfy this requirement for the intermediate itself. That is the clause that ends the argument before it starts — the receiving team stops guessing, the clearing agent stops substituting, and a supplier with something to hide stops bidding.
Why put it in writing when a phone call feels faster? Because the swap happens in transit, between people who never spoke. Your purchase order reaches a documents department; the drum reaches a warehouse. The only thing both of them read is the paper.
And one admission from our side of the trade, since buyers rarely hear it: manufacturers issue conformity certificates more often than we issue refusals to issue them. They are quick, they are polite, and they close emails. The discipline is on the buyer’s side — knowing which document the decision needs, and asking for it before the shipment, not during the audit.
The auditor in Chennai did not catch a bad supplier. He caught a folder full of promises where the numbers should have been. Both documents looked the same, were signed by the same company, and only one of them had ever contained a measured value.
That is the real gap between promise and proof — and it is closed with one clause, five checks, and the refusal to release a lot on a page that never showed a number.
