How the Tested Index scores peptide vendors

What is the Tested Index trying to measure?

The Tested Index asks: across many results, over time, how reliably does a vendor ship what the label says? It answers with a score from 0 to 100 and a letter grade. This lesson follows the formula exactly as the scoring code applies it. The versioned method is on methodology, which reads its numbers from the same configuration the code uses.

Rankings come only from published records, never from hand-set ranks, and nothing a vendor pays for changes a score (how rankings work).

Which batches count?

Each batch is checked in this order, and the first rule that applies decides.

1. Not an anonymous purchase: excluded. Vendor-submitted and buyer-submitted records are published but never scored, for the reasons in lesson 12. This mirrors the covert, mystery-shopper sampling the WHO describes for medicine quality surveys, where "the identity and purpose of the buyer should not be generally known by the outlet being evaluated" (WHO TRS 996, Annex 7).
2. Published more than 12 months ago: excluded as outside the window.
3. Identity failed: included, and the batch scores 0. If the vial holds the wrong molecule, nothing else about it matters.
4. Identity, purity or amount missing: excluded as incomplete.
5. Everything else: included and scored as below.

How is each batch scored?

A batch that passed identity gets four component scores from 0 to 100, combined with fixed weights.

Purity uses the HPLC figure from lesson 3 and the HPLC-UV method. The 90 to 99 range keeps the difference between 95% and 98% visible.

Amount uses the signed deviation from the label, (measured minus label) / label x 100, from lesson 4. The 2% tolerance absorbs ordinary fill and measurement variation. Underfilling costs twice as much as overfilling because the buyer pays for milligrams that are not there. Both directions count: the US FDA warns that illegally marketed semaglutide and tirzepatide "could contain too little, too much or no active ingredient at all" (FDA).

Consistency asks whether a vendor's fills are predictable. For each passing batch, take measured divided by label. The coefficient of variation is "the ratio of the standard deviation to the mean" (NIST Dataplot reference), here expressed as a percent, using the sample standard deviation. It is the same idea as the relative standard deviation labs use for precision under ICH Q2(R2) (ICH Q2(R2)). One consistency value is computed per vendor and applied to each of its scored batches.

Transparency is the only component not measured from a vial, so it carries 10%: openness counts without outweighing what is in the product. The legit checklist asks similar questions from the buyer's side.

The batch score is 0.35 x purity + 0.35 x amount + 0.20 x consistency + 0.10 x transparency, rounded to one decimal place only after the sum, so carry unrounded component values through it.

How are batches combined into a vendor score?

Recent batches count more than old ones. A batch's weight is 0.5 raised to the power of its age in months divided by 6: a batch published this month weighs 1, one from six months ago weighs 0.5, one from twelve months ago weighs 0.25. Age is measured from the date the record was published.

The vendor score is the weighted average of its batch scores, rounded to one decimal place. Identity failures stay in the average at 0, with their full recency weight. A vendor is ranked only with at least 3 scored batches in the window, and identity failures count towards that 3. With fewer, the vendor is listed as insufficient data, with no score or grade. Ranked vendors are ordered by score.

What does a worked example look like?

The following is illustrative only. Demo Vendor A and every figure below are invented to show the arithmetic; they do not describe any real vendor or record. Demo Vendor A has a transparency rating of 0.6 and four tirzepatide records, all on 10 mg labels:

Eligibility. A4 is vendor-submitted, so it is excluded however good it looks. A3 is exactly at the 12-month edge and still inside the window. Three batches are scored, the minimum for a ranking.

Purity. A1: (98.2 minus 90) / 9 x 100 = 91.1. A2: 99.4 is above 99, so 100. A3: (95.5 minus 90) / 9 x 100 = 61.1.

Amount. A1 is 3.0% under label, 1 point beyond tolerance, at 4 points per percent: 100 minus 4 = 96. A2 is 4.0% over, 2 beyond tolerance, at 2 points per percent: 100 minus 4 = 96. A3 is 8.0% under, 6 beyond tolerance: 100 minus 24 = 76.

Consistency. The ratios are 0.97, 1.04 and 0.92. Their mean is 0.9767 and their sample standard deviation is 0.0603, so the coefficient of variation is 6.17%. Consistency = 100 minus 61.7 = 38.3, applied to all three batches. The swing from 8% under to 4% over is what costs this vendor most.

Transparency. 0.6 x 100 = 60.

Vendor score. (79.1 x 1 + 82.3 x 0.5 + 61.6 x 0.25) / (1 + 0.5 + 0.25) = 135.65 / 1.75 = 77.5. That is grade C.

Two variations show how the rules bite. If A3 had been published one day earlier, it would fall outside the 12-month window, leaving two scored batches: Demo Vendor A would show as insufficient data, with no grade at all. And if a fifth anonymous purchase, three months old, had failed identity, it would score 0 with a weight of 0.5 to the power of 0.5, about 0.707. The vendor score would fall to 135.65 / 2.457 = 55.2, grade F, while consistency would not change because it uses passing batches only.

What does a grade not tell you?

A grade summarises published records. It is not a prediction about the next order, a certificate for any batch, or a statement about safety or suitability for any use. It is recalculated as records are added and frozen monthly in the changelog. Your vial may not come from a scored batch at all: a batch in the ICH Q7 sense is a quantity "expected to be homogeneous within specified limits" (ICH Q7), and a different batch is a different run. So lesson 15 and lesson 11 still apply to every vial you hold.

A single compound's Index page applies the same formula to that compound's records only; the records themselves sit on compound pages such as tirzepatide. The lesson on amount vs label covers the amount measurement behind 35% of each batch score.

Check your understanding