How we review products

Most supplement reviews online are rewritten sales copy. The reviewer never examined anything — they paraphrased the vendor's page, added a star rating, and attached an affiliate link. We are an affiliate too, so the only thing that makes our review worth more than that is method. Here is ours, in full, so you can judge whether it is worth anything.

What we actually do

1. Read the seller's own page, completely Every claim, the FAQ, the small print, the shipping terms, the refund policy. Not a summary of it — the actual page.
2. Separate the formula from the marketing What is measurably in the product: ingredients, doses, format, how you take it. This is fact and we record it as fact.
3. Check the citations If a page lists references, we look at them. Are they peer-reviewed papers or consumer blog posts? Does each one support the claim it sits next to? Do the volume and page numbers match the year given? This is where most supplement pages fall apart, and it takes about twenty minutes.
4. Separate strain-level evidence from product-level evidence Research on an ingredient is not research on a product containing it. A study on a strain at a stated dose tells you nothing about a blend that will not disclose its dose.
5. Name the pressure tactics Fake scarcity, struck-through anchor prices, unverifiable review counts, unnamed “doctor-formulated” credentials. These are not evidence and we call them what they are.
6. Write the drawbacks first If the cons list is shorter than the pros list, we have usually not looked hard enough.

What we do not do

• We do not lab-test anything. No purity assays, no CFU counts. We could not verify a label's accuracy and we never imply that we have.
• We do not accept free product, payment, or review copy from sellers. Our only commercial relationship with a vendor is the affiliate programme, disclosed on the page.
• We do not give numeric scores. A single number out of ten collapses “the strains are well studied” and “the dose is undisclosed” into one digit that hides both. It also produces a tidy star rating for search results, which is exactly why so many affiliate sites publish one. We would rather you read the drawbacks.
• We do not invent studies, statistics, endorsements or customer testimonials. If we cannot verify it, it does not appear.

Which sources count

Checking the citations only helps if we are honest about what we replace them with. The order below is fixed, and a page may not skip a level because the answer at a lower level is more interesting.

1. A Cochrane review, if one exists They state their own certainty in plain words, and that wording — “low-certainty evidence”, “may reduce” — is usually the most accurate sentence anyone has written on the question. We quote it rather than paraphrasing it upward.
2. Then the abstract of record PubMed, Europe PMC or PMC. Participant numbers, journal, volume and year come from the indexed record, so a mis-cited paper on a sales page cannot propagate into ours. This is how we caught the ProDentim reference list citing a 2018 paper's volume and page numbers under a 2012 date.
3. Then national guidance England's Delivering better oral health toolkit, or an equivalent, when the question is what a health service actually recommends rather than what one trial found.
4. Then a regulator's own opinion document Read at the regulator's site, not as reported by a brand. The EU cosmetics committee's opinion on nano-hydroxyapatite is on the page because it was opened there.
5. Never a claim that exists only on a page selling the thing If it cannot be traced past the seller, it is reported as the seller's claim, in those words, or not at all.

Where the full text was paywalled or behind a bot check, the page says which record we read instead, and the finding is given qualitatively rather than printing a figure we could not see with our own eyes.

Every study we rely on is linked from a References section at the end of the article, by DOI and PubMed ID.

Who paid, in the same sentence as the finding

This is the rule that changes the most pages. When we report a trial, the funder and the authors' declared conflicts go in the sentence that reports it — not in a footnote, not in a reference list at the bottom, where a reader has already formed a view before reaching them.

It applies symmetrically. Company-funded research is not automatically wrong, and independent research is not automatically right; the reader is simply entitled to know before they weigh it. In consumer dentistry an unusual share of the evidence is produced by the companies selling the product, and a systematic review is sometimes written by the authors of the trials it pools. Where that is the case we say so as a fact about the paper — the hydroxyapatite page is the clearest example — and we do not dress it up as an accusation.

When we get it wrong

We correct the page, we say on the page what changed, and we leave that admission in place. Nothing gets quietly edited and re-dated. A claim that turns out to be weaker than we made it sound counts as an error and gets the same treatment as a wrong number.

The most recent one is worth reading precisely because it is not flattering: our flossing page described the evidence as low quality, which was true and incomplete — across 35 trials and 3,929 adults, no trial measured decay between the teeth at all. The full list of corrections we have made is on the about page.

Who writes these

Reviews on this site are written by Jose Pollman. He is not a dentist, a hygienist, or medically qualified in any way, and nothing here is a substitute for seeing someone who is.

The method above is what stands in for that. It is a reading and checking discipline, not clinical judgement, and it has limits — we cannot examine a product, only the claims made for it and the sources offered in support. Where that is not enough to answer a question, the page says so rather than guessing.

The conflict of interest, stated directly

We earn a commission when someone buys through our link. That is a real incentive to be positive, and no policy makes it disappear. What we can do is make the incentive work against itself: a review that names what is unverified is more useful than one that does not, and a reader who trusts the page is worth more than one who bounces off it.

You do not have to take that on faith. Read our ProDentim review and check whether it reads like something written to sell you a bottle.