How to commission independent testing, a practical procedure

Every article about supplement quality ends with “get it independently tested”, and then stops, which is where the actual difficulty begins.

Because a laboratory will test whatever you send against whatever you ask for, and if the sample, the scope or the method is wrong, you will pay for a certificate that answers a question you did not have.

So this is the procedure: how to choose the laboratory, what to ask for, how to take the sample, what it will and will not tell you, and what to do with the result.

This is orientation, not legal or regulatory advice. Accreditation scopes, prescribed methods and limits change. Confirm the current requirements for your product and market with a competent regulatory consultant.

First: decide what question you are answering

And this is the step people skip, which makes everything after it wasteful.

Four quite different questions get collapsed into “test it”.

“Is what the label says true?” → a finished-product assay of the declared actives.

“Is it safe?” → heavy metals, pesticide residues, microbiology, mycotoxins, and in some cases a screen for undeclared pharmaceuticals.

“Is it the plant it says it is?” → botanical identity, by microscopy, chromatographic fingerprint, or DNA methods.

“Will it still be true in eighteen months?” → a stability study, which is a different exercise entirely.

Each has a different scope, cost and method. Commissioning “a full test” without choosing produces a long report that misses the thing you were worried about.

Choosing a laboratory

Five criteria, in order of how much they matter.

Accreditation for the specific test, not in general. NABL accreditation is scope-specific: a laboratory accredited for microbiology is not thereby accredited for heavy metals by ICP-MS. Ask for the scope document and look for your test on it.

Independence from your manufacturer. A laboratory your manufacturer recommends, routes samples through, or owns is not independent. Choose your own, and do not tell the manufacturer which one.

Experience with botanical matrices. Herbal extracts are difficult samples, complex, variable, and prone to interference. A laboratory that mainly tests water or dairy may be accredited and still struggle with a polyherbal tablet. Ask what they test regularly.

Whether they will state the method and the limit of detection. A report that says “not detected” with no limit of detection is not a result.

And turnaround time, in writing, with what happens if they miss it.

For export, add one more: whether the destination accepts reports from this laboratory. Some markets require specific recognitions, and a report from the wrong laboratory is a cost repeated.

Taking the sample, where most of the error lives

This is the part that decides whether the result means anything, and it gets almost no attention.

Buy from the open market, not from the manufacturer. If you are verifying your own product, buy your own product from a retailer or a marketplace, as a customer. A sample the manufacturer hands you is a sample they chose.

Do not announce the test. An announced test is a test of their best batch.

Take more than one batch. One result is a data point; three batches across different manufacturing dates tell you whether there is a process or an accident. If you can only afford one, take the most recent.

Record the chain of custody. Photograph the sealed pack with the batch number and expiry visible, note where and when you bought it, and keep the receipt. Without this, a disputed result goes nowhere, because the manufacturer’s first response will be that the sample was not theirs or was mishandled.

Send it sealed and unopened. Opening a bottle to decant a sample introduces contamination and destroys the argument that the product was as sold.

Keep a duplicate. Buy two packs of the same batch and keep one sealed and stored properly. If the result is bad and contested, a retained duplicate for a second laboratory is what resolves it.

And for a disputed result, consider sending to a second laboratory blind, same batch, no mention of the first result.

Specifying the scope

Write the request, do not describe it on a phone call.

For an assay: name the compound class, the declared percentage or quantity you want verified, the matrix (finished product), and the method you want used. Ask for the result expressed both as a percentage of the extract and as milligrams per serving, because the second one is what your label claims.

For heavy metals: lead, arsenic, cadmium and mercury, individually, by ICP-MS or AAS, with the limit of detection stated, on finished product. And state which limit you want the result compared against, the Indian limit, or your export destination’s.

For microbiology: total plate count, yeast and mould, coliforms, E. coli, Salmonella, Staphylococcus aureus, and ask what else the applicable specification requires.

For pesticides: a multi-residue screen, and ask how many compounds the panel covers, because panels vary widely.

For mycotoxins: aflatoxins at minimum, and ochratoxin A for some materials.

For botanical identity: say what you suspect the substitution is, if you suspect one. “Confirm this is Asparagus racemosus and not Chlorophytum borivilianum” is a far better request than “check identity”, because it tells the laboratory which comparison to make.

And for undeclared pharmaceuticals: this needs to be asked for explicitly. It is not part of any standard panel. If a product works implausibly fast and strongly, a screen for steroids, NSAIDs or sedatives is the relevant test, and it is the one nobody commissions.

Reading the result properly

Six things to check before you conclude anything.

Does the batch number on the report match the pack you sent?

Is the matrix stated as finished product?

Is the method named, and is it the method you asked for?

Is the limit of detection stated, and is it low enough to be meaningful relative to the limit you care about?

Is the result expressed in the same units as your label claim, or do you need to convert? A percentage of extract and a milligram per serving are different statements, and a product can look compliant in one and not the other.

And is the laboratory’s accreditation shown for this specific test on this report?

Then interpret with the measurement uncertainty in mind. A declared 5% that assays at 4.7% is a different situation from one that assays at 1.6%. The first may be within analytical and batch variation; the second is a different product. Ask the laboratory for the uncertainty on the method rather than treating the number as exact.

What it will not tell you

Being honest about the limits of the thing being recommended.

It covers the sample, within the batch. Not the batch, strictly, a poorly blended lot can give a passing sample and a failing lot.

It says nothing about other batches.

It does not establish efficacy. A product that is exactly what the label says may still do nothing useful, and no test addresses that.

It does not cover anything you did not ask for. A clean heavy metals report has no bearing on pesticides, and a clean assay has no bearing on microbiology.

And a single clean result is not a quality system. Per-batch release testing by the manufacturer, with your own verification occasionally, is the system. A one-off certificate is reassurance.

What to do with a bad result

In order, and the order matters.

Verify before acting. Second laboratory, retained duplicate, blind. A false positive is possible and acting on one is expensive.

Preserve everything, the remaining product, the packs, the photographs, the receipts, the reports.

Then raise it with the manufacturer in writing, with the report attached, asking for their release testing data for that batch. Their response tells you as much as the result did.

If it is a safety result, stop selling that batch while you resolve it. Not after you resolve it.

Take advice on your obligations, which depend on the product category, the severity, and whether stock has reached consumers. Recall and reporting obligations are real and this is not the place to improvise.

And decide separately about the relationship. A bad result on one batch with a documented root cause and a process change is recoverable. A bad result the manufacturer disputes without data, or a second bad result, is not.

If you are a consumer rather than a brand

Commissioning a test on one bottle is usually not worth the cost to you personally, the result covers that batch, you cannot act on it commercially, and the money is better spent on the next six months of a product you chose carefully.

The high-value version for you is simpler: email the brand with the batch number on your pack and ask for the finished-product assay and the heavy metals report for that batch.

The brands that send it are demonstrating the only thing worth demonstrating. The ones that send a general marketing certificate without a batch number have also answered.

The short version

1. “Get it independently tested” is where the difficulty starts, not ends
2. First decide which of four questions you are answering
3. Label truth, safety, botanical identity, or shelf life, different tests
4. “A full test” without choosing misses the thing you were worried about
5. Accreditation is scope-specific, ask for the scope document
6. Choose a laboratory independent of your manufacturer
7. Do not tell the manufacturer which laboratory you used
8. Ask whether they test botanical matrices regularly
9. Require the method and the limit of detection to be stated
10. For export, check the destination accepts that laboratory
11. Most of the error lives in sampling, which gets no attention
12. Buy from the open market, as a customer
13. Do not announce the test, an announced test samples their best batch
14. Take three batches across different dates if you can
15. Photograph the sealed pack, note where and when, keep the receipt
16. Send sealed and unopened; keep a duplicate pack stored properly
17. For a disputed result, send blind to a second laboratory
18. Write the scope, do not describe it on a call
19. Ask for assay results both as a percentage and as mg per serving
20. Heavy metals: four metals individually, finished product, LOD stated
21. State which limit the result should be compared against
22. Pesticide panels vary widely, ask how many compounds
23. For identity, name the suspected substitution explicitly
24. Undeclared pharmaceuticals must be asked for, no panel includes them
25. A product that works implausibly fast deserves exactly that screen
26. Check batch number, matrix, method, LOD, units and accreditation
27. Convert to your label’s units before concluding anything
28. Interpret with measurement uncertainty, 4.7% against 5% is not 1.6%
29. The result covers the sample, within the batch, for that test only
30. It never establishes efficacy
31. A single clean result is reassurance, not a quality system
32. Bad result: verify first, with a second laboratory and the duplicate
33. Preserve product, packs, photographs, receipts and reports
34. Raise it in writing and ask for their release data for that batch
35. For a safety result, stop selling that batch while you resolve it
36. Take advice on recall and reporting obligations rather than improvising
37. One documented failure with a process change is recoverable; a disputed one is not
38. As a consumer, ask the brand for the batch report instead of paying for a test

Related guides: how to choose a contract manufacturer, supplement label compliance in India, supplement costing and margins. Related reading: heavy metals in herbal supplements, why the standardisation percentage is the most important line, reading a certificate of analysis, and what a GMP certificate does and does not mean.