How the Vendor Benchmark Works

A transparent walkthrough of how PeptideBenchmark combines Finnrick and Peptide Critic into one ranked board.

---

What this board is actually doing

PeptideBenchmark is not pretending to be a lab, a pharmacy, or a regulator. The vendor board is a normalized benchmark layer built from public source records.

Right now the two upstream sources are:

• Finnrick vendor records
• Peptide Critic vendor directory

Each source has its own scale, coverage, and blind spots. Our job is to make those differences legible instead of hiding them behind a single mystery score.

The three numbers that matter

Every vendor card surfaces three different ideas:

Benchmark score

This is the normalized source score.

• If a vendor appears on both sources, we blend Finnrick’s 10-point score with Peptide Critic’s rating after converting the Peptide Critic value onto a 10-point scale.
• If a vendor appears on only one source, the benchmark score stays single-source until more coverage exists.

We no longer show a public letter grade on vendor cards because it created too much confusion with the actual board ordering. The public UI now keeps the benchmark layer numeric so readers can compare Benchmark score and Rank score directly.

Rank score

This is the ordering score used on the leaderboard.

Rank score now does two jobs:

• first, it pulls thin source evidence back more aggressively so a perfect Peptide Critic-only vendor does not automatically outrank a deeper consensus record
• second, it allows a bounded transparency/trust modifier when PeptideBenchmark has actually reviewed meaningful public verification, public archive depth, or structured gated certificate evidence for that vendor

That means benchmark score and rank score are intentionally not the same thing.

Confidence band

Confidence shows how much underlying signal is carrying the score.

Higher confidence usually means:

• more Finnrick test records
• more Peptide Critic reviews
• or coverage from both sources instead of just one

It can also be helped directionally by quality cues like source-backed Peptide Critic metadata, but source overlap and evidence depth still matter more than decorative perfect scores.

On the public site, some one-source entries are labeled Single-source in the UI so readers do not mistake limited coverage for a negative quality judgment. Under the hood, the model still computes an internal confidence band that drives ranking.

Why a perfect score is not always a top rank

A vendor can show a very strong raw score and still rank below a competitor with broader evidence. That is intentional.

The board is designed to reward signal quality, not just signal intensity.

For example, a Peptide Critic-only vendor with a perfect rating but limited source depth should not automatically sit above a consensus vendor supported by dozens of tests and reviews across both sources.

Likewise, a vendor with unusually strong public verification or public archive depth may deserve to rank above a similarly scored vendor whose trust surface is much thinner. That is why the board now allows a bounded transparency/trust modifier instead of pretending all source ties are equal.

What the board does not prove

Even a strong benchmark profile does not prove:

• current batch identity
• current sterility
• current endotoxin status
• fulfillment quality on your next order
• legality or medical suitability for human use

This board is a research and sourcing tool, not a substitute for due diligence.

What we want the site to become

The long-term goal is to keep the benchmark layer transparent while adding stronger editorial work on top:

• clearer source methodology
• scheduled source refreshes
• original guides that explain what a score can and cannot tell you
• eventually, first-party review standards that are disclosed separately from source aggregation

If a future page makes an editorial claim, it should be labeled as editorial. If a score comes from a source sync, it should be labeled as source-derived.

The long-term goal is to keep that separation clean:

• Benchmark score stays source-derived
• Rank score becomes the board’s best directional ordering value
• and first-party trust/transparency evidence stays visible as a bounded overlay, not a hidden replacement for the source record