The verifiability gap: 1 in 3 peptide lab reports has no link to the lab’s own copy

A lab report is only as good as your ability to check it. A PDF on a store’s product page is a copy, and copies can be edited, reused from another batch, or simply out of date. The strongest reports carry a link or code that opens the same report on the laboratory’s own website, so you are reading the lab’s record rather than the seller’s.

So we counted. For every report a store currently posts, we asked three things: does it point to the lab’s own copy, how old is it, and is it still there a few weeks later?

Of the 13,923 reports stores currently post, 9,255 (66.5%) carry a link hosted by the laboratory. The other 4,668 rely on the store’s copy alone. Which lab signed the report matters a great deal: some labs give every report a public lookup page, others offer none.

The market leans on a handful of labs. Freedom Diagnostics, Janoshik and Vanguard Laboratory sign 58.3% of all current reports between them.

A missing link is not always a dead end. When we checked 100 reports by hand, about 58% of the unlinked ones still printed a lab lookup code or QR code on the document itself (every Janoshik and Freedom Diagnostics report we read does). Reports with neither a link nor a code came to about 14% (11% to 17%). Those are the ones a buyer genuinely cannot check.

Most stores do better than the average report suggests: 67 of 129 link at least 90% of their reports to the lab. But 11 stores have no lab-hosted link on any report the tracker records. That is not the same as uncheckable: in our hand check, several of those stores print a lab lookup code on the document or link the lab elsewhere on the page.

Between July 14, 2026 and October 5, 2026, 3,631 of the 17,622 reports seen on store websites (20.6%) vanished from the address where they were found. Some of that is housekeeping: 1,237 of them (34%) are the identical file, or the same report number from the same lab, still posted by the same store somewhere else. The tracker counts 2,394 more (13.6%) as gone, but it over-counts: in our hand check, most of those were still on the store’s page as a re-uploaded file. Our corrected estimate is that about 5% of reports truly disappear (3% to 9%), mostly replaced by a newer lot or removed with a discontinued product.

The longest-watched group shows the trend best. Of the 1,214 reports first seen in July, only 10% are still where they were found and 63% are gone. That group is small and one store accounts for 49% of its removals, so read it as a direction rather than a rate.

A report describes the sample that went to the lab on the day it was tested. The older it is, the less it can tell you about stock on the shelf today. Half of the dated reports were issued between 84 and 263 days before we pulled the data, and 1,516 (12.6%) are more than a year old.

Most reports do name a batch: 93.3% carry a batch or lot identifier. Whether that batch is the one a store is shipping today is something no public dataset can show.

The tracker marks a link as verified when it resolves at the lab. To check that the links work for a real visitor, we picked 50 current reports with a lab link at random and opened each one from our own server on October 5, 2026.

44 opened at the lab and 6 could not be reached from our server (6 of them Janoshik, whose site turns away automated requests from our server). We record those as unchecked, not as broken. Of the links we could reach, 100% opened. Opening is not the same as matching: we did not compare what the lab shows with the store’s copy.

NuuVive sells research peptides, so we have a stake in this question. Here is our own row, scored the same way as of the same day. It is kept out of every industry figure above.

Where we do well: every peptide we list links to its report on Janoshik’s own site. Where we don’t: our reports are older than the median, and they describe the supplier’s tested material, not the specific lot in your box. We are not in the tracker’s dataset, so the “still posted” measure is not scored for us. Our reports are on the lab reports page.

Hand check

Before publishing we checked 100 records by hand (2026-10-05), sampled at random across labs and statuses, against the original documents and lab lookup pages. The tracker named the right lab 95% of the time and the report date within a week 99% of the time, and its link field matched what the store page shows 99% of the time. Two of its labels needed correcting, and the figures above use our corrected versions: many unlinked reports still print a lab lookup code, and most reports it marks as disappeared were re-uploaded rather than removed.

Data source

All industry figures come from the PeptideBenchmark public COA API, which catalogues the lab reports peptide stores publish and tracks each one over time. We pulled its complete default feed (17,622 records: 13,923 current, 68 retained, 3,631 disappeared) on October 5, 2026, data generation 8818f98fc78fb197297eba28. PeptideBenchmark first saw these reports between July 14, 2026 and October 5, 2026. We thank PeptideBenchmark for publishing it openly.

“Public read-only evidence data. Preserve record IDs, PeptideBenchmark permalinks, provenance labels, and source attribution when republishing.”PeptideBenchmark API reuse policy, as published in the API on the day of the pull

Our CSV keeps PeptideBenchmark’s record ID and permalink on every row, as that policy asks. Store names, store web addresses, batch numbers and report numbers are left out; each store gets a code (V001 and so on) that is assigned fresh in each edition.

Definitions

• Current report: a report still posted where PeptideBenchmark found it on the day of the pull (13,923).
• Lab link: the report carries at least one link hosted by the laboratory (its lookup page or file). Matched to the lab means PeptideBenchmark confirmed the link belongs to the lab’s own lookup (61.8% of current reports). Neither means anyone compared the content at the lab with the store’s copy.
• Report age: days from the date printed on the report (or, if missing, its certificate, test or sample-received date) to the day of the pull. 86.6% of current reports have a readable date. Month-only dates count as the 15th; dates before 2015 or after the pull are treated as unreadable.
• Gone and moved: a report is disappeared when it is no longer at the address where it was found. It counts as moved when the same store still posts the identical file (same SHA-256 fingerprint) or the same report number from the same lab (one listing for one missing report, and only when the number looks like a real report number); otherwise it counts as gone.
• Store: one of the 129 vendors PeptideBenchmark tracks.

Limitations

• These are the reports stores chose to publish, from the stores PeptideBenchmark chose to track. They are not a random sample of the market. PeptideBenchmark discloses an affiliate relationship with some of the stores it tracks.
• Fields such as the lab, dates and batch were read from PDFs and images by PeptideBenchmark’s software. We did not re-read the documents by hand for this edition, so small extraction errors will be in the figures.
• A report that was re-saved with different bytes and a new number would count as gone, not moved, so “gone” can overstate removals. A page the tracker failed to reach can also look like a removal.
• Half of all reports were first seen on or after September 29, 2026, so most have been watched only briefly. The persistence figures will firm up with each monthly edition. The single store with the most removals accounts for 18% of all reports counted as gone.
• Our link spot-check used one server on one day. Janoshik turned away automated requests from that server, so its links could not be opened there; we recorded them as unchecked, not as broken.

Download the full data (CSV, 17,622 rows). One row per report, with the fields behind every figure on this page, so each number can be recomputed.

You are welcome to reuse our analysis and charts under CC BY 4.0: credit “NuuVive, The verifiability gap” with a link to this page, and credit PeptideBenchmark as the source of the underlying records, keeping their record IDs and permalinks.

Updated monthly. We re-pull the data on the 1st of each month and this page updates itself. This is the October 2026 edition, pulled October 5, 2026.

Spotted an error? Email support@nuuvive.com and we will correct it and note the change here.