How we test our data before it reaches a page

One fact, one source

Every number on a data page is stored with the source it came from: a trial registry entry, a journal article with its PMID, a drug label or a regulator's notice. Pages are generated from that store, so a number cannot appear on a page without its source travelling with it. The numbered source list at the bottom of each page is built from the same records.

Our methodology lists the source types we accept and the ones we do not, such as vendor product pages and certificates of analysis that cannot be traced to a lab.

Unknowns stay visible

When we do not have a sourced value, for example a half-life by route or a storage time after mixing, the page shows a dashed [pending] chip instead of a number. We would rather show a gap than a plausible guess. Many peptide pages are full of them today, and that is deliberate: it tells you exactly how much is known.

The chips are also how we measure progress. Each page has a completeness score based on how many of its planned facts are filled and verified.

The publish gate

Every page has an editorial status (draft, review or published) and a completeness score from 0 to 1. A page is offered to search engines only when it is published and at least 80% complete. Until then it still exists, so links to it work, but it carries a noindex tag and is left out of our sitemaps.

This is our guard against publishing hundreds of thin pages. The site grows only as fast as facts can be verified.

How calculator math is tested

All calculators and converters share one small library for reconstitution math, unit conversion and blend splits. That library has automated tests that run on every change, covering worked examples, unit conversions, U-100 and U-40 syringe scales and edge cases such as a draw larger than the syringe.

These tests are the only calculator check we can point to today. We have not recorded an independent rebuild of the calculators by a second person, so we do not claim one. Try the reconstitution calculator or the mcg to mg converter.

Literature and quote checks

Evidence summaries start from PubMed. We query NCBI's E-utilities, record each search with its query, date and result count, and cache every abstract we use. Trial registry entries come from the ClinicalTrials.gov API and drug labels from DailyMed, also cached.

A verification script then checks every quoted statement. Each quote must appear word for word in the cached abstract, label or registry record it points to, and every number in our one-line restatement must appear in that quote, so no figure is derived or invented. Every study we cite must also have its abstract cached. The check exits with an error if any of these do not hold.

Evidence badges follow a fixed rule, not a judgement call. A randomized or controlled trial in people maps to "human RCT"; other human studies map to "human observational"; animal work maps to "animal"; cell work maps to "in vitro". Reviews and meta-analyses are counted but never used to set a grade.

Regulatory data

Legal status changes fast, so every regulatory entry is a dated event with a link to the primary notice: an FDA page, a Federal Register entry, an advisory committee record or the WADA Prohibited List. Status labels on peptide pages are derived from those events, not written by hand. The full list is in the legal tracker.

Search-interest data

The trend chips you see on peptide cards (for example, "+172% searches" on retatrutide) come from US monthly search volumes in Google Keyword Planner (pulled in October 2026), comparing June to August 2026 with the same months of 2025. Some peptides show no data because the keyword tool suppresses them; we leave those blank rather than model a number. Search interest measures curiosity, not safety or effectiveness.

Community reports

Readers want honest user reports, including "it did nothing". The site has an Anecdotal evidence level for them, and the rule is that such reports carry that badge and never support a dosing figure. The site does not collect or display community reports today, and no launch date has been set.

Lab tests of products

We do not commission or run lab tests of peptide products. If a page shows third-party test results, it links to the original report, shows the test date and the lab, and does not restate the result as our finding. We never publish accusations about a named seller, and sellers can reply to anything we publish about them.

Dates you can trust

"Last reviewed" at the top of a page means our medical reviewer checked the whole page against its sources on that date. "Updated" means content changed. Sitemaps carry the real review date, never the build date. If you find an error, report it; confirmed corrections are dated and listed publicly.