How We Pick
Certification first. And a list of things we do not do.
Last updated September 1, 2026
The claim we refuse to make
We have not tested these filters. There is no lab here, no test bench, no rack of units we have been running for six months. Every publication in this category says otherwise about itself, and most of them are describing a level of testing that would cost more than the sites earn.
So we compete on something we can actually do and you can actually check. We read the certification listings, pull the manufacturer performance data sheets, cite the primary standards, and do the cost-per-gallon arithmetic in the open. Every number on this site is traceable to a source you can open.
Here is the honest scoreboard for that approach:
| What we claim | Count |
|---|---|
| Products we claim to have tested ourselves | 0 |
| Ratings we invented | 0 |
| Sponsored placements sold | 0 |
| Free products accepted from manufacturers | 0 |
| Products whose specifications carry a source link | 21 of 21 |
| Products whose certification status we state explicitly | 21 |
| Products holding a third-party certification listing | 15 |
| Pages published | 62 |
The first four rows are the ones that matter, and every one of them is a claim we are withholding rather than a capability we lack. Publishing a zero there is the point: everyone in this category says they tested twenty filters, we have tested none, and we say so on every page.
What we do instead
- Read the certification listing, not the box. A product is certified to a specific standard for specific contaminants at a specific rated capacity. We record all three, and we distinguish certified to from tested to every single time — see the standards guide.
- Pull published manufacturer specifications. Rated gallons, flow rates, ppm ceilings, pressure windows, dimensions. Every specification on this site comes from a manufacturer or certifier document, linked on the product.
- Cite primary sources for facts. EPA rulemaking, NSF standards scope, performance data sheets. Where a fact came from a secondary source, we name it.
- Do the arithmetic in public. Cost per gallon is filter price divided by rated gallons. We show the division rather than a verdict, so you can redo it with the price you actually see.
- Disqualify someone on every commercial page. A "skip this if" section is mandatory, not decorative. If we cannot name who should walk past a product, we do not understand it well enough to recommend it.
Prices, and why you may see none
Prices on this site render only from a live data layer refreshed on a schedule, and they carry the date they were verified. When no live price exists, the button reads "Check price" and no number is shown.
Right now, no products carry a live price. Our retailer data access has not been issued yet, so rather than print a number we remember or estimate, every buy button reads "Check price" and takes you to the retailer to see the real one. That is deliberate, and it is disclosed rather than hidden.
A remembered price is a wrong price. A price older than 48 hours disappears from this site automatically rather than sitting there going stale.
How a product gets on the site
- It has to be currently sold. Discontinued models get removed, not left up for traffic.
- Its specifications have to be published somewhere we can link. No source, no listing.
- Its certification status has to be stateable precisely, including "no certification listing published" where that is the truth.
- It has to be the right answer for somebody specific. We do not list products to fill a grid.
Certification claims are dated and re-checked
Manufacturers revise performance data sheets, and a certification that was accurate last year may not be accurate now. Every page carries a "certification claims verified" date, separate from the last-updated date, because a page can be edited for clarity without anybody re-reading a listing.
When we find we were wrong, it goes on the corrections log with a date, whether or not anybody noticed.
What money changes
Nothing about a recommendation. Some links on this site earn a commission and some do not, and where a product we rank first earns us nothing we say so on the page — the SpringWell CF1 and WS1 are both examples.
We do not accept payment for placement, we do not accept free products, and no manufacturer reviews anything before publication. The full detail is in the affiliate disclosure and the editorial policy.
What this approach cannot tell you
Being honest about the limits of the method matters as much as the method:
- Real-world performance on your specific water. Certification is tested under standard conditions. Your source water is not standard conditions.
- Long-term reliability. We have not run these units for three years, so we do not know which valve fails first.
- Taste. Genuinely subjective, and we say so rather than manufacturing a verdict.
- Build quality in the hand. Not assessable from a data sheet.
Where those matter to a decision, we say we do not know rather than filling the gap. A published methodology with holes in it is more useful than a confident review of something nobody examined.