The fine print is often the point. See our methodology →
Our
Methodology.
Every grade you see on this site is built from a defensible rubric. No opinions without evidence. Here's how it works.
The framework, visually
Four panels showing how a candidate source moves through the rubric to become a Tier S, A, or B citation, or get rejected outright.
1How a source moves through the system
01
Candidate source
Anything that could back a claim enters here — a trial, a review, a safety assessment, a regulatory notice. Nothing is trusted on its name alone; every source still has to earn its place.
02
5 gating checks
Authority. Methodology. Independence. Currency. Verifiability. Failing any one knocks it out.
03
Score (0 to 15)
Points added across the same 5 dimensions. Higher score, stronger evidence.
04
Tier S / A / B
12+ = S. 8-11 = A. 5-7 = B. Below 5, not shown.
05
Display on ingredient page
Sorted strongest-first. Tier tags are computed but hidden from end users.
2The 15-point scorecard(max points per dimension)
Authority
5 pts
Top score: Cochrane / WHO-JECFA / IARC / EFSA scientific opinion
Methodology
4 pts
Top score: PRISMA meta-analysis or systematic review
Independence
2 pts
Top score: No industry funding for the cited claim
Currency
2 pts
Top score: Published or updated within 5 years
Verifiability
2 pts
Top score: DOI or PMID that resolves to a working page
Total possible15 pts
3What the score buys you
S
Independent scientific evidence
12 to 15 pts
Cochrane reviews, systematic reviews & meta-analyses, PubMed-indexed & top-decile-journal trials (NEJM / BMJ / JAMA / Lancet), pharmacopoeia monographs, and WHO / JECFA / IARC & EFSA scientific opinions.
A
Government institute or clinical guideline
8 to 11 pts
NIH ODS fact sheets, ICMR or NIN reports, NICE / USPSTF clinical guidelines, AHRQ reviews, ISSN position stands.
B
Regulatory qualifier & supporting evidence
5 to 7 pts
National regulators (FSSAI, FDA, Health Canada, TGA) as a permitted / banned compliance check only; CSE India & Consumer VOICE investigations; Examine.com & DrugBank curation.
Not surfaceable
Below 5 pts
Not strong enough to back a user-facing claim. We may still note these in the audit log, but they don't appear on ingredient pages.
4Instant-veto conditions(any one knocks the source out, no matter how many points it would have scored)
Retracted
Cross-checked against Retraction Watch.
Predatory journal
Listed on Beall's List archive, Predatory Reports, or delisted from DOAJ.
Citation broken
DOI / PMID / URL does not resolve to a working page.
Manufacturer-employed authors
Every listed author works for the manufacturer of the product, without independent replication.
Methodology incompatible with claim
Single-arm trial backing an efficacy claim that needs controls, or a mechanism-only paper backing a clinical outcome.
Outside hard currency window
Over 15 years for safety opinions, over 25 for foundational sciences, unless explicitly overridden.
Evidence rating scale

Each documented effect is rated on a 4-point scale based on the quality and consistency of human-trial evidence:

Strong
Multiple synthesis-tier reviews (Cochrane, meta-analyses) converge on a positive finding. Effect is replicated across independent research groups.
Moderate
At least one good systematic review or meta-analysis with a consistent direction; multiple RCTs supporting the effect.
Limited
Some human-trial evidence with mixed results, or a single trial awaiting independent replication.
Insufficient
Mostly mechanism-only or animal-study evidence; no meaningful human-trial support.
Dose validation

We show the studied dosage range, the doses actually used in the trials that demonstrated the effect, rather than a generic intake figure. The amount that produces a benefit is often very different from the amount that merely avoids a shortfall, and that difference matters.

We also show the Upper Limit (UL) where one is established, so you can see the safe ceiling alongside the studied range.

Form quality matters

Two products listing "magnesium 400 mg" can deliver 40 mg or 280 mg of usable magnesium depending on the form (oxide vs glycinate). Our forms table flags absorption tier and surfaces the best-in-class option for every ingredient.

Source tiering

Behind every claim is a tier-graded source, and the tiers rank independent science above regulators. Tier S is independent peer-reviewed evidence: Cochrane reviews, systematic reviews and meta-analyses, peer-reviewed primary studies (PubMed and top-decile journals like NEJM, BMJ, JAMA and the Lancet), pharmacopoeia monographs, and the scientific risk assessments of WHO/JECFA, IARC and EFSA. Tier A is government nutrition institutes and clinical guidelines (NIH ODS, ICMR-NIN, NICE / USPSTF, AHRQ). Tier B is the supporting layer: reputable secondary curation, civil-society lab investigations, and national regulators used as a compliance qualifier.

The audit (internal) tracks every source against a 5-dimension rubric, Authority, Methodology, Independence, Currency, Verifiability, so we can't silently inflate a citation.

Regulators qualify, they don't grade

A food regulator declaring an ingredient permitted is a legal status, not proof that it works, or that it's safe at the dose printed on the label. Permitted-lists lag the science by years, and several regulators have documented gaps in enforcement and independence.

So FSSAI, the US FDA, Health Canada and the TGA feed our regulatory cross-check, is the ingredient banned, restricted, recalled, or permitted across jurisdictions, but they no longer lift an ingredient's evidence grade. They sit at Tier B as a qualifier; the trials and syntheses do the grading.

One distinction we do keep: the scientific assessment bodies, WHO/JECFA, IARC and EFSA's panels, sit at Tier S, because their carcinogenicity reviews and toxicology opinions are structured evidence synthesis, not market-permitting decisions, and they're often the best evidence that exists for long-term safety questions no trial can answer. We weigh them as synthesis, not as infallible: these panels can carry their own conflicts and have been known to read the same data differently, so a regulatory opinion informs a verdict, it doesn't end the argument.

What we don't do

No medical advice. Everything here is descriptive of the evaluation, not a recommendation to consume. Always consult a qualified healthcare professional before changing your supplement routine.

No brand affiliations. We don't sell, recommend, or affiliate-link any product or brand.