Methodology

Here's the short version: you can recompute any score we publish, because nothing is hidden. Each sub-score is just the sum of visible score-events - one rule firing on one fact we can check - and each event carries an evidence tier and a plain-English line you can read right on the scorecard. AI does the legwork of reading spec sheets and certification databases for us; it never decides the score. That's the part you can trust to be consistent.

Evidence tiers

Every fact behind a score is tagged with how solid it is, so you can weigh it yourself. Strongest at the top:

A - independent certification (NSF, WQA, IAPMO, AHAM, CARB, Energy Star) or an accredited-lab test. This is the only tier that earns certification credit - someone outside the company checked the work.
B - a published manufacturer spec or a recognized industry standard. Useful, but it's the maker's own number.
C - inferred or estimated. We give it low weight, and it never decides a score on its own.
Seed - a placeholder value used while we build a scorecard. It's always labeled as such, and it gets replaced by A or B evidence as we finish verifying.

Non-negotiables

These are the lines we won't cross, no matter how popular the product:

A claim is not a certification. "Tested to NSF standards" earns zero certification credit and is labeled uncertified - a phrase on the box is not the same as being listed in the registry.
Total cost of ownership is computed from real, sourced filter prices. If we can't find the price, you'll see a labeled gap - we don't fill it with a guess.
Hard fails (ozone generators, uncertified lead/PFAS claims, overstated room sizes) cap a dimension, and you'll still see exactly why in the breakdown - nothing gets quietly buried.
Affiliate revenue never moves a score.

Verification coverage

When you see Verified on a product, it means we checked its rating against an official source ourselves - AHAM Verifide for CADR, the NSF / WQA / IAPMO registries for water certifications, or the published ANSI/ASHRAE 52.2 MERV for furnace filters - not just the manufacturer's Performance Data Sheet. Coming up empty counts as a finding too: if the registries return no listing, we record that the claim is "tested to," not certified, rather than leave it hanging. Right now, 59 of 61 scored products are Verified.

Category	Verified
Air purifiers	25 of 25
Water filters	25 of 25
Furnace filters	9 of 11
All products	59 of 61

The rest stay Partial on purpose, not because we forgot them: these are the cases where the public record genuinely doesn't settle the question - a manufacturer data sheet that contradicts itself on the rating, say, or a seller-applied number with no accredited test behind it. When the answer is murky, we'd rather show you the doubt than hand you a false certainty.

Air Purifier v1.0

Dimension	Weight	Measures
Verified Performance	30%	AHAM CADR, true sealed HEPA, air changes per hour, honest room-size match.
Total Cost of Ownership	25%	Computed annual filter cost vs the room-size class median. The number nobody surfaces.
Certification	15%	AHAM Verifide, Energy Star, CARB, published lab efficiency.
Safety	15%	Ozone and ionizer behavior. Ozone generators hard-fail.
Practical Fit	15%	Noise, filter-change indicator, energy draw. Smart features are neutral.

Water Filter v1.0

Dimension	Weight	Measures
Verified Contaminant Reduction	35%	Held NSF/WQA/IAPMO certifications for the contaminants that matter. A claim is not a certification.
Total Cost of Ownership	25%	Computed cost per gallon vs the format-class median. The number nobody surfaces.
Certification Independence	15%	Certified on the official database vs self-claimed 'tested to' marketing.
Capacity & Flow Fit	15%	Cartridge life and flow appropriate to the format.
Practical Fit	10%	Install, footprint, and source-water readiness.

Furnace Filter v1.0

For furnace filters, the rating you can rely on is MERV (ANSI/ASHRAE 52.2) - it's the only standardized one, so it's the only one we trust. MPR (3M) and FPR (Home Depot) are in-house marketing scales the seller controls, and the same proprietary number can stand for a different MERV depending on how thick the filter is - so don't treat them as a stand-in. We score the true MERV and reward honest labeling - the furnace-side version of "tested to is not certified to."

Dimension	Weight	Measures
Verified Filtration	30%	True ANSI/ASHRAE 52.2 MERV - the only standardized filtration rating. Higher MERV captures finer particles.
Total Cost of Ownership	25%	Computed annual filter cost vs the depth-class median. Cheap 1-inch filters replaced often can cost more than deep media.
Rating Honesty	20%	Does the brand print the true MERV, or hide it behind a proprietary MPR/FPR scale? MPR/FPR is not MERV.
Airflow Fit	15%	Depth vs MERV. A high-MERV 1-inch filter restricts airflow and can strain the blower; deep media at the same MERV does not.
Practical Fit	10%	Replacement cadence and reusability. Frequent swaps are a maintenance and cost burden.

Bands

Here's how to read a composite at a glance: Excellent 8.5-10 · Strong 7.5-8.4 · Mixed 6.5-7.4 · Limited under 6.5. You'll never see one of these numbers on its own - we always show the sub-scores that add up to it.