How the scores work
Every number on this site comes from the data printed on the vendor page and the formula on this one. There are no manual adjustments and no term you cannot recompute by hand.
The standard
A single sample is judged pass or fail. It passes only when all of the following hold. There is no partial credit, because a vial that is pure but half empty is still a bad vial.
-
99.5%
Minimum purity by HPLC
Area percent of the main peak at the lab’s standard wavelength.
-
±25%
Fill accuracy
Measured quantity divided by the label claim must sit between 0.75 and 1.25.
-
Identity
Confirmed by mass spectrometry
The observed mass has to match the compound on the label.
-
Lot
Certificate matches the vial
Scored under trust rather than per sample, but a sample with no lot on the vial cannot be matched to any certificate.
Purity score
The purity score is the share of a vendor’s samples that passed the standard, weighted by how recent each result is. A result loses half of its weight every 240 days, so a failure from last winter matters less than a failure from last month.
Vendors with only a handful of results are pulled toward the board-wide pass rate. We add 2 virtual results at the baseline rate to every vendor history. One lucky result cannot produce a perfect score, and one bad result cannot produce a zero.
Product leaderboards list the vendors with independent results for that compound, ordered by overall score, with the pass counts and average purity for the compound shown alongside.
weight(result) = 0.5 ^ (age_in_days / 240)
pass_rate = sum(weight * passed) / sum(weight)
purity_score = 100 * (sum(weight * passed) + k * baseline) / (sum(weight) + k)
where k = 2 Trust score
Trust is everything you can check before paying. Each item is worth a fixed number of points and the total is out of 100. Review platforms are shown on vendor pages but do not score, because reviews can be bought and lab reports cannot.
| Check | Points |
|---|---|
| Certificates readable before checkout (a free account wall still counts; email on request does not) | 20 |
| Lot number on the certificate matches the vial | 20 |
| Testing lab named on the certificate | 12 |
| Certificate can be re-checked on the lab’s own site | 13 |
| Mass spectrometry identity, not HPLC alone | 10 |
| Trading history, 2 points per year | up to 10 |
| No public enforcement actions | 15 |
Delivery score
Measured as the median number of business days from order to doorstep for US addresses. One day scores 100 and every additional day costs nine points, with a floor of 20.
| Median business days | Score |
|---|---|
| 1 | 100 |
| 2 | 91 |
| 3 | 82 |
| 4 | 73 |
| 5 | 64 |
| 7 | 46 |
| 10 | 20 |
Value score
Every listed vial is converted to dollars per milligram. For each product we take the median across all vendors, then express each vendor’s price as a ratio to that median. The vendor’s value score is based on the median of its ratios, so one expensive niche product does not drag down a vendor that is cheap on everything else.
A vendor priced 30% below the market median scores 100. A vendor priced at the median scores 70. Value is weighted least of the four pillars on purpose. A cheap vial that fails the standard is not a bargain.
ratio = median over products of (vendor $/mg / market median $/mg)
value_score = clamp(100 - (ratio - 0.70) * 100, 15, 100) Overall score
The four pillar scores are combined with fixed weights. Purity carries the most weight because it is the only pillar measured in a lab. Trust is next because it is what lets you check the vendor yourself.
overall = 0.40 * purity + 0.25 * trust + 0.20 * delivery + 0.15 * value Bands: 85 and above is strong, 70 to 84.9 is solid, 55 to 69.9 is mixed, below 55 is weak.
What is left out
- Vendor-supplied certificates. They are linked on vendor pages for reference but never count toward purity.
- Customer reviews and social media sentiment. Shown, not scored.
- Catalog size. A vendor with eight well made products outranks one with eighty inconsistent ones.
- Results with undocumented provenance. If we cannot show where a sample came from, it is not on the site.
Worked example
Atomik Labz currently sits at the top of the board. Here is its score rebuilt from the four pillars shown on its page.
| Pillar | Score | Weight | Contribution |
|---|---|---|---|
| Purity | 98.4 | 0.40 | 39.36 |
| Trust | 100 | 0.25 | 25.00 |
| Delivery | 100 | 0.20 | 20.00 |
| Value | 70 | 0.15 | 10.50 |
| Overall | 94.9 |
Data status
This site is running on preview data.
Vendor names, domains, and the operational facts we could confirm from public sources are real. The lab results, delivery medians, and review counts that drive the scores are placeholder values used while our first round of testing is completed. Nothing on the board should be read as a measurement of any vendor until this notice is gone.
Snapshot Oct 27, 2026. 20 vendors, 12 products, 160 results.