How We Test and Choose Products

What has to be true before a product appears on this site.

Last updated: SEP 2026

Most product recommendations online are written by people who have never handled the product, ordered by which retailer pays the most. This page sets out what we do instead, so you can judge our recommendations against a standard rather than taking our word for them.

The short version

• We start from the evidence for the ingredient or mechanism, never from the product.
• We never accept payment, free products, or any other consideration in exchange for coverage.
• We check review authenticity, not just star ratings.
• Where we have used something ourselves, we say so. Where we have not, we say that too.
• We recommend the cheapest product that meets the standard, not the most expensive one we could justify.
• Every recommendation includes a real drawback. If we cannot find one, we have not looked hard enough.

Step one: the evidence comes first

Before we look at a single product, we establish what the research supports.

If an ingredient has no good evidence behind it, no product containing it appears on this site, regardless of how well it sells or how highly it is rated. A five-star collagen cream is still a product that cannot do what it claims.

This is why our product recommendations sit inside articles about evidence rather than in standalone buying guides. The recommendation is downstream of the research, and it always will be. If you ever find a product on this site without an accompanying explanation of why the underlying approach works, we have made a mistake and we would like to know.

Step two: the product has to match the evidence

Knowing what works is not the same as knowing what to buy. A product has to actually deliver the thing the research supports.

The right form. Where research is on a specific form of a compound, we recommend that form. Creatine monohydrate rather than creatine HCl, because monohydrate is what was studied.

The right concentration. Where trials used a particular strength, the product has to fall in that range. A 2% niacinamide serum is not what the 5% studies tested.

Stated quantities. Proprietary blends that hide how much of each ingredient is present do not qualify, because there is no way to know whether the evidenced amount is there.

No padding. Products that surround one evidenced ingredient with a dozen unevidenced ones are usually selling the list rather than the effect.

Step three: review verification

We use customer reviews as a signal of consistency and reliability. We do not trust star ratings on their own, and neither should you.

Review manipulation is widespread. Analysis cited when the review-checking tool Fakespot closed in 2025 found that around 43% of best-selling Amazon products had reviews judged unreliable or fabricated.

Those independent checking tools no longer exist. Fakespot shut down in July 2025 and ReviewMeta is no longer operating, so we verify manually. For every product we look at:

• The real positive percentage. We open the ratings breakdown and add the five and four-star percentages together. Our threshold is 85% or above. An average rating can conceal a large block of one-star reviews.
• Listing age. We check the Date First Available. A listing under twelve months old with thousands of reviews is a warning sign, not a recommendation.
• The negative reviews specifically. We read at least twenty one and two-star reviews looking for a recurring, specific complaint. A repeated fault matters far more than the average.
• Recent versus lifetime. We sort by most recent. If recent reviews are markedly worse than the lifetime average, the product has usually been reformulated or the seller has changed.
• Reviewer patterns. Clusters of five-star reviews posted within days of each other, from accounts with no other review history, are the standard signature of a manipulated listing.

A product can clear the rating threshold and still fail this step. Several have.

Step four: first-hand use, where we have it

We are explicit about the difference between products we have used and products we have selected on evidence and review data.

Where we have used something ourselves, the article says so directly, states how long for, and shows our own photographs.

Where we have not, the article says that instead, and the recommendation is described as selected on the evidence and on verified review data.

We will never describe a product as tested, trialled or used if it was not. Building a first-hand testing library takes time and money, and we would rather be slow than inaccurate.

Step five: the honest drawback

Every product recommendation on this site includes a section on what to know before buying, drawn from the product itself or from the pattern in its negative reviews.

This is not a formality. A recommendation with no downside attached is an advertisement, and we would rather tell you that a good product leaves grit in the bottom of the glass or stings a freshly shaved face than pretend it does not.

If we cannot find a genuine drawback, we have not read enough reviews.

What we will not do

We will not accept payment to feature, review or rank any product.

We will not accept free products in exchange for coverage.

We will not let commission rates influence selection or ordering. Rates play no part in either.

We will not recommend a product whose underlying ingredient has no evidence behind it, however well it sells.

We will not recommend something we would not use ourselves or suggest to a friend.

We will not describe something as tested unless it genuinely was.

How we make money

Some links on this site are affiliate links. If you buy through one, we may earn a commission at no additional cost to you.

Every article containing affiliate links says so clearly at the top of the page, before any link appears. That income pays for hosting, for access to research, and for the products we buy to test.

It does not influence what we recommend. The clearest evidence of that is how often our recommendation is the cheapest option available, and how often we tell readers that a product category is not worth buying at all.

Our Editorial Standards page covers the wider question of how we research and correct everything else on the site.

When we get it wrong

Formulations change, products are discontinued, sellers change hands, and new evidence arrives. We review product recommendations at least every twelve months, and sooner when something significant changes. Each recommendation shows the month it was last checked.

If we recommend something and it turns out to be poor, tell us. We will change it and say why, with a dated note explaining what changed.

Questions or corrections about any recommendation on this site can go to our contact page. We read every one.

Anup Shrestha
Editor, All About Him