How to Read a Certificate of Analysis
Most brand owners file the certificate of analysis without reading it, which makes it paperwork instead of a control. Read the header for traceability, the methods for comparability, and the figures for margin. Each tells you something different.
What the Document Is For
This guide is for brand owners in India receiving certificates of analysis from manufacturers or suppliers.
A certificate of analysis states what was tested, by which method, against which limit, and what the result was. It is the evidence that a batch meets its specification, and it is what you would rely on if that batch were ever questioned.
Filed unread, it provides no protection, because a certificate that reports a result just inside specification, or omits a parameter you require, or uses a method you did not agree, tells you something important that you only learn by reading it.
Reading one properly takes a few minutes per batch, and it is the single highest-value quality habit a small brand can adopt.
Section by Section
What to check in each part.
| Section | What to Verify |
|---|---|
| Product name | Matches exactly what you ordered, including grade or strength |
| Batch or lot number | Present, and matching the physical goods |
| Manufacturing date | Present and consistent with the batch |
| Expiry or retest date | Present, and the remaining shelf life is acceptable |
| Quantity | Batch size stated |
| Specification reference | Which specification the results are judged against |
| Parameters tested | Every one your specification requires, none missing |
| Method per parameter | Named, not blank |
| Limit per parameter | Stated alongside the result |
| Result per parameter | An actual figure wherever a figure is possible |
| Testing date | Present, and sensible relative to manufacturing date |
| Laboratory | Named, with accreditation where relevant |
| Authorised signatory | Named, with a role, and signed |
Compare the parameter list against your own specification line by line. A missing parameter is the most consequential omission and the easiest to overlook, because the document looks complete.
Reading the Figures
The numbers carry more than pass or fail.
Where in the Range
A result at the edge of specification is a pass, but a different situation from one comfortably inside. Note where each result sits.
Across Batches
Keep the figures and compare batch to batch. Drift in one direction over several batches is a signal worth raising before it becomes a failure.
“Complies” Without a Figure
Acceptable only where the test genuinely has no numeric result. Where a figure is possible, ask for it.
Below Detection Limit
A legitimate result, but ask what the detection limit is, since below the limit of a sensitive method means more than below the limit of an insensitive one.
Recording the figures in a simple sheet, batch by batch, converts a filing obligation into a monitoring system. Trends only become visible when the numbers sit next to each other.
Methods Matter
This is where comparability lives.
A result is only meaningful alongside the method that produced it. Two laboratories testing the same material by different methods can report different figures, both correct for their method.
So the method name should appear against each parameter, it should be the method your specification names, and it should stay the same between batches. A method that changes between batches makes the figures non-comparable, and that change should be notified rather than discovered.
For assay results in particular, ask whether the method is specific to the compound claimed or measures a broader class of compounds, because the two give very different pictures of what the material actually contains.
And confirm the laboratory is accredited for the specific methods used, not merely accredited in general.
Warning Signs
Any one of these is worth a question.
- A parameter in your specification missing from the certificate.
- Methods blank or stated only as “in-house” with no further detail.
- “Complies” against parameters that should produce a figure.
- Limits not stated, so results cannot be judged.
- A testing date before the manufacturing date.
- A batch number that does not match the physical goods.
- No laboratory named.
- No authorised signatory, or a signature with no name or role.
- Identical figures across multiple batches, which real testing rarely produces.
- A specification reference that does not match the specification you agreed.
- A certificate issued for a different grade or strength than ordered.
Checklist
Per certificate received.
- Check the product name, grade and strength match your order.
- Check the batch number against the physical goods.
- Check manufacturing and expiry dates and remaining shelf life.
- Check the specification reference is the one you agreed.
- Compare the parameter list line by line against your specification.
- Confirm a method is named against every parameter.
- Confirm the methods are the ones your specification requires.
- Confirm limits are stated alongside results.
- Note where each result sits within its range.
- Ask for figures where “complies” appears and a figure is possible.
- Ask the detection limit where a result is reported as below it.
- Confirm the laboratory is named and accredited for these methods.
- Confirm a named signatory with a role has signed.
- Record the figures in a batch-by-batch sheet.
- Review that sheet periodically for drift.
- Raise any discrepancy before the batch is dispatched or sold.
Want Certificates You Can Actually Read?
Ask our team for a sample certificate of analysis with methods and limits stated before you place a first order.
Frequently Asked Questions
Why read the certificate of analysis at all?
Because it is the evidence that a batch meets specification and what you would rely on if the batch were questioned. Filed unread it provides no protection, and a missing parameter or a borderline result only becomes visible when you read it.
What is the easiest thing to overlook?
A parameter from your specification missing from the certificate. The document still looks complete, so compare the parameter list line by line against your own specification rather than scanning it.
Is “complies” an acceptable result?
Only where the test genuinely produces no numeric result. Where a figure is possible, ask for it, because a pass statement conceals whether the result sat comfortably inside specification or at its edge.
Why do methods matter?
Because a result is only meaningful alongside the method that produced it. Two laboratories using different methods can report different figures, both correct. The method should be named, should match your specification, and should stay the same between batches.
What should I do with the figures?
Record them in a simple batch-by-batch sheet and review it periodically. Drift in one direction over several batches is a signal worth raising before it becomes a failure, and trends only appear when the numbers sit next to each other.