Home / Why Cupel

What makes a certificate worth trusting

Most peptide certificates cannot be checked by the person holding them. This is what one that can be looks like beside one that cannot, and the nine questions that separate them.

A certificate you can check, beside one you cannot

The one on the left is a composite of what certificates in this sector tend to look like, not any one laboratory’s. Each numbered mark is something a buyer cannot check, explained below.

A typical certificate · composite

LAB LOGOtell 5

Certificate of Analysis

Product Semaglutide 5 mg

Lot 2026-1012tell 1

Purity 99.2%tell 2

tell 3

Date 10/12/2026 Analyst J.S.tell 4

To verify this certificate, contact the seller.tell 7

nothing heretell 6 tell 8

Ours

What you receive

A certificate of analysis, published under its own number

The accession number.Assigned when your sample is logged, with a check digit that catches a mistyped number.
The sample as received.Your lot reference and the received and issued dates and times, so a buyer can match the report to the vial.
Identity and purity, with the specification column.Each charge state within 10 ppm of what the formula predicts; purity from three injections, every peak at or above 0.05% named.
The evidence.The chromatogram and the mass spectrum, the method parameters, and what could not be resolved, in a row of its own.
Two signatures.The analyst and the reviewer, named, with the date each signed.
Two hashes and a QR code.Anyone can check the report at its address without contacting us.

The page itself, element by element: how to read a certificate.

No number that resolves to anything Either no accession at all, or one that is just a counter. Nothing to type in anywhere. Ours carries a number with a check digit and a public lookup that answers for it.
A purity figure with nothing under it 99.2%, and no chromatogram, no retention times, no replicate injections. A number with no data is an assertion. Ours prints the trace, the retention times, all three injections and the RSD they produce.
A chart with no axes A small picture of a peak, no scale, no units, no labels. It cannot be read, only recognised. Ours is drawn with axes and units, and every peak is named by its retention time to a hundredth of a minute.
Initials where a signature belongs “J.S.” Nobody is accountable for the number. Ours names the analyst who ran it and the reviewer who released it, each with the date they signed.
An accreditor’s logo with no scope The badge is real; the scope behind it covers something else. Ours shows no badge, says so, and publishes the scope table instead.
No hash, no way to detect tampering A PDF that can be edited and re-sent. Ours prints a SHA-256 of the report’s data and publishes the data and the report itself beside it, so the figures on any PDF can be checked against the published record.
Verification means asking the seller “Contact us to verify.” The party with the most reason to say yes is the only one who can. Ours verifies at a public URL and by QR, with no login and no email.
Nothing about what could not be resolved Every certificate reads as if the method saw everything. Ours has an “Open point” row on every certificate, filled in even when the answer is that nothing is outstanding, so that a peak we could not resolve has a place where it must be said.

Nine questions to ask any laboratory

Put them to us and to anyone else who offers to test a peptide. These are our answers; each is checkable on this site.

Nine questions to ask any laboratory, with our answers.
Ask any laboratory	Our answer
Are your prices published and fixed?	Yes, per sample, on one page
Can I read your scope document, the list of analytes your methods are validated for?	Yes, as a table, per analyte, with the validation status of each; today none is validated, and each row says so
Can I verify a report without an account?	Yes: paste the number, no login
Does a mistyped accession number get caught before it is looked up?	Yes: a check digit catches a wrong or swapped digit
Is the report’s hash published somewhere I can compare it against?	Yes: printed on the report, with the data it hashes published beside it
If you get a result wrong, does the original stay published?	Yes: retracted, linked to its correction, kept at its address
Can the customer who paid for a low result make it disappear?	No: the register refuses a deletion, and a removal would show at its next integrity check
Do you state what your methods cannot determine?	Yes: eight limitations, written out
What happens if you miss your turnaround?	The invoice is cancelled, or the payment refunded, and you still get the results

Built into the register

Five commitments. The first two are built into the register itself and the third into every report, rather than into a policy that could be revised quietly; the last two are undertakings, written here so that a change would have to be made in public.

We do not delete reports, for ourselves or for the customer who paid for one. The register is built to refuse a deletion and to show one at its next integrity check; a result you dislike is published the same way as one you like.
Corrections are issued as new reports. The superseded report stays reachable at its original address, marked retracted, linking forward to what replaced it.
Every report carries a SHA-256 of its own data, printed on the certificate, and the data is published beside it. Recomputing the hash shows whether the published data is the data the number was issued with; the certificate file itself is checked on the verify page against the fingerprint the register recorded when it issued, and the figures on any paper against the published report. It does not vouch for us: we publish the report, the data and the hash, so it cannot show that we measured correctly.
Measurement is our only line of business: we undertake to hold no inventory, to take no affiliate payment and no reseller margin, and to say so on this page if that ever changes. That also means nothing subsidises the price list, so we cannot undercut a laboratory that sells what it tests.
We do not accept payment for a favourable result, and we do not offer vendors an embargo period, a pre-review, or a chance to withdraw a sample after the data exist.

How to read one of ours, element by element, is on the reading page. The method behind every report, column to source voltage, is published on the methods page; what it cannot determine is on the limits page; the accreditation we do not hold is stated in the footer of every page.