How We Review Supplements

Evidence first. Hype last. We start from the published human research and work outward to the product, rather than starting from a product and hunting for studies that flatter it. When the evidence is weak, we say so — including for supplements that are popular, well-reviewed and pleasant to take.

Each review is scored across five weighted components. The weighting is fixed across categories so that scores are comparable between a magnesium product and a fish oil.

• Evidence35% of score

  How good is the human research?

  We weight randomised controlled trials and systematic reviews above small pilot studies, and human studies above animal or cell work. Study size, duration, funding source and how closely the participants resemble adults over 50 all move this number.

• Formulation20% of score

  Does the product match the research?

  The right compound, in a form the body can absorb, at an amount close to what was actually studied. Underdosed 'proprietary blends' and unstudied novel forms lose points here.

• Testing & Transparency20% of score

  Can the label be trusted?

  Independent third-party verification, published certificates of analysis, full ingredient disclosure and clear manufacturing information. We record what a company publishes — we do not run our own laboratory tests and never imply that we do.

• Safety15% of score

  What could go wrong, and for whom?

  Documented side effects, upper intake limits, drug and supplement interactions, and specific cautions for kidney or liver disease, pregnancy, bleeding risk and upcoming surgery.

• Practical Value10% of score

  Is it worth the effort in real life?

  Servings per day, pill size and burden, taste, availability, and whether the realistic benefit justifies adding one more thing to a daily routine. We assess practicality, not cost — we publish no prices.

Separately from the product score, every claim carries an evidence verdict. These labels describe the state of human research, not our enthusiasm.

• Strong — consistent findings from multiple well-conducted randomised trials or high-quality systematic reviews.
• Promising — encouraging human trial results that are limited by size, duration or the populations studied.
• Mixed — trials disagree, effects are small, or results depend heavily on who was studied.
• Weak — little human evidence, or evidence that mostly fails to support the claim.
• Not Recommended — the claim is unsupported, contradicted, or the risk profile outweighs any plausible benefit.

• We prioritise randomised controlled trials, systematic reviews and meta-analyses in humans over animal, cell or observational work.
• We prefer research conducted in adults over 50 where it exists, and we say plainly when the participants were much younger than our readers.
• We never invent a citation. When a reference has not yet been checked line by line by an editor, it appears on the page as a clearly labelled citation placeholder until it is verified.
• We never invent a certification, a laboratory result or a brand fact. We report what a manufacturer publishes and describe it as such.
• We do not run our own laboratory tests and we never imply otherwise.

• We do not publish prices, price comparisons or checkout links, so nothing on this site is a sales page.
• We do not accept payment for a ranking, a score, a verdict or a mention.
• We do not use conspiratorial framing about medicine or the pharmaceutical industry. Supplements are compared against real clinical alternatives on their merits.
• We do not write in a way that could be mistaken for a diagnosis or a treatment plan. See our medical disclaimer.
• We do not claim that a supplement cures, treats or prevents disease.

Every review carries a “last reviewed” date. We revisit cornerstone reviews at least annually and sooner when a significant trial or safety signal is published. If you believe something here is wrong, tell us on the contact page — corrections are made openly and the review date is updated.

SupplementDecider does not sell supplements and accepts no payment for evidence ratings, rankings or verdicts.