Editorial standards and corrections

How a claim gets onto this site

Every substantive claim — sample size, effect size, p-value, confidence interval, dose, guideline grade, journal, year — is checked against the paper itself, not against a summary of the paper. Where a source cannot be retrieved, the claim is removed rather than softened or replaced with a plausible-looking figure.

Five specific things we look for, because they are the errors we have actually made:

• A non-significant result written up as a finding. If a confidence interval crosses the line of no effect, we say so in the sentence rather than in a footnote.
• A within-group change presented as a placebo-controlled one. “40% improvement” often means 40% against that group's own starting point, with the placebo group improving almost as much. We check which comparison a number belongs to.
• A conclusion running backwards from its source. We read the paper's own conclusion and check ours against it.
• Omitted nulls. If a trial's primary outcome was negative and a secondary one positive, both appear.
• Claims that flatter us. The category we are most suspicious of, because nobody inside a company argues against its own marketing.

What we will not write

• No disease claims. Supplements are not treatments. Nothing we sell diagnoses, treats, cures or prevents any condition, and we will not imply otherwise.
• No dose characterisations of doses we do not disclose. Two of our five products contain a proprietary blend whose individual amounts our manufacturer will not release. We will not describe those amounts as “modest”, “targeted” or “clinically dosed”, because we do not know them.
• No implied certifications. We have not published a certificate of analysis, named a testing laboratory, or enrolled in a third-party certification scheme. Until we can point at a document, we do not claim one. In September 2026 we removed testing claims from all five product pages for exactly this reason.
• No range-wide disclosure claims. Only Sleep Formula publishes every ingredient at its exact dose. We do not let that fact stand in for the whole range.

Corrections

In 2026 we audited every article on this site against its own sources. The audit found errors in most of them — roughly 170 corrections across the library. That is an uncomfortable number to publish and it is the honest one.

What we did with it:

• Corrections are visible, not silent. Where an error overstated a benefit or flattered us, the article carries a dated correction note explaining what it previously said and what was wrong with it. Eighty-two articles currently carry one. We do not quietly edit a page that readers may already have acted on.
• Unverifiable claims are removed, not rewritten. In one case we cited a meta-analysis that does not exist. We removed it and said so on the page rather than substituting a different study.
• We correct claims about other companies too. One article wrongly said a competitor hid its doses in a proprietary blend. Its label lists every amount. That correction is published on the article, with an apology.

If you think something here is wrong, tell us. Point at the claim and, if you have it, the source. We would rather be corrected by a reader than keep an error.

Clinical review: where we actually are

We want to be precise about this, because it would be easy to imply more than is true.

The articles on this site have not been reviewed by a clinician. They are written from primary sources and checked against them, but that is editorial verification, not clinical review. Where an article has been clinically reviewed, the reviewer's name appears on that article. Where no name appears, no clinical review has happened.

We hold back the articles where that distinction matters most. Several finished pieces on neurological and diagnosed mental-health topics are written but unpublished, waiting for a suitably qualified reviewer, and we would rather leave a gap on the site than publish clinical content nobody qualified has read. We are working on arranging that review.

Separately, and because it does not depend on that review to be useful, we publish a composition reference for clinicians: what is in each product and at what dose, the two products whose blend amounts our manufacturer will not release, the interactions we think matter most in practice, and a plain list of what we have not tested. It is a statement of composition rather than a clinical recommendation, which is why it can stand while the articles wait.

How we handle our own commercial interest

The clearest test of an evidence claim is what a company does when the evidence does not help it. Some examples currently published on this site:

• An article on sleep apnoea that tells readers no supplement will help them, including ours, and to get tested instead.
• A review of riboflavin and CoQ10 for migraine — both with real guideline support, neither of which we sell.
• An article on physical activity quoting a World Health Organization recommendation, graded strong, that polyunsaturated fatty acid supplements should not be recommended for cognitive decline. We sell fish oil. The recommendation is on the page.
• A comparison of exercise against our own products which concludes that if you can only do one thing, it should not be the one we sell.

We do not claim to be free of commercial bias. We claim to make it visible.

This page describes our editorial process. Nothing on this site is medical advice, and no article can assess your individual situation. Statements about our products have not been evaluated by the Food and Drug Administration, and our products are not intended to diagnose, treat, cure or prevent any disease. Speak to a doctor or pharmacist about symptoms, diagnosis and treatment, and before taking any supplement alongside prescribed medication.

Last updated 30 September 2026.