System Finder Methodology

Original research by the Pure Whole House Team · Last updated September 30, 2026

Most water-filter buying guides are rewritten manufacturer brochures. We built the System Finder — a filterable database of 49 whole-house systems — on a different foundation: documented specifications, live certification listings, and observed retail prices. This page explains exactly how it works, so you can judge our numbers for yourself.

How we chose the 49 systems

• 11 brands that actually sell whole-house treatment in the US: SpringWell, Kind Water, Aquasana, iSpring, Pelican/Pentair, US Water Systems, Home Master, APEC, 3M Aqua-Pure, GE, and Whirlpool.
• We cover the systems people actually buy for city water, well water, hard water, and specific problems (chlorine, chloramine, iron, sediment, PFAS) — not every SKU each brand sells.
• When a system is discontinued, we remove it and log the change in our price change log.

Where our data comes from

• Manufacturer spec sheets and install manuals — flow rates, tank sizes, media types, warranty terms.
• Live NSF, WQA, and IAPMO certification directories — not manufacturer marketing pages. Every system in the Finder carries a “certification verified” date stamp.
• Published retail prices — manufacturer sites and major retailers, recorded as ranges on a stated date.
• Every figure is labeled documented (found in a manufacturer document, certification listing, or live retail price) or estimated (inferred from a comparable source, with the basis labeled). A dash means it isn’t published anywhere we could find — not that it doesn’t exist.

How we estimate 10-year cost of ownership

The sticker price is the least informative number on a water system. What matters is what it costs to own. Our estimate:

10-year total = equipment price + installation + (annual upkeep × 10)

We also show an annualized cost — the 10-year total divided by 10 — wherever the 10-year total appears in the Finder: the same estimate, expressed per year.

For the full side-by-side numbers, see our 2026 Whole-House Water Filter Cost Index — all 49 systems ranked by 10-year cost, with a citation guide for journalists and researchers.

• All inputs are ranges, so the total is a range — honest about uncertainty instead of pretending at precision.
• Annual upkeep comes from replacement media/cartridge costs and change intervals in manufacturer documentation.
• Where intervals are gallon-rated, we assume 300 gallons/day household use.
• Installation is the widest range because it depends on your plumbing; we show typical professional-install ranges, not DIY best cases.

Cost per 1,000 gallons

The 10-year total tells you what a system costs to own. Cost per 1,000 gallons tells you what the treated water costs: our signature metric for comparing systems of different sizes and prices on equal footing.

Cost per 1,000 gal = 10-year total ÷ 1,095

The divisor comes from a stated household assumption: a 4-person household using 75 gallons per person per day, or 300 gallons/day (the same assumption behind our gallon-rated upkeep intervals). That is 109,500 gallons/year, or 1,095,000 gallons over 10 years; dividing by 1,000 gives 1,095 thousand-gallon units.

• Because the 10-year total is a range, we use the midpoint of the range for this calculation, rounded to the nearest cent.
• Two systems have no published upkeep data and therefore no 10-year total; they show no cost-per-1,000-gallon figure rather than a guess.
• Your number will move with actual water use: a 2-person household will see a higher cost per gallon, a large family a lower one. The metric is for comparing systems, not predicting your bill.

Certification verification

• Each system shows a “certification verified” date — the last time we checked its certification claims against the live NSF, WQA, or IAPMO directory, or against the manufacturer’s certificate.
• Where marketing outruns certification, we say so directly in the system’s certification note — for example, when a retailer claims “NSF certified” but only the components (not the system) are certified, or when a brand names no standard at all.
• Most whole-house systems have no system-level certification. That’s normal for the category, and we show it plainly rather than hiding it.
• Full claim-by-claim detail lives in our Certification Database.

Certification coverage

Certification verification tells you when we checked. Certification coverage tells you what we found: a short, comparable summary shown on every system card and in the side-by-side comparison:

• “NSF 42, NSF 372 (2 standards)”: the system itself holds those listings, checked against the live directories.
• “IAPMO R&T to NSF/ANSI 42 (per owner’s manual)”: the owner’s manual states system-level certification; we verified the manual, not the directory listing.
• “No NSF/WQA certification claimed”: the honest majority. Most whole-house systems have no system-level certification, and we say so plainly.

Every coverage line carries its verification date. A coverage line is a record of what the certification databases said on that date, not an endorsement, and not a substitute for checking the directory yourself before buying on the basis of a certification.

How we build an evidence chain

Every claim on this site sits at the end of a chain. When you read a statement about a system, you should be able to trace it back:

1. Manufacturer specification: what the brand publishes (spec sheets, manuals, product pages). This is a claim, not a fact.
2. Certification record: whether an independent body (NSF, WQA, IAPMO) has tested and listed the system for that claim. A claim with a listing outranks one without.
3. Independent testing: published third-party test data where it exists. We do not have a lab and we do not pretend otherwise; when no independent test exists, we say so.
4. Interpretation: what the first three mean for your buying decision, in plain language, with the uncertainty stated.

Where the chain breaks, a marketing claim with no certification record, a specification with no published test behind it, we mark the break explicitly instead of smoothing over it. That is the point of the certification notes, the data-quality flags, and the “no system-level certification” labels: a visible chain with visible gaps beats a confident story with hidden ones.

Price verification

• We re-check prices across the database regularly and stamp every database page with the observation date.
• Every price change, discontinuation, and buy-link change is logged publicly in our price change log.
• Prices are ranges observed on a stated date. Always check the current price before buying — ours go stale, and we tell you exactly when.

How we estimate flow from bathrooms

The Finder’s Bathrooms question converts your bathroom count into a required service flow (gallons per minute), then keeps only systems whose published service flow meets or beats it:

• 1 bathroom → 7 GPM · 2 → 9 · 3 → 11 · 4 → 13 · 5 → 15 · 6+ → 17 GPM

This is a rule-of-thumb estimator, not a plumbing calculation. It uses the midpoints of published whole-house sizing guidance: the Pure Water Gazette’s service-flow chart (1–2 baths ≈ 5–7 GPM, 2–3 baths ≈ 7–10 GPM, 3–4 baths ≈ 10–12 GPM, 4–5 baths ≈ 12–14 GPM for typical households) and multi-bathroom peak guides (around 10–13 GPM peak for three bathrooms) — rounded to 7 GPM for the first bathroom plus about 2 GPM per additional one, on the assumption that not every fixture runs at once. Homes with body sprays, soaking tubs, or frequent simultaneous showers should size up. Systems that publish no flow rate are never excluded by this filter: missing data widens your results, never narrows them.

Salt vs salt-free preference

The Hardness treatment preference filters hardness and scale treatment systems only — salt softeners, salt-free conditioners, descalers, and filter-plus-conditioner combos. Choose Salt softener and only true salt softeners remain; choose Salt-free and only salt-free technologies remain. Systems that don’t treat hardness at all (carbon tanks, cartridge filters, UV combos, RO) pass through unaffected — filtering them on a hardness question would be dishonest. So if your problem is chlorine rather than hard water, this question changes nothing; that’s by design.

Maintenance tolerance

Plain language for the yearly-upkeep filter: “I don’t want to think about it” keeps only systems under $100/yr upkeep; “Some upkeep is fine” keeps everything up to $200/yr; “I’ll do whatever it takes” keeps all systems. The upkeep figures themselves come from replacement-media costs and change intervals in manufacturer documentation.

Water test numbers: guidance, not filters

You can enter your hardness (grains per gallon) and iron (ppm) from a water test. These numbers never filter systems out — and that’s deliberate. Manufacturers don’t publish per-system hardness capacity (grains) or rated iron removal (ppm), so any cutoff we invented would be fiction. Instead, your numbers feed a “Your likely setup” guidance panel: hardness is classified on bands adapted from the USGS scale (soft under 1 gpg, slightly hard to 3.5, moderately hard to 7, hard to 10.5, very hard above), iron is compared against the 0.3 ppm EPA secondary standard, and the panel tells you plainly what each number implies — including when the honest answer is that you don’t need treatment. The panel is labeled as guidance, not a filter, and it suggests which problem filters to add.

Why this system / why we excluded the rest

Every matching result carries up to three “Why this system” bullets, each derived from a filter you actually set — your bathroom count’s flow need, your salt preference, the problems it treats, your upkeep tolerance, certifications you required, your water source. If no filters are set, there are no bullets: we won’t invent reasons.

Below the results, “Why we excluded N systems” reports exactly which filter(s) each non-matching system failed, grouped into counts (“7 over budget · 4 can’t hit your flow rate”) with an expandable list naming every excluded system and its failing reasons. Nothing is hidden: if a filter you set is doing the excluding, you’ll see it.

What we don’t do

• No lab, no hands-on testing. We don’t own these systems, haven’t installed them, and don’t run water tests. Anyone who tells you they “tested” 49 whole-house systems is selling something.
• No pay-for-rankings. The Finder has no “best overall” winner. Systems are labeled “Best for [use case]” and ranked only by the filters you choose — usually 10-year cost.
• We tell you when you don’t need a filter. If your water doesn’t have a problem our systems solve, we’ll say so.

Use our data

Journalists, researchers, and bloggers: you’re welcome to cite our cost-of-ownership figures. Please credit “Pure Whole House” with a link to purewholehouse.com/system-finder/. The table below is our full 49-system dataset as of September 30, 2026 — equipment price ranges and estimated 10-year totals. For the filterable version with flow rates, certifications, and upkeep detail, use the System Finder.

Affiliate disclosure: we may earn a commission if you buy through our links, at no extra cost to you. It doesn’t affect what we publish — the numbers above are the same either way.