How is the composite score calculated?

The composite is the plain arithmetic mean of the five rubric dimensions, rounded to one decimal place. All five are weighted equally. There is no hidden weighting, no bonus, and no adjustment applied after the mean is taken.

What decides the order vendors appear in?

Composite score, descending, and nothing else. Ties break alphabetically so the order is stable between builds. There is no pin, no boost, and no commercial tier in the ordering function. It is a single sort in a public file, and changing that behaviour would take a visible code change.

Can a vendor pay for a higher score or a better position?

No. No position on this board has ever been sold. Some outbound links on this site are commercial and that is disclosed at /legal/disclosure, but no commercial arrangement is readable by the scoring or ordering code.

Why did a vendor's score change?

Every score that moves carries a dated entry naming the evidence that moved it. Entries are one of four kinds: a listing, meaning the vendor entered the directory at that score; a recalibration, meaning the rubric scale changed and the whole board was restated on one date, which is not a gain or a loss for anyone; a movement, meaning the vendor itself changed and we saw the evidence; or a correction, meaning we scored something wrong and the vendor did not change at all. Only a movement is displayed as a delta. A correction is labelled as our error, because presenting our own mistake as a vendor rising or falling would be a second dishonesty on top of the first.

Why are some vendors marked provisional?

A provisional entry is scored from published documentation rather than from material we have handled. It is an honest desk assessment, the analytical dimension is capped while it stands, and the entry is re-scored once we have run a lot ourselves.

Do you sell peptides?

No. researchuseonlypeps holds no inventory, ships nothing, and fulfils no orders. We are a reference and review publisher. Everything listed here is for laboratory research use only and is not for human or veterinary use.

Methodology

How we score peptide vendors

We do not sell peptides. We hold no inventory and ship nothing, which means we have no batches of our own to defend and nothing to protect by scoring a competitor down. What we have instead is a rubric, and the only thing that makes a rubric worth anything is publishing it before the scores rather than after them.

Five dimensions, weighted equally. The composite is their mean. The board is sorted on that composite and on nothing else. Everything below is already enforced in the code that renders the vendor board, and where the two ever disagree, the code is the thing that is true.

operations scored

with per-batch certificates we checked

dimensions, equally weighted

The five dimensions

Each is scored 0 to 5. Each carries the same weight as the other four. A vendor that is excellent at one and absent at another lands in the middle, which is the correct place for it.

Analytical purity

Is the material what it says it is, at the strength it claims?

Scored from HPLC and mass-spectrometry results we have read, weighted toward results we have handled a lot against rather than results a vendor has published about itself. A chromatogram we can inspect beats a percentage we cannot.

A vendor cannot pass 4.0 here on published documentation alone. Above that we have run a lot ourselves.

Certificate transparency

Can a buyer connect the paperwork to the vial in their hand?

Per-batch certificates, a named analysing laboratory, a lot number that reconciles to the vial, and a route to the document that does not require emailing anyone. This is the dimension most vendors lose ground on, and the cheapest one to fix.

One certificate published under a product name with no batch traceability scores as undocumented, not as documented once.

Handling and fulfilment

Does the material arrive in the condition the certificate describes?

Cold-chain handling where the compound needs it, packaging that survives transit, realistic dispatch times, a stated reship policy, and origin disclosed so a buyer can judge customs risk for themselves.

Speed alone does not score here. A fast shipment of a warm lyophilised peptide is a handling failure that happened promptly.

Support and disputes

What happens after you ask a question they would rather not answer?

Response time is the smallest part of this. What moves the number is whether a specific analytical question gets a specific answer, and what a vendor does when a batch is challenged.

A vendor that answers quickly and vaguely scores below one that answers slowly and precisely.

Value against verified purity

What does a milligram of documented material actually cost here?

Price normalised against purity we have evidence for, not against label claim. A cheap vial of undocumented powder is not good value at any price, and this dimension is where that gets recorded.

Being expensive is not a penalty. Being expensive without documentation is.

The composite

composite = (quality + coa + shipping + support + value) / 5

Rounded to one decimal place. No weighting, no bonus, no adjustment applied afterward. The current leader is Axiom Labz at 3.9, and it holds that position because that number is the highest one, not because of anything else.

Ordering is a single sort on the composite, descending, with ties broken alphabetically so the board is stable between builds. There is no pin, no boost, and no commercial tier anywhere in that function. If that ever stops being true, the change has to happen in that function, in a diff, in public.

What the numbers mean

4.5 to 5.0Evidence we handled ourselves: Reserved. We have run material from this vendor, and the results support the claim. No vendor reaches this band on paperwork alone.
4.0 to 4.4Strong and independently documented: Per-batch third-party certificates that reconcile, checked by us, with no open concerns.
3.0 to 3.9Adequate, with gaps we can name: The midpoint of the scale. A competent operation with specific things missing, each stated in the entry.
2.0 to 2.9Material gaps: Documentation is partial, inconsistent between batches, or unverifiable at the point it matters.
Below 2.0We would not use this ourselves: Reserved for operations where the missing evidence is the point rather than an oversight.

What we refuse to let count

A rubric is defined as much by what it excludes as by what it measures.

Payment of any kind. No position on this board has ever been sold, and there is no tier that can be bought.
Affiliate or commercial relationship. Some outbound links here earn a commission, disclosed at /legal/disclosure, and the ordering function cannot read that fact.
Event attendance, press coverage, and creator posts. These appear under Company activity on a vendor page, labelled not scored, and they never touch a dimension.
Volume, popularity, or how often a name comes up in forums.
Our own convenience. A vendor being difficult to deal with is scored under support, not taken out on the composite.

When a score moves

A list that reshuffles and never says why is an advertisement wearing a ranking's clothes. So every score that moves on this board carries a dated entry naming what we saw, and those entries are one of four kinds, because four different things move a number and they are not the same statement.

Listing: The vendor entered the directory at that score. Not a gain, just an arrival.
Recalibration: The rubric scale changed and the whole board was restated on one date against one scale. The vendor did not change. Nobody gains or loses ground at a recalibration, and we do not render a delta for one.
Movement: The vendor itself changed and we saw the evidence. This is the only kind displayed as a delta, because it is the only kind that is a performance.
Correction: We were wrong and the vendor did not change. A correction names what we got wrong, what we did to find out, and the page anyone can use to repeat the check. It is labelled as our error rather than rendered as a delta, because showing our own mistake as a vendor rising or falling would launder it into a performance. The first one on this board cost the site owner's own commercial partner the top certificate score, which is the direction a correction has to be able to run in for the log to be worth anything.

The log starts on 28 August 2026, the cycle it was introduced. We do not backfill movements from before that date. We did not record them at the time, and a reconstructed ranking history is exactly the thing this log exists to prevent.

Rubric v2

Adopted 28 August 2026. Version 1 ran compressed into its top third, which left no headroom to record a vendor improving and made the difference between good documentation and excellent documentation unreadable. Version 2 re-anchors the midpoint at 3.0 and reserves 4.5 and above for evidence we have handled ourselves rather than read about. The entire board was restated against it on one day, each vendor carrying a dated recalibration entry, so that no vendor appeared to gain or lose ground because of a change we made to our own ruler.

Questions

How is the composite score calculated?: The composite is the plain arithmetic mean of the five rubric dimensions, rounded to one decimal place. All five are weighted equally. There is no hidden weighting, no bonus, and no adjustment applied after the mean is taken.
What decides the order vendors appear in?: Composite score, descending, and nothing else. Ties break alphabetically so the order is stable between builds. There is no pin, no boost, and no commercial tier in the ordering function. It is a single sort in a public file, and changing that behaviour would take a visible code change.
Can a vendor pay for a higher score or a better position?: No. No position on this board has ever been sold. Some outbound links on this site are commercial and that is disclosed at /legal/disclosure, but no commercial arrangement is readable by the scoring or ordering code.
Why did a vendor's score change?: Every score that moves carries a dated entry naming the evidence that moved it. Entries are one of four kinds: a listing, meaning the vendor entered the directory at that score; a recalibration, meaning the rubric scale changed and the whole board was restated on one date, which is not a gain or a loss for anyone; a movement, meaning the vendor itself changed and we saw the evidence; or a correction, meaning we scored something wrong and the vendor did not change at all. Only a movement is displayed as a delta. A correction is labelled as our error, because presenting our own mistake as a vendor rising or falling would be a second dishonesty on top of the first.
What is rubric v2?: The current scale, adopted 28 August 2026. Version 1 ran compressed into its top third, which left no headroom to record a vendor improving and made the gap between good documentation and excellent documentation unreadable. Version 2 re-anchors the midpoint at 3.0 and reserves 4.5 and above for evidence we have handled ourselves. Every vendor on the board was restated against v2 on the same day, each carrying a dated recalibration entry.
Why are some vendors marked provisional?: A provisional entry is scored from published documentation rather than from material we have handled. It is an honest desk assessment, the analytical dimension is capped while it stands, and the entry is re-scored once we have run a lot ourselves.
Do you sell peptides?: No. researchuseonlypeps holds no inventory, ships nothing, and fulfils no orders. We are a reference and review publisher. Everything listed here is for laboratory research use only and is not for human or veterinary use.

Read the rubric applied

The method above is only worth something where you can watch it operate.

The vendor board →Compare vendors side by side →The testing standard we hold vendors to →How to read a certificate of analysis →Lab results we have checked →

Think we have a score wrong? Tell us what we missed at editor@researchuseonlypeps.com. A correction with evidence attached moves a number faster than anything else on this page.