How a product gets onto this site, or does not.

The decision chain, in order

Every page on this site — the home page, the search, a comparison, a long read, a product page — applies the same sequence of steps. It is published here so you can check we never get round it.

1. Understand the person. The 15-question assessment, or what you write in the search bar. Nothing is transmitted: it all stays in your browser.
2. Understand the problem, not just the area. “I sleep badly” is not enough. Waking at 3am and taking an hour to fall asleep do not call for the same products. It is the declared symptom that decides, not the category.
3. Start with what is free. Every area has its no-purchase levers, ranked by effect and sourced. They appear before the products.
4. Rule out what has no business being there. A product only enters a list if it genuinely targets the need in question, or if a declared symptom points to it. Otherwise it does not appear, even with a good level of evidence.
5. Rank according to the profile. Age, sex, training volume, chronotype, diet, caffeine tolerance, contraindications. Two different people do not get the same list.
6. Take account of what you already have. A product you own keeps its real rank — a good product stays a good product — but it leaves the shopping list and we show you what complements it.
7. Adapt to the budget AND to how much the problem matters. You can be tight everywhere and willing to invest on one subject. The budget is asked per problem, not once and for all.
8. Say why, for you. Every product is explained in the context where you are reading it. Magnesium does not say the same thing in a sleep piece and in a stress piece.
9. Flag what makes the purchase premature. If a free action makes a product pointless until it is done, that is written on the product’s line, before the buy button.
10. Compare three options. Not three prices for the same product: three defensible choices, with what to check on each.
11. Explain how to use it. Dose, timing, relation to meals, to sleep and to exercise, how long before judging, and how to read the data the device produces.
12. Let the profile change. You can redo the test, change your answers, export your data or erase everything, at any time.

Filter 1: the three levels of evidence

This is the heart of the site, and the question we get asked most. A “level of evidence” answers one thing only: has the science really shown that this product does what it is said to do?

And here is the detail that changes everything: the rating is not about the product in general, it is about one specific promise. The same product can therefore carry two very different levels of evidence depending on what you expect of it.

Solid evidence

Several serious studies, done in humans, all pointing the same way. In practice, at least one of these two is needed:

• A meta-analysis. That is a study gathering every study already published on a question and calculating the average result. It is the highest level, because it corrects for the flukes of a single study: if twenty studies say the same thing, it is no longer luck.
• Or several randomised trials reaching the same conclusion. “Randomised” means participants were allocated at random: half receive the real product, the other half a copy with no active ingredient, and nobody knows which group they are in.
  Why go to all that trouble? Because of the placebo effect: when you believe you are taking something effective, you often feel better, even with an empty capsule. It is real, and it is measurable. Comparing the real product with the copy is the only way to know whether the effect comes from the product, or simply from believing in it.

You can rely on it.

Emerging evidence

Studies exist, but there are few of them, they are short, or they were done in small numbers. It is promising, but it is not established. Most supplements on the market sit in this category, including products we recommend.

Limited evidence

Little or no serious research, or results that contradict each other.

The obvious question then is: why keep a product on the site when we say ourselves that the evidence is weak?

Because there is a difference between “unproven” and “useless”. We keep it only if all three of the following hold at once:

1. It costs little. If it does not work, you have lost very little.
2. It cannot harm you. No known risk, no serious contraindication.
3. It genuinely does something, however modest, and even if it is not the effect the manufacturer advertises.

Take an evening chamomile tea. The evidence that it “improves sleep” is very weak: a few small trials, inconsistent results. But: it costs a few cents a cup (1), it cannot hurt you (2), and it forces ten minutes sitting quietly, without a screen, before bed — which is in itself a good ritual (3). All three conditions hold. We keep it, and we write plainly that the evidence is weak: the effect probably comes from the ritual, not from the plant.

Now, the limit. A product at this level will never come ahead of a product with solid evidence in a ranking. It does not matter if it is cheaper, better rated by customers, or more likeable. If you want to sleep better, low-dose melatonin (solid evidence on phase shift) will always appear before a 39 € “relaxation” herbal blend (limited evidence), even if the second earns us more. Evidence always comes first.

Across the whole site, every “Solid”, “Emerging” or “Limited” evidence badge is clickable and brings you back here. And the Sources page lists every study cited, one by one, with a link to open it.

Filter 2: is there enough active ingredient in it?

This is the filter that eliminates the most products. And it is also the hardest to check when you are not in the trade: it is all written on the label, but not legibly.

The principle is simple enough. A good ingredient, in an insufficient quantity, does absolutely nothing. Three traps come up constantly.

Trap 1: the big number is not the right number

A magnesium supplement never contains pure magnesium. Magnesium is always bound to another molecule to stay stable, a little like table salt is a compound of sodium and chlorine. The result: the label often shows the weight of the whole compound, not the weight of magnesium.

Indicative orders of magnitude, varying with the formulation.

What to look for on the packaging: the words “elemental magnesium”, or “of which magnesium”. It is the only figure that really counts.

For reference: an adult needs roughly 200 to 350 mg of elemental magnesium a day, food included. If a label only shows the total weight, you have no way of knowing whether you are taking 50 mg or 300 mg. That is exactly the point.

Trap 2: an ingredient that never gets into your body

Swallowing an active ingredient is not enough. It still has to cross from the gut into the blood. Curcumin is the perfect example: in its raw form it passes through the intestine and comes out almost entirely. You can swallow 1,000 mg and almost none of it will reach where it is supposed to act.

What to look for: a formulation statement on the packaging. “Phytosome”, “micellar”, or the addition of piperine (a black pepper extract). Those are three ways of forcing the passage into the blood.

Trap 3: the ingredient present in a laughable quantity

Some labels list fifteen ingredients. It looks impressive on paper. But if a study showed an effect with 600 mg of an ingredient and the product contains 15 mg, that ingredient serves one purpose only: appearing on the list. It will have no effect.

Why do brands do this? Because a long ingredient list sells better than a short one, and because every milligram costs money.

What to do: check how many capsules the stated figure corresponds to. It is written in very small print, often as “per daily dose of 3 capsules”.

Filter 3: who must not take it

Every product page lists its real contraindications, before the buy button.

The personalised assessment goes further. If you declare a pregnancy, anticoagulant medication, a kidney or thyroid problem or bipolar disorder, the site moves the products concerned down your ranking. And it shows you which ones it ruled out, with the reason. We would rather explain than hide something from you.

Filter 4: what it costs per day

The price on the label does not mean much. What counts is what one day of use costs you, at the dose that actually works. Two examples, with the sum:

The tub that costs more to buy works out fifteen times cheaper in use. And in this particular case, it is also the one with the better level of evidence. That is why we never rank by shelf price.

How we assess, in practice

For every product, we look at:

• The scientific studies on the claimed result. All our sources are listed on the Sources page, with a link to open them.
• The declared composition: real dose of the active ingredient, chemical form, additives, allergens.
• Independent certifications, where they exist. For example Creapure for creatine, IFOS for fish oils, Informed Sport for the absence of doping substances. These are simple criteria to check: a product has it, or it does not.
• Traceability: declared origin of the raw materials, whether there is a batch analysis.
• The legal framework: is the health claim printed on the packaging (“contributes to the reduction of tiredness”, “supports immunity”) even authorised in Europe? Far from all of them are. European regulation No 432/2012 gives the official list of those allowed to be written on packaging.

For a supplement that is the only valid method anyway. How one person felt after two weeks on a product proves nothing: they do not know what would have happened without it. Only studies on hundreds of people can answer that question.

The GX index, and where its percentages come from

Every product carries an index out of 100. It is not a mood rating: it is an average of four scores, calculated exactly the same way for every product. We say “weighted” because those four scores do not all count equally: some weigh more than others in the final result. The four scores and the full sum are shown on every page: you can redo it with a calculator.

The percentages are our choices. Here is why these ones:

• Level of evidence, 40 %. The heaviest weight, because it is the only criterion capable of cancelling everything else. If a product does nothing, its price and its tolerability do not matter.
• Dose and form, 25 %. Just behind, because an insufficient dose cancels the evidence. The most common trap in this aisle is a product that does contain the right ingredient, but ten times less than the studies used. The evidence exists, the product does not benefit from it.
• Effect per euro, 20 %. This criterion serves one purpose only: separating two products that work equally well. Between two equally effective products, the cheaper one comes first. Logical.
  But it only works in that direction. A cheap product that does nothing never climbs the ranking because of its price. Psyllium costs 2 € a week and it has solid evidence: it rises. A 2 € supplement with no evidence stays at the bottom, price notwithstanding. That is why this criterion weighs less than evidence and dose.
• Tolerability and risk, 15 %. The lightest weight, and that often surprises people. The explanation is simple: genuinely risky products are not rated, they are ruled out before they reach the scoring stage. So this criterion only separates shades of comfort, for example a product that causes bloating against one that does not.

This index contains no sales data, no customer rating, no commission. It is a structured judgement, published so it can be argued with. If you find these proportions questionable, write to us: that is exactly why they are public.

Our independence rules

• No sponsored content. No brand can buy a place in a ranking.
• No samples accepted in exchange for a favourable review.
• Every link to Amazon is marked as commercial. Technically it carries a tag invisible to you (rel="sponsored nofollow") but readable by Google, meaning “this link is paid”. Search engines require it, and it stops a site artificially inflating its ranking with paid links. You can check it yourself: right click on a “See on Amazon” button, then “Inspect”.
• Our best-rated products are among the cheapest in the catalogue, so the ones that earn us least. Psyllium at 15 €, vitamin B12 at 14 € and potassium-enriched salt at 6 € are among our best rated.
• A reported error is corrected, dated, and noted at the top of the page concerned.

What this method does not do

It does not replace a blood test, a diagnosis, or a prescription.

It does not predict your personal reaction. Two people can respond very differently to the same product, and that is the rule, not the exception.

And it does not have an answer for everything. On many questions the science is still too weak to recommend anything. In that case we recommend nothing, and we write it down.