Skip to the method

Home / How This Site Reads a Label

Methodology

How This Site Reads a Label

The procedure runs in a fixed order so a reader can rebuild any row from scratch, or throw it out with a reason.

This page publishes the method before it publishes anything read by it. The record date comes first, then the reading procedure, then the three-state rule that governs every cell. None of it depends on trusting this site, which is the point.

Answers: how does this site compare cjc-1295 sellers Results recorded 2026-09-01

The record date comes first

Every finding on this site is a reading of one search, pulled on one date. That date is 2026-09-01, and it is stamped on every page for a reason: a single pull is not a pattern across time, and this site does not pretend one search describes the market on any other day.

The stamp moves only when the set is recorded again. Until then, every count, every split and every state is a fact about 2026-09-01 and nothing else. This is the procedure behind the table this procedure produced.

The reading procedure

A title is read for its parts, not paraphrased for its meaning. The procedure is the same for every listing in the set:

  1. Record the domain as plain text. It is never linked, because the listing sits outside what this site verified.
  2. File the result type: research supply, marketplace, clinic service, community thread or video.
  3. Count the compound names the title claims.
  4. Note whether a status token is present or absent, without defining it.
  5. Note the name order and the connector, then set both aside as decoration.
  6. Record whether the listing text mentioned a certificate, in one of three states.

Titles are described structurally and never reproduced, because several of the recorded titles carry tokens this site does not publish. The one exception is a community thread title, which carries none and is quoted once elsewhere on the site.

The three-state evidence rule

Every evidence cell holds one of three states, and the three are never quietly collapsed into two:

  • Recorded: the listing text stated the thing on the day recorded.
  • Not recorded: the listing text did not state it on the day recorded.
  • Not applicable: the column does not apply to that kind of result.

Not recorded is never rendered as no. Silence in a search snippet is a tooling outcome, not evidence that a seller lacks a document, and reading it as a no would be exactly the kind of unsourced claim this site exists to flag. A blank stays a blank, and it carries a footnote rather than a verdict.

Whose criteria these are

The documentation criteria this site reads against are its own. They are not attributed to a named engine or a third party, because the pulls that shaped them have no public address a reader can open, and a criterion a reader cannot check is not worth citing to anyone.

There is a precedent for putting your own name in your own table rather than exempting it. The supplier this site links publishes a comparison page titled "Best Peptide Companies, Compared on What They Publish" that states plainly it is one of the companies in its own table and that it wrote the rubric. This site takes that as the model for scoring its own commercial link in public, and it does not reuse that page's five criteria, because this site reads label structure, which is a different axis. Read that page as the precedent it is at the supplier's comparison page.

What is deliberately left out

No amount is reproduced, and neither is a price, because a copied price is stale the week after it is copied and an amount is outside what this site publishes at all. Position in a table is not a score and cannot be bought. Rows sit in the order the results were recorded, which is mechanical, unpaid, and a verdict about nothing.

Who does the reading is set out at who reads these labels, and the commercial interest behind it is set out in full, because it is the commercial interest, stated.