We read 18,699 supplier certificates. Here is what we found.
Online peptide shops post lab certificates (COAs) as proof of quality. For most buyers, the certificate is the only evidence they ever see. We searched the websites of the 518 shops we track. 401 of them post a certificate we could read. With shops we tracked earlier, we read every certificate from 416 shops. Almost all report purity. Few report contamination.
Certificates collected 23-24 September and 7-8 October 2026. Lab checks September to October 2026, on the certificates collected in September. Product listings checked 8 October 2026.
The findings in brief
Each finding has its own chart, base and method. Select one to jump to it, or copy its link to share it.
- Only about 1 in 3 certificates show any contamination test
- Almost half of shops show no contamination test on any certificate
- Almost 8 in 10 UK shops show no contamination test
- Semaglutide certificates are the least likely to show a contamination test
- We could match a contamination certificate from the seller to only about 1 in 9 products for sale
- Only 1 in 3 posted certificates could be found on the lab's own system
- Certificates made out to an unrelated company are rare: we found 64 in 11,231
- 9 in 10 certificates claim 99% purity or more
- Nearly 1 in 5 certificates was more than a year old
Only about 1 in 3 certificates show any contamination test
Share of 18,699 readable certificates that show each kind of test. Contamination shares are floors.
- Purity, identity or amount94.1%
- Any contamination test33.9%
- Endotoxin (bacterial toxins)28.4%
- Sterility or microbial count16.4%
- Heavy metals14.4%
- All three contamination tests9.2%
Almost every certificate reports purity: 94.1% name a purity, identity or amount test. Only 33.9% show a result for bacterial toxins, microbes or heavy metals. That is a floor: our reader misses some results hidden in images. 62.9% show purity and nothing else.
Endotoxin is the most common contamination test, on 28.4% of certificates. Only 9.2% show all three contamination tests.
Purity and safety are not the same thing. A purity test measures how much of the powder is the intended peptide. It does not look for bacteria, the toxins bacteria leave behind, or metals. A very pure vial can still carry them.
How we counted
A certificate counts if it shows a result for endotoxin, sterility or microbial count, or heavy metals. Disclaimers such as "not tested" and offers such as "available on request" do not count.
This is a floor. Some certificate designs hide test names in images. 2.2% name a contamination test but show no readable result. If every one of those had a hidden result, the share would be 36.2%.
Counting each certificate once, even when several shops post it, gives the same picture: 33.0% of 16,658 unique certificates. Some labs issue the endotoxin result as a separate document, so we also counted per product: of 5,680 cases where a shop posts a certificate for a peptide, 40.4% include at least one contamination certificate for that peptide.
Almost half of shops show no contamination test on any certificate
416 shops that post at least one readable certificate. Each square is one shop.
205 of 416 shops post certificates, but none of them shows a contamination test. Among shops with at least five certificates, it is 144 of 337 (42.7%).
Testing is the shop's choice. The same labs offer both purity-only and contamination testing. At the biggest labs that do both, 67.0% of shop and lab pairs are all or nothing: the shop orders contamination tests on under 10% or over 90% of its certificates.
How we counted
A shop counts here if no certificate we collected from its website shows a contamination result. A shop may test more than it posts.
The all-or-nothing figure looks at each shop's certificates from each lab, for shops with five or more certificates at labs with 300 or more certificates that issue both kinds. That gives 342 shop and lab pairs.
Almost 8 in 10 UK shops show no contamination test
Shops with no contamination test on any certificate, by shop country
- UK (91 of 116 shops)78.4%
- EU (18 of 27 shops, small sample)66.7%
- All shops (205 of 416)49.3%
- US (63 of 205 shops)30.7%
91 of 116 UK shops show no contamination test on any certificate they post. For US shops it is 63 of 205 (30.7%).
The gap holds when we count only shops with five or more certificates: 70.5% of 78 UK shops against 28.6% of 182 US shops.
Counted per certificate the gap is smaller: 26.3% of UK certificates show a contamination test, against 37.2% of US ones. A few UK shops test thoroughly. Most show no contamination test at all.
How we counted
Shop country comes from our supplier records. The EU sample is small (27 shops). Shops in other countries, or with no known country, are in the total but not shown as their own bar.
Semaglutide certificates are the least likely to show a contamination test
Share of certificates naming each peptide that show a contamination test. The 12 peptides with the most certificates.
- KPV (713 certificates)40.7%
- NAD+ (588 certificates)39.8%
- MOTS-c (753 certificates)37.8%
- Tesamorelin (775 certificates)37.0%
- GHK-Cu (1,406 certificates)36.1%
- TB-500 (1,587 certificates)34.2%
- BPC-157 (2,049 certificates)33.8%
- Retatrutide (1,695 certificates)33.4%
- Ipamorelin (950 certificates)33.2%
- CJC-1295 (882 certificates)30.8%
- Tirzepatide (1,117 certificates)28.8%
- Semaglutide (647 certificates)16.7%
Only 16.7% of semaglutide certificates show a contamination test, and 14.8% an endotoxin test.
The other popular weight-loss peptides show one more often. Retatrutide certificates show a contamination test 33.4% of the time and tirzepatide 28.8%. For all other peptides it is 34.8%.
How we counted
We matched each certificate to the peptides it names. A certificate that names several peptides counts for each of them.
On the September certificates, an independent reviewer recounted this figure with a different peptide matcher and got a similar result. It is not driven by a single shop: leaving out any one of the shops with the most semaglutide certificates changes it by less than one point.
We could match a contamination certificate from the seller to only about 1 in 9 products for sale
6,308 peptides listed for sale by 429 shops (each shop's listing counted once), checked 8 October 2026
- Has any certificate from the sellerat least 32.2%
- Has a contamination certificate from the sellerat least 11.6%
A shop's certificate page can look full while the product you are looking at has none. Of 6,308 peptides listed for sale, we could match 32.2% to a certificate from the shop selling them, and only 11.6% to a contamination certificate from that shop.
Both figures are floors: the real shares are at least this high, and may be higher where a shop's certificate covers a blend, or names the product in a way we could not match.
How we counted
We count a product as covered only by a certificate from the same shop, for a single peptide, that names a lab and tests the peptide itself. Blends are left out. Certificates made out to an unrelated company, or with a report number that is not a possible date, do not count.
Products are live, priced listings in our database on 8 October 2026. Certificates were collected 23-24 September and 7-8 October 2026. Shops change their listings daily, so this figure moves a little from day to day. The certificates added by our October re-scan have not been through the client check yet.
Only 1 in 3 posted certificates could be found on the lab's own system
11,231 certificate postings (one certificate on one shop's website) put through our lab check
- Found on the lab's own system33.8%
- Has a link or QR code our automated check could not confirm26.1%
- Names a lab, but no link or QR code32.1%
- Names no independent lab we recognise7.9%
Some labs keep an online register, or host their own copy of each report, so anyone can check it. We found 33.8% of certificate postings there. That is a floor, for the reason below.
Another 26.1% carry a link or QR code our automated check could not confirm. Many lab portals only work in a web browser, so this does not mean something is wrong: in a small hand-checked sample, 4 of 8 were on the lab's system after all. For 40.0%, there was nothing to look up online.
Where we could compare the shop's copy with the lab's own copy, they almost always matched: the client name was the same on 1,133 of 1,136.
How we counted
We followed each certificate's link or QR code to the lab's register or lab-hosted copy. For one large lab whose reports carry a search code, we looked up each code on the lab's own server. Most of the copy comparisons come from that lab.
The found share is a floor: some certificates we could not confirm are on a lab's system. Three labs issue 59.1% of the certificates that name a lab, so a buyer's chance of being able to check depends a lot on which lab the shop uses.
The 4,731 certificates added by our October re-scan have not been through this check yet, so they are not in this figure.
Certificates made out to an unrelated company are rare: we found 64 in 11,231
Every posting where the client named on the certificate was not the shop's trading name, checked by hand against the shop's own website. Number of postings.
- The shop's own company, under its legal name151
- A brand or partner the shop names openly62
- Another company, and we found no link to the shop64
Each certificate names the client who paid for the test. Most cases where that was not the shop had a simple explanation. That left 64 postings (0.6% of 11,231) at 8 shops where the certificate is made out to a different company and we found no link between the two. Such a certificate may be a real lab report, but it does not show that the vial for sale came from the batch that was tested.
Two warning signs were rarer still. In 3 cases, the lab's own copy of a report names a different client from the copy the shop posted. And 36 files from a single shop carry report numbers that cannot be real. On this lab's reports the number starts with the date the sample arrived, and these numbers are not possible dates. None of the 1,838 certificates the lab itself holds has such a number. None of the 30 we looked up is on the lab's server: the lab has no record of them.
We do not name shops on this page. We will give shops and labs a chance to respond before naming anyone.
How we counted
Our software flags a posting when the client on the certificate is not the shop. Those flags overstate the problem: 76.6% had a simple explanation once we checked the shop's website. So we only report the hand-checked figure.
The report-number check covered 2,587 certificates carrying one lab's branding. The 36 files carry 27 different report numbers.
The 4,731 certificates added by our October re-scan have not been through this check yet, so they are not in this figure.
9 in 10 certificates claim 99% purity or more
Claimed purity on the 13,062 certificates that state one
- 99.9% to 100%13.1%
- 99.5% to under 99.9%50.0%
- 99% to under 99.5%27.4%
- 98.5% to under 99%3.9%
- 98% to under 98.5%2.7%
- 97% to under 98%0.8%
- 95% to under 97%0.9%
- 90% to under 95%0.6%
- Under 90%0.6%
90.5% of certificates that state a purity claim 99% or more. The median claim is 99.65%. Only 2.9% report less than 98%.
High purity is good news. But when nearly every certificate says the same thing, the purity number stops telling shops apart. The tests that differ between shops are the contamination tests.
How we counted
Purity is what the certificate claims. We did not test any product ourselves. Some labs write a ceiling, such as "more than 99.8%"; we read those as the ceiling value.
Nearly 1 in 5 certificates was more than a year old
Age of the 15,693 dated certificates on the day we downloaded them
- Up to 3 months26.3%
- 3 to 6 months24.4%
- 6 to 12 months29.8%
- 1 to 2 years13.5%
- Over 2 years6.0%
Most certificates are recent: the median was 5.9 months old when we downloaded it. But 19.5% were more than a year old, and 6.0% more than two years old.
A certificate describes the batch that was tested, which may not be the batch on sale today. The older it is, the more likely that is.
How we counted
Age is the download date minus the date printed on the certificate. 15,693 of 18,699 certificates had a readable date.
What this means for buyers
A short checklist for reading any certificate. A strong certificate does all six.
- It names the independent lab that did the test.
- It names the shop as the client, or a supplier the shop names openly.
- It can be looked up on the lab's website, through a link, QR code or register number. Use it.
- It reports identity and amount, not only purity.
- It includes an endotoxin result, and ideally sterility and heavy metals. If it only shows purity, it says nothing about contamination.
- It is for the exact product and batch on sale, and it is recent.
Judge the shop, not the lab logo: the same labs issue both purity-only and full reports, and the shop chooses which to order. Our guide to reading a certificate explains each part in more detail.
How we did it
- Collection. Our crawler saved every file that might be a certificate from the websites of 551 research-peptide shops, some of which we no longer track: 32,169 distinct files, collected 23-24 September and 7-8 October 2026. In October we searched again the shops where September gave us no readable certificate, this time opening product pages too. That added 4,731 certificates and 177 shops.
- Reading. We extracted the text of every file, using OCR for images. 18,699 files are readable lab certificates from 416 shops. The rest are product photos, labels or web pages, or have no readable text, and are not counted.
- Classifying tests. A rule-based classifier records which tests each certificate reports and whether it shows a result. We checked it by hand on 55 certificates; it got 2 wrong.
- Checking with labs. For each certificate posting we followed its link or QR code to the lab's register or lab-hosted copy, and compared the two copies where we could.
- Hand checks. We read 54 certificates across every lab-check outcome, and checked every case where the client was not the shop against the shop's own website.
- Products. We matched certificates to the peptides they name, and compared that with what each shop lists for sale in our price database.
- Independent recount. On the September certificates, a second script, and then an independent reviewer with their own script, recomputed the headline figures by a different route. They agreed. With the October certificates added, the headline figures match a separate recount made during the re-scan.
Bases
Test findings count each certificate file once (18,699). Lab-check and client findings count each certificate on each shop's website (11,231 postings checked), because “was this issued to this shop?” only makes sense per shop. Shop findings count the 416 shops with at least one certificate.
Limits
- Shops with no readable certificate are not in the shop findings. 117 of the 518 shops we track gave us no certificate we could read. Some post none. Some post only images our reader could not read, or keep certificates behind a login or send them on request. Shop findings use the 416 shops with at least one readable certificate.
- The October re-scan is not fully checked yet. In a hand check of 20 random files it added, 1 was not a certificate, so a small share of the added files may not be certificates. Its certificates have not been through the lab check or the client check, so those findings use the September certificates only.
- Only what shops publish. A shop may test more than it posts. “No contamination test” means none in the certificates we collected.
- Many figures are floors, not exact counts. Our reader misses some results hidden in images, so contamination and endotoxin shares are floors. So are the lab-check share and the product shares.
- Not found is not a finding against a shop. Many labs have no public lookup, and some portals only work in a browser.
- “Sterility” is a broad group. It includes true sterility tests, PCR screens and microbial counts.
- Small samples. The EU figures rest on 27 shops. Shop country comes from our supplier records.
- A snapshot. Shops change their pages often. We did not buy or test any product for this analysis.
Download the data
Every aggregate figure on this page, with its base. No shop, lab or file names.
Cite as: Peptide Scanner, “We read 18,699 supplier certificates. Here is what we found.”, October 2026, peptidescanner.co/reports/coa-findings-2026/, with the finding's link.
Full report and comment
Journalists and researchers can ask for the full report, the data definitions behind every figure, or a comment. Email media@peptidescanner.co.
Peptide Scanner compares the prices and certificates of research-peptide shops. It sells nothing and takes no orders. See also our disclaimer and review policy.