// The process

How we test

Every result follows the same chain, documented and time-stamped, and publishes with the raw lab report attached. Nothing below changes because of who a vendor is.

01

Buy blind

We order as a normal retail customer, through anonymous addresses and rotated payment methods, off the same public shopping cart everyone else uses. No vendor may send us samples directly. A vendor who doesn't know a vial is headed for a lab can't cherry-pick it.

02

Capture the claim

In the same session we place the order, we archive the vendor's published COA and product page — the PDF where one exists, plus a full-page screenshot — with a SHA-256 hash, the capture timestamp and the source URL. That snapshot is what a comparison is made against.

03

Test & compare

The sample goes to an independent accredited lab under a coded ID. It runs HPLC-UV purity, LC-MS identity and net content, LAL endotoxin, sterility and a fentanyl screen. Both COAs publish together with the discrepancy table.

Chain of custody

Every sample carries timestamps at each handoff: purchased, received, shipped to lab, received by lab, results in, published. Those dates appear on the result. A result without a complete chain doesn't publish.

The lab receives coded sample IDs rather than vendor names wherever the workflow allows, so a result isn't shaped by whose vial it is.

How verdicts are decided

Verdicts are thresholds applied to measurements, not judgments about a vendor. They are published here so a reader can check our arithmetic against the attached documents:

  • Pass — measured purity and mass fall within methodology margin of the vendor's own COA and the product's label.
  • Underdosed / marginal — a measurable shortfall that is real but not gross: typically a purity gap of a few points, or a vial short of its labeled mass.
  • Fail — a significant discrepancy: purity materially below claim, mass materially below label, identity not confirmed, or a contaminant finding.

How the Pep Score is calculated

A rank is arithmetic applied to attached lab documents. The score is 70% quality and 30% verified value:

  • Pass rate (30) — across all tested batches, weighted toward recent results.
  • Purity accuracy (20) — average gap between claimed and measured purity.
  • Label accuracy (15) — average gap between labeled and measured milligrams.
  • Transparency (5) — the vendor publishes a real COA, from a named lab, with a batch number that matches what shipped.
  • Verified value (30) — verified $/mg against the median for that same peptide, so semaglutide vendors are compared to semaglutide vendors.

Ranking rules: a vendor needs at least two tested batches to receive a score at all (one test shows as Provisional). A verified fail caps the score at 49 until a later blind retest passes. Scores decay after twelve months without a fresh test, so passing once buys nothing permanent. Tiers are 90+ Verified Elite, 75–89 Recommended, 50–74 Mixed Results, below 50 Not Recommended.

Limitations — what a result does not tell you

A result describes the specific batch we bought, on the date shown. Batch-to-batch variance is real and expected; a pass is not a guarantee about a vendor's next production run, which is exactly why every vendor returns to the testing queue rather than earning a permanent badge.

Analytical methods have margins of error. Where a measurement falls inside that margin we say so rather than presenting it as a discrepancy. We report what the lab reported; we don't extrapolate to claims about safety or suitability for any use.

Independence: vendors cannot pay to be tested, retested sooner, ranked, or removed. We take no vendor sponsorships and no affiliate commissions from tested vendors. Vendors at Verified Elite or Recommended may license the “Pep Tester Verified” mark for their own marketing, at a flat published fee — the mark is earned in the lab first, licensing status is shown on the vendor's profile, and the scoring pipeline has no visibility into who licenses it. A failed retest revokes the licence immediately.

Questions about methodology are welcome, and corrections are published. See also why this testing exists and our terms.

Become a Member