Methodology

This page explains where our figures come from, how we calculate, and how we score the products we compare. It is designed so that you can check our work, not so that we can ask you to take our word for it.

1. Where our nutritional data comes from

The values shown on our 1 020 food pages come, for the vast majority, from the Ciqual food composition table, published by Anses (the French food safety agency). We use the 2025 version, which covers 3,484 foods and 74 constituents.

Ciqual does not cover everything. We then apply a hierarchy, and each page shows the database its figures come from, with the exact line it is matched to. At 26 September 2026, across 1,023 pages:

We never mix two databases within the same page: a page matched to USDA takes all its values from USDA, including its vitamins and minerals. A food that neither Ciqual nor USDA measures does not get a value reconstructed from assumed ingredients.

Source reference, as Anses requests it: Anses. 2025. Table de composition nutritionnelle des aliments Ciqual.

What we do with these values

We display them as published, with no additional rounding and no added decimal places. A value missing from Ciqual stays blank on the page: we never show a zero in place of unknown data, and we do not fill gaps with estimates.

The values shown on our recipes have been calculated ingredient by ingredient from the Ciqual table since September 2026. For each line of the ingredient list, the quantity is converted into grams, then multiplied by the composition of the matching food page; the total is divided by the number of portions. Each recipe states its method.

Three conventions that we prefer to spell out:

• The calculation is based on the ingredients as listed. A recipe that says 200 g of dry rice is calculated on dry rice.
• Household measures use average weights. A tablespoon is 15 ml, converted according to the ingredient’s density: 14 g for an oil, 21 g for honey, 8 g for flour. A medium onion is 110 g, an egg 50 g. For foods sold whole, we remove the inedible part: the flesh makes up about 30% of the weight of mussels in their shells.
• Aromatics removed before serving are not counted. A bay leaf, a bouquet garni or a cinnamon stick add nothing to the plate.

What we do not calculate. Around forty recipes keep an estimate, and say so: either an ingredient does not exist in Ciqual, such as a curry paste or a seafood mix, or its fat melts during cooking, such as duck breast or bacon pieces, and the weight served is no longer that of the ingredient as weighed. Rather than publish a made-up figure, we keep the estimate and flag it.

These values remain orders of magnitude: a real onion weighs between 80 and 150 g. They are now derived from a public, verifiable source, which was not the case before.

Branded products and foods missing from the tables

Ciqual does not list branded products. For the few pages concerned, the values come from the manufacturer’s nutrition declaration, via the Open Food Facts database. The source is shown on the page.

When a food is missing from both public tables, we prefer to say so rather than cobble together a figure. Three pages are in this situation today: skyr, gochujang and coconut sugar. They show the order of magnitude of products sold in France, and their « source » field stays blank, which is our way of flagging the lowest level of reliability. We will replace them as soon as a public table measures them.

When no line describes the food exactly, we match it to the closest measured line and we write on the page which one, and what the gap costs. Salted butter caramel, for example, takes the values of « confiture de lait » (milk jam): the same base of milk and cooked sugar, a little less fat than a version enriched with lightly salted butter. We prefer a measured line and a stated gap to an invented figure.

It also happens that the table only holds the raw food when the page describes the cooked food. We then apply a cooking weight loss only if the table publishes the cooked line, whose composition is measured. Applying a weight loss to a raw composition would give a wrong result, because the weight is lost as water and fat, not in proportion to the energy.

2. How our calculators work

Our 18 calculators are based on published formulas, and every formula has its limits. Here are the main ones.

The calculation method is restated on each tool’s page, with the related frequently asked questions. These calculators give useful orders of magnitude to get your bearings: they replace neither a clinical measurement nor a professional’s advice. No data entered is collected or stored.

3. How we evaluate a product

Our supplement comparisons follow a four-step protocol.

Step 1: selecting the products

We choose the best-selling products, the most searched-for ones, and those recommended by health professionals. A comparison covers 5 to 10 products, at least two of them with no affiliate link. This is what ensures it reflects the market and not our list of partners.

Step 2: analysing the formula

This is the core of the work. For each product, we examine the full list of ingredients, the actual dose per serving, the form of the active ingredient used, and the presence of additives.

Each claim made by the manufacturer is checked against the literature. When a brand advertises a « clinically proven » dose, we check that the amount per serving really matches that of the studies cited. A gap between the two is common.

Step 3: practicality review

We assess how practical the product is day to day from its format, its directions for use and the information published about it: taste, solubility, digestive tolerance, format, ease of use. This is not a clinical trial and we do not claim to measure efficacy. Each article states clearly which criteria rest on the product’s technical data sheet alone.

Step 4: scoring and updating

Each product receives a score out of 20, calculated using the weighted grid below. The ranking follows from this score, and nothing else.

A comparison is revised when a product is reformulated, when a notable product arrives on the market, when a price changes significantly, or when a reader reports an error. The date of the last check is shown at the top of each article.

4. Our scoring grid

Weighted score out of 20. The weightings are adjusted according to the product category, and the version used is stated on each comparison.

The price per gram of active ingredient deserves a word. Comparing two supplements by the price of the bottle makes no sense if one is twice as concentrated as the other. We therefore systematically bring the price back to the quantity of active ingredient actually supplied, and that is the value shown in our tables.

5. Our sources

In order of priority:

• Primary studies and meta-analyses: PubMed, Cochrane
• Health agencies: Anses, EFSA, and their nutrivigilance opinions
• Drug interactions: Vidal
• Nutritional values: Anses Ciqual table (2025), then USDA FoodData Central (SR Legacy) for foods that Ciqual does not measure

Manufacturers’ technical data sheets are used to establish a composition, never to prove efficacy. Sources are linked directly in the body of each article, not tucked away in a bibliography at the bottom of the page.

6. Affiliation and independence

Some links on the site are affiliate links: we earn a commission on purchases made through these links, at no extra cost to you.

Three rules govern this practice.

• A product’s score is calculated before any commercial link is added.
• An affiliated product gets no bonus; a non-affiliated product is never left out.
• Each comparison includes products with no partnership, to remain representative of the market.

Affiliate links are identified as such in the code of our pages, and a transparency box appears at the top of each comparison and each review.

7. Report an error

If you spot a wrong value, a missing source or an out-of-date price, write to us. We check, we correct, and we update the date shown at the top of the article. Contact us.