Why We Started Testing Third-Party COAs Ourselves

This publication didn't start with a mission to independently verify lab results. It started with a single certificate of analysis that didn't add up — a purity figure that looked suspiciously clean next to a chromatogram with an oddly shaped baseline — and a decision to find out whether that was a fluke or a pattern.

The Certificate That Started It

The COA in question wasn't fake, exactly. The lab was real, accredited, and the underlying test had genuinely been run. But the specific chromatogram image included in the public-facing certificate had been cropped in a way that excluded a visible secondary peak present in the raw data we later obtained directly from the lab — a peak that didn't change the overall purity conclusion much, but that should have been disclosed rather than trimmed out of the image.

What That Taught Us

The problem wasn't fraud in the way people usually mean the word. It was that the gap between “laboratory data exists” and “consumer sees an accurate representation of that data” has more room for things to go missing than anyone had accounted for. Nobody along that chain had lied. Several people had each made a small, individually defensible choice about what to include, and the cumulative effect was a certificate that told a cleaner story than the underlying data did.

The Methodology That Came Out of It

That story is the direct reason this publication now regularly commissions independent split-sample testing rather than relying solely on supplier-provided certificates — the practice behind pieces like our recent multi-lab purity comparison. It's slower and more expensive than simply reporting what a COA says. It's also the only way we've found to reliably catch the gap between what a certificate shows and what the underlying data actually contains.

One Year Later

We've since run this same split-sample process across more than a dozen compounds. Most results have matched their certificates closely. A meaningful minority haven't — not through anything as dramatic as the original story, but through the same pattern of small, individually reasonable omissions that add up to a cleaner-looking number than the raw data supports. That pattern, more than any single dramatic finding, is the actual story of this publication's first year.