Verified Supplement Data

Editorial standards

We try to make every claim checkable. No brand pays for a pick. Health claims link to their source, product numbers come from the label, and testing claims are checked against the tester’s own list. When we get something wrong, we fix it and log it.

By Erin Rose Updated October 11, 2026 Who writes this

Run by Erin Rose. Who writes and checks our pages →

1. How we make money and stay independent

We earn a commission when you buy through our Amazon links. As an Amazon Associate, we earn from qualifying purchases. We are not paid by supplement brands. No brand pays for a ranking, a pick, a review or a mention.

  • Picks come from our five-check score: dose, independent testing, cost per dose, label clarity and recalls. The rules are public so you can check them.
  • Our footer discloses the affiliate links on every page.
  • We often recommend the cheaper product, or say the evidence does not support buying at all, when that is what the data shows.

2. What we cite

We use the strongest evidence we can find, in this order:

  1. Reviews that pool many trials, such as Cochrane reviews and meta-analyses of randomized trials.
  2. Single randomized trials, when no review exists.
  3. Large population studies, such as CDC NHANES, to say who is low in a nutrient, not whether a supplement works.
  4. Government and testing bodies: NIH Office of Dietary Supplements, FDA, USP and NSF.

We do not cite marketing claims, stories, or “studies show” without naming the study. Where the evidence is weak or mixed, we say so.

3. How we check a claim

For a product recommendation, we check each step against its own source:

  1. The dose. We find the dose used in the strongest trials on PubMed and record the study.
  2. What the product delivers. We check the active amount per serving on the label, such as elemental magnesium, not the weight of magnesium glycinate.
  3. Testing. We look for the exact product in the tester’s own public list and record the date. If we cannot find it, we say the brand claims it rather than calling it tested.
  4. Cost. We work out the cost of the studied dose, so a cheap bottle with a small dose does not win on price alone.

4. AI and automation

We use software, including AI tools, to collect labels, registry listings and prices, do the math, and draft pages. Automated checks run before the site ships: every buy link must resolve to an Amazon product, and pages in our current design must pass our layout and wording checks.

We also check every PubMed ID on the site against PubMed. The last full check, on October 11, 2026, found all 1,319 IDs were real, and we fixed 2 wrong author names. Our data is published as a free, citable dataset.

5. When we get something wrong

  • When a published fact turns out to be wrong, we fix it and log it on our corrections page with the date. We have logged 27 so far. Routine price updates are not logged. We never backdate or invent entries.
  • Every page footer has a “Report it” link. We aim to review reports within five business days.
  • We print the real date our data was last reviewed. See how fresh our data is. The price at checkout always wins.

6. Medical advice and expert review

We are not doctors, and we will never put a clinician’s name on a page they did not review. Pages about a dose, a drug interaction or a health condition must trace to a review, a trial or official guidance. They point you to that source and do not replace your doctor or pharmacist.

This is health information, not medical advice. Talk to a qualified professional before starting, stopping or changing a supplement, especially if you take medicine, have kidney disease, or are pregnant or nursing.

Think we got something wrong? Email [email protected]. Confirmed errors are fixed and logged publicly.