Test data last updated 25 Jul 2026, 9,361 tests

Methodology

Finnrick rankings summarize published test evidence for a vendor and peptide. A higher rating means recent, classifiable samples have more consistently met the selected Finnrick standard for identity, purity, fill accuracy, and, in standard modes, batch labeling quality.

Independence

What Vendors Can and Cannot Change

Tests are not altered

Finnrick does not rewrite, remove, or soften test results because a vendor objects to them. If a factual correction is needed, it is handled as a correction record; the underlying test history remains part of the evidence base.

Rankings are not for sale

Finnrick does not accept payment to change rankings, suppress failures, promote vendors, or alter the scoring method for a vendor.

How a vendor improves

A vendor with a poor ranking has two real paths to improvement: fix the product or process issue, then have the product tested more regularly so newer evidence can demonstrate that the problem has been resolved.

Inputs

What Finnrick Tests

Finnrick sources samples from vendors and sends them to third-party labs. High Performance Liquid Chromatography is the primary method requested for peptide samples. Lab results are used to evaluate identity, purity, and fill accuracy: whether the expected active substance is present, how pure it is, and how much material is in the container compared with the vendor's claim.

Finnrick also reviews batch identification and quantity claims. Information printed directly on the vial is strongest because it stays with the product at the point of use. Physical shipment inserts can be acceptable. Website-only claims are weaker because they are separated from the product after purchase.

Safety Standards

What Counts as a Passing Sample

A sample must have enough information to evaluate identity, purity, quantity accuracy, and labeling. If a sample is missing one of those essentials, it is not counted in the rating for that standard.

Standard	Purity	Quantity	Labeling
Purity focus	at least 99.5%	within plus/minus 35% of claim	no label requirements
Finnrick Standard	at least 99.5%	within plus/minus 25% of claim	label score at least 0.5
Purity focus strict (Members)	at least 99.8%	within plus/minus 35% of claim	no label requirements
Finnrick Super strict (Members)	at least 99.8%	within plus/minus 15% of claim	label score at least 2.0

Underfill and overfill both fail when they exceed the allowed range. A failed identity result fails the sample. Purity focus modes use the compound result and fill accuracy only; batch labeling does not affect those modes.

The two strict standards are included with every paid Finnrick membership, alongside original COA downloads.

Labeling

How the Labeling Score Works

The labeling score measures whether the sample can be tied back to a batch and whether batch-specific claims are communicated in a durable, product-adjacent way. The score ranges from 0 to 2.

If only a label claim is present

2 points: the batch identifier is printed on the product container.
1 point: the batch identifier is provided on a separate physical support shipped with the container.
0 points: the batch identifier is website-only, purchase-page-only, or absent.

If both a label claim and batch claim are present

The score is the sum of two parts: batch identifier quality and batch claim communication quality.

Batch identifier: 1 if printed on the container, 0.5 if on a separate physical support, 0 if website-only or absent.
Batch claim communication: 1 if printed on the container, 0.5 if on a separate physical support, 0 if website-only.

For Finnrick Standard, a score of 0.5 or higher is acceptable. For Finnrick Super strict, a score of 2.0 or higher is required. Purity focus modes do not apply the labeling score.

Rating Percent

Natural Rating vs Final Rating

Finnrick shows two related ideas: the natural rating and the final rating. The natural rating is closest to the raw evidence. The final rating is what leaderboards use, because it handles thin data more carefully.

Step 1

Classify each sample

Each test is marked pass, fail, or incomplete under the selected standard. Incomplete tests are shown in the test history but do not count toward that standard's pass rate.

Step 2

Weight recent evidence more

Newer tests count more than older tests. The current evidence half-life is 730.5 days, so a test that old counts about half as much as a new test.

Step 3

Adjust thin histories

If a vendor has only a small number of classifiable tests for a peptide, the final rating is pulled toward the peer baseline for that peptide. This prevents one lucky or unlucky test from looking too certain.

In the formulas below, w_i is the age weight for sample i, p_i is 1 for a passing classifiable sample and 0 otherwise, and c_i is 1 when the sample is classifiable under the selected standard.

Time weighting

w_i = 2^{-\frac{\text{age}_i}{730.5}}

Natural rating

R_{natural} = \\frac{\\sum_i w_i p_i}{\\sum_i w_i c_i}

Posterior mean

R_{mean} = \frac{\sum_i w_i p_i + s\,R_{prior}}{\sum_i w_i c_i + s}

Prior strength

s \in [0, 20]

Evidence prior decay

s = \frac{q_{product}\,24}{1 + \frac{N_{effective}}{q_{product}\,8}}

Product evidence scale

q_{product} = \operatorname{clamp}\left(\frac{M_{product}}{M_{P90}}, 0.3, 1\right)

How to read the final rating

As a vendor accumulates more recent classifiable tests, the final rating moves closer to its natural rating. With limited evidence, the adjustment is stronger. Mature product boards can start at up to 24 peer-baseline pseudo-samples, while thinner product boards scale that prior down based on their median vendor evidence.

The base evidence half-life is 8 effective tests. The same product evidence scale also shortens that half-life for thinner boards, so their prior decays sooner.

Leaderboards

How Rankings Are Ordered

Each leaderboard ranks vendors for one peptide. Vendors are ordered by final rating first. Ties are broken by effective sample volume, classifiable sample count, passing sample count, total test count, and vendor name.

The ranking is not an endorsement or medical recommendation. It is a compact summary of Finnrick's published testing evidence.

Bad results stay visible

A vendor cannot negotiate a failure away. The practical way to recover from a bad result is to fix the cause and build a newer, stronger testing record for the same product.

back to leaderboards view tests