The number is real. The sample might not be.
A certificate of analysis is a factual document. The laboratory measured what it was given and reported the result honestly. The question that decides whether the number means anything is the one the certificate almost never answers: who selected the material that reached the lab?
If the processor drew the sample, packed it and posted it, then the COA describes that parcel. Whether the parcel represents the twelve drums on your purchase order is an assumption — not a measurement.
Three ways a sample stops representing the lot
- Selection. Material is drawn from the best drum rather than at random across the lot.
- Timing. The sample is taken before a blending or drying step that changes the profile.
- Substitution. A retained sample from an earlier, better batch is sent in place of the current one.
None of these require fraud. A processor under delivery pressure, choosing the drum nearest the door, produces the same outcome as one acting deliberately.
What independent sampling changes
When the laboratory draws its own sample from the sealed lot, the chain from material to number has no discretionary step in it.
That is the whole point. We do not hand a bottle to a lab. An accredited laboratory attends, draws to guideline from the sealed lot itself, and reports directly. The result is not more flattering — often it is less — but it describes the material you are actually buying.
What to ask your supplier
- Who physically drew the sample, and from where in the lot?
- Was the lot sealed at the time of sampling?
- Is the laboratory accredited, and did it receive the sample directly?
- Can the COA be traced to a specific lot number on the packing list?
A supplier who can answer all four without hesitation is telling you how they work. A supplier who cannot is telling you something too.
We built our workflow around a lot that came back. It was not fraud — it was a small gap between what the paperwork said and what the lab found. Everything we do now traces to closing that gap.