Readable evidence.
Visible limitations.

Identify the exact product

We distinguish the brand, cartridge, reference number and the system in which it is used. Two products from the same brand may have different compositions and performance.

A fluoride add-on, a membrane version or a new generation does not automatically transfer its results to the main product. Certification tied to a specific tank is not extended to all compatible tanks.

Collect and date the sources

We consult manufacturer pages and instructions, public laboratory reports, and accreditation or certification registers. A document hosted by a retailer may supplement the evidence; its origin is then stated.

Each product page lists the links used and the access date. Detailed results link to their own sources. A summary is identified as a summary, even if it bears a laboratory’s name.

When a selection of documents forms the basis of a product page, that scope is stated. Information absent from the selected sources is marked “not documented”; this does not prove that the product is ineffective.

Distinguish four types of information

Manufacturer claim

A specification or reduction is advertised without a usable result confidently attributable to this model in the documents reviewed.

Summary table

The manufacturer publishes figures. They are retained as published data, with their volume and limitations, without being presented as a new independent check.

Test report

The document allows a sample, laboratory, method and results to be examined. Its existence does not, by itself, certify the product or its entire service life.

Product certification

We look for a specific model and performance in an official register, accounting for configuration and restrictions. Materials certification remains distinct from reduction certification.

Examine the four transparency criteria

Each criterion is worth a maximum of 2.5 points. The same weighting applies to every product.

The laboratory is assessed on documented, relevant accreditation, not its name alone. A current scope does not retrospectively validate every earlier test: dates and methods must be cross-checked. Quel Filtre performs no factory audits.

Calculate a comparison score out of 20

The fixed score adds two components: documentation transparency out of 10 and contaminant coverage out of 10. Visitor-selected filters change the product selection, never the score.

Documentation transparency: 10 points

We add the points for the four criteria above, each worth 2.5. The breakdown and reasons appear under “Understanding this assessment” on the product page.

Contaminant coverage: 10 points

The comparator uses a fixed list of 25 parameters or groups, including parasites. Each entry earns 0.4 points when a result is attributable to the model, has a cited source and has one of these statuses: public test report, manufacturer-published table or reduction in the official listing.

Claims alone, uncertain or limited-scope results, references to another model, non-isolated aggregate results and tests without a reduction measurement earn no points. Materials certification alone earns no coverage points. Evidence quality is assessed separately in the 10 transparency points.

Each entry counts once, regardless of the number of molecules named in its group. Groups are not all mutually exclusive: PFAS, PFOA and PFOS are three separate entries in this rubric. The selector’s seven visual groups help navigation; they are not each worth one seventh of the score. Any change to the list of 25 entries requires a revision of the rubric.

Coverage is not a filtration rate. A low reduction and a high reduction can each count as one entry. A start-up result is not equivalent to a result after several thousand litres. The score does not weight contaminants by their hazard and guarantees neither complete removal nor safe drinking water. Rates, organisms, volumes and reservations must be read in the detailed results.

Zero coverage points mean that no result is credited for that entry under this rubric; they do not prove ineffectiveness. Entirely unestablished documentation remains “not assessable”. Scores may change with the sources; individual points are available on each product page.

Read the figures with their conditions

• Substance or organism: we retain its name. A lead test does not describe all metals; an MS2 test is not a direct measurement for rotavirus.
• Volume: a result at start-up, after 5 L or after 3,000 L is presented with that volume. Advertised capacity remains distinct from the volume tested for each parameter.
• Stages: a minimum over the cycle and a final value are different pieces of information. We do not merge them into a uniform reduction rate.
• Conditions: inlet concentration, pH, flow rate, configuration, ageing and any surrogates are stated when available. Different protocols limit direct comparisons.
• Quantification limit: “less than” does not mean absolute absence. We avoid percentages more precise than the available concentrations support.
• Discrepancies: a calculation inconsistency, conflicting version or unidentified model is flagged. We do not invent missing conditions to reconcile documents.

Compare by group, then by parameter

These seven groups make searching easier. “Chemicals” includes chlorine, PFAS and nitrates, while other groups highlight common needs. The groups are therefore not a mutually exclusive chemical classification.

If several groups are selected, a product must have at least one accepted published result in each group. This does not mean that all their components are filtered. A claim alone or unestablished attribution is generally insufficient for inclusion.

For viruses, choosing “Rotavirus” or “MS2” limits the selection to the recorded organism. A published result with reservations may remain accessible, with an explicit status. Parasite tests and additional parameters are available in the product details.

Understand the colours and crosses

Selector icons represent categories only. Result colours appear on cards, product details and comparisons. Each group is clickable in product details and the comparison table to access figures, volumes and sources.

Write a review, then revise the documentation

Each review describes what the evidence establishes, its strengths, its limitations and the questions to check for a particular use. We distinguish transparent evidence from measured performance; the highest overall score is not automatically the best choice for every contaminant.

Updates are made when checks take place and documents are added. The access date is stated; it does not imply continuous monitoring of manufacturers or registers. Unverified prices remain labelled as such.

Quel Filtre reviews documents: we perform neither physical product testing nor manufacturing audits. Private working documents are not published; only the cited public sources are accessible from the site.