Open document
Methodology
This page explains exactly where every number on this site comes from. If something cannot be explained here, it should not be in the ranking.
01 We compare in the unit it is sold in
Each category is compared by price per kilo if it is sold as a powder and by price per capsule if it is sold as softgels or tablets. The two units are never mixed in the same table: €30/kg and €0.07/capsule are not comparable numbers, and putting them in the same column would be a made-up ranking. A product whose page does not say how much the pack contains does not enter the comparison. Vitamin D is the exception: it is scored per 1,000 IU, because a 50,000 IU softgel and a 400 IU one are not the same capsule; one whose name does not state its IU only competes with others that do not.
The price is the one a buyer pays today, not the crossed-out one, and it comes from the store page on the day the data was collected. Flavour variants of the same product are grouped into one row so the table does not show the same thing eight times.
And on top of that, where possible, the cost per effective dose. Price per kilo compares packs, not what you take home: a tub that is cheap per kilo with half the active ingredient per serving is not cheap, it is half a tub. When the store publishes how much active ingredient each dose carries, we also work out how many effective doses the pack gives according to the reference dose cited, and that figure appears on the product page. It does not move the ranking, because most stores do not publish that data and ranking with it would leave half the table out; but when it is there, it is what really separates two tubs that cost the same per kilo.
That is why whey concentrate and whey isolate are two separate categories and not one: a kilo of concentrate carries about 750 g of protein and a kilo of isolate about 880, so comparing them in the same table by price per kilo would always put the concentrate ahead for carrying more filler. In the cost per dose on the product page that difference is discounted, with the typical purity of each category cited in the dose table.
No data, no mark. If the store does not publish how much active ingredient each dose carries, the product appears with its price and a zero: we assign it no typical dose and assume no formula. Multivitamins go entirely this way, because there is no single reference dose for twenty micronutrients at once.
02 The four verification levels
| Level | What it means | Weight in quality |
|---|---|---|
| 4 Third-party certification | The certification is backed by someone other than the brand. Either we checked it with the body that issues it (Creapure QS code, batch in the Informed Sport list), or the product carries in its name a trademark that cannot be used without a third party behind it: Creapure, which requires a licence agreement with the maker of the active ingredient, or IFOS, which tests the batch in an independent lab. Each product page shows which of the two it is and on what date. | 100 % |
| 3 Lab report published by the brand | The brand itself publishes a lab report. It is provided by the interested party, so it is worth less than an independent third party. | 85 % |
| 2 Claimed on the page | The certification appears on the product page or the label, with no way to check it. | 70 % |
| 1 No certification | No certification of any kind on record. | 60 % |
Level 4 comes from two things, and the product page says which. The first is a check of ours against the body that issues the certification: the QS code on the pack at creapure.com, or the batch in Informed Sport’s public list. The second is that the product carries Creapure or IFOS in its name. Creapure is a registered trademark of Alzchem and only a licensee can put it in a product name. IFOS is Nutrasource’s batch testing programme: an independent lab tests them and publishes the report for each batch, so it is not a logo you just add either. Neither is decorative: they are claims that legally bind the brand and that the store signs off on by listing them in its catalogue. We only compare stores that answer for what they publish, so we treat it as what it is, a third-party certification, and not as a loose logo printed on the label.
What it does not mean. That the name states Creapure does not mean we have held that tub in our hands or typed in its QS code: that can only be done product by product, with the physical pack in front of you. On each product page you can see exactly where its level 4 comes from, with the date and the link to the proof, and if it comes from a checked QS code it says so in those words. And it only counts if it is in the name: the long description of a product page shows "Creapure" from other products in the same store, and that claims nothing about this one.
Why level 3 is worth less than level 4. A lab report published by the brand itself is commissioned and paid for by the interested party: it picks the lab, picks the batch and picks whether to publish the result. A Creapure QS code or a batch in the Informed Sport list are issued and maintained by a third party the brand does not pay to approve, and anyone can look them up. We are not saying a brand lab report is false: we are saying it proves less.
Why level 2 is so low. A certification printed on the label or mentioned on the store page proves nothing: there have been cases of logos used with no certification behind them. A certification only rises to level 4 if we have checked it against the body that issues it, and the date of that check is recorded.
03 The formula
The formula
score = 35 % · cost+ 35 % · quality+ 20 % · requirements+ 10 % · reviews
The final score runs from 0 to 100 and is split like this:
Verification level x chemical form of the active ingredient x actual composition x how well the doses match (the last two only when the store publishes them; if not, nothing is claimed about the formula). The studied form scores 100 %, an alternative form 75 % and an undeclared form 60 %.
The price per kilo (or per capsule) compared with the cheapest in its category, which takes 100 %. A product with no declared pack size gets none of this part.
Of its category’s requirements that could be checked on its page, how many it meets. It is the question neither price nor certification answers: is this what it says it is?
The rating buyers give it at the store that sells it, normalised to 5 (some stores rate out of 10). It weighs little on purpose: they are unverified reviews moderated by the store itself. And it is damped with the category average, as if each product started with 10 borrowed reviews: that way a single five-star review does not overtake a product with hundreds. A product with no reviews counts as the average: not publishing them is not a fault.
Underdosing. A key ingredient present below 60 % of its minimum effective dose marks the product as underdosed and multiplies its quality by 0.5. It is the penalty for fairy dusting: putting a famous ingredient in token amounts just to be able to list it on the label.
All of these weights live in a single configuration file in the project, and this page is generated from that same file: they cannot say something different from what the calculation does.
04 What each tub really contains
Two proteins at the same price per kilo are not the same product. When the page publishes its nutrition table, we read how much of the tub is active ingredient: if it declares 82 g of protein per 100 g of powder, those are the grams you pay for, and the cost per effective dose is computed with that figure and not with the category average. A composition better than the typical one raises quality by up to 15 % and a worse one lowers it by up to 25 %.
From the ingredient list on the label we also read the additives: artificial sweeteners, colourings, fillers and anti-caking agents. Each one takes 4 % off quality, and between them they cannot take off more than 12 %. They are neither illegal nor dangerous: they are what separates a clean label from one that fills and colours.
What the store does not publish does not count against it. A product whose page carries no nutrition table and no ingredient list is scored as before, with the typical purity of its category and no penalty at all. Punishing it for what its store does not write would be making the data up.
05 Each category’s requirements
Price per kilo does not tell a kilo of creatine from a kilo of creatine with a third of maltodextrin in it: the second comes out cheaper and delivers less. Certification does not see it either, because it certifies there are no doping agents, not that there is no filler. Requirements are the third question, and they are worth 20 % of the mark.
Each category has its own and they are public: they are on every product page, one by one, with a yes or a no and the reason. A whey has to carry at least 70 g of protein per 100 g, carry no added free amino acids and no cheaper proteins blended in, and not be cut with flours. A magnesium has to come in a form that is absorbed and not as oxide. A probiotic has to identify its strains and declare its CFU. A pre-workout cannot hide the doses in a proprietary blend.
A requirement that cannot be checked does not score. "No filler" can only be claimed by reading the ingredient list on the label: that it does not appear in the product name proves nothing, and taking it for granted would reward the store that publishes least. The mark is "of the ones that could be judged, how many it meets", and a product where none could be judged gets its category average: neither reward nor penalty.
06 Where the reference doses come from
Every figure has its source cited and is checked by hand. No dose is used without a published reference.
Who checks this. Nobody with a medical qualification. These figures are checked against their source by the same person who writes the site, and they have no healthcare qualification. That is why each dose carries the DOI of the work it comes from: so you do not have to trust whoever copied it, but can read the original. You will not find a "reviewed by a professional" badge here until there is a real professional behind it, with a name and a registration number.
07 The recommendation badges
A badge from this site certifies an editorial criterion: that a product meets an objective, published condition. It promises no effect on your body, your performance or your health, and it should not be read that way.
| Badge | Earned when |
|---|---|
| Verified level 4 | Its certification is backed by a third party: checked against the body that issues it, or a licensed ingredient trademark declared in the product name. With a date and evidence you can check on the product page. |
| Best value | It has the highest score in its category and beats 70 out of 100. If the best in a category does not reach that threshold, that category goes without a badge: it is not handed out out of obligation. |
Both thresholds come from the same configuration file that does the scoring, so the answer to "why does this product have the badge" is always its score breakdown, visible on its page. There are no negotiated or sponsored badges.
08 Affiliate links
Some store links are affiliate links and may earn a commission. Affiliate links are applied after the ranking is computed, in a file the scoring engine does not read. A product with no affiliate programme is linked just the same and takes exactly the place its score gives it. There is an automated test that swaps the affiliate links and checks the order does not move.
09 What we do not do
- We do not say a product improves anything. Effects belong to the ingredient at the studied dose, with its citation. A product only "contains X grams of Y".
- We do not charge for position. No commercial deal moves a product in the ranking.
- We do not mark a certification as verified without having checked it at its source.
- We do not make up data the store does not publish: if the dose per serving is missing, we say it is missing and the product scores worse for it.
10 Known limitations
- Prices change daily and promotions distort the ranking. The collection date is in the footer of every page and on every product page.
- Some stores block automatic data collection. They are not forced: they simply do not appear, and that is stated in the project repository.
- Verifying Creapure requires the QS code on the physical pack, so it can only be done product by product and by hand.
Data last updated: 2026-10-15. The weights on this page are read from the same configuration file that computes the ranking.