Methodology
Every score is calculated from verifiable label data. This page explains exactly how — down to every point.
Overview
Every supplement in our database (116,409+ products) receives a SupplementScore from 0 to 100. The score is split into 4 categories, each based on data from the product's physical label as recorded in the NIH Dietary Supplement Label Database (DSLD).
No brand can pay for a higher score. The algorithm is deterministic — same label data always produces the same score.
1. Dosage Transparency (0–35 points)
Measures how transparently the product discloses its ingredient doses. This does not evaluate whether doses are clinically effective — it evaluates whether you can see what you're getting.
How points are awarded:
Data source: supplementFacts[].amount and supplementFacts[].dailyValuePercent from the product's Supplement Facts panel.
2. Label Transparency (0–25 points)
Measures how completely the product discloses its contents. Proprietary blends that hide individual doses are penalized.
How points are awarded:
Data source: supplementFacts, servingsPerContainer, servingSizes, targetGroups from the label.
3. Formulation Cleanliness (0–25 points)
Starts at 25 points (full score) and deducts for each unnecessary additive found in the "Other Ingredients" list.
Deductions:
Bonuses:
Data source: otheringredients[].name and targetGroups from the label.
4. Ingredient Profile (0–15 points)
Evaluates quality signals from the active ingredient list.
Formula focus:
Branded ingredients (up to +5 pts):
+3 points for each clinically studied, patented ingredient form detected in the name. These indicate the manufacturer invested in verified, bioavailable forms:
KSM-66 · Sensoril · Albion · Quatrefolic · BioPerine · Meriva · Longvida · Cognizin · Suntheanine · Creapure · Carnipure · Ferrochel · TRAACS · K2Vital · Aquamin
Penalty:
Data source: supplementFacts[].ingredientName and supplementFacts.length from the label.
Evidence Levels
On product pages, we display research summaries for individual ingredients. Each claim is assigned an evidence level based on the quality and consistency of published research. These levels reflect the state of evidence for the ingredient in general — not for any specific product or dose.
Evidence levels are not part of the SupplementScore calculation. They are displayed separately as informational context about the ingredients a product contains.
What We Don't Score
We intentionally exclude categories where we lack reliable, verifiable data:
- Third-party certifications (USP, NSF) — DSLD doesn't include this. We'd rather omit than guess.
- FDA safety record — we display FDA adverse events separately, but automated brand-matching isn't reliable enough for scoring.
- Price / value — DSLD has no pricing data. We'll integrate this when retailer data is available.
- User reviews — not enough reviews yet for statistical significance.
When we add reliable data for these, we'll expand scoring. Honesty over completeness.
Data Source
All scoring data comes from the NIH Dietary Supplement Label Database (DSLD) — a U.S. government database containing 116,409+ product labels with full Supplement Facts, Other Ingredients, serving information, and dietary certifications. Data is public domain (CC0 license).
Each product in DSLD was entered from the physical label by NIH staff. This is not scraped, estimated, or AI-generated data — it's what's printed on the bottle.