Sources & method
Where every number comes from
A rating you can't audit is just an opinion with a number on it. This page is the full method: which public sources we collect, how the data is checked and normalized, how the score is built, and what would make us change it. One thing up front: WaterKnown does not currently run its own lab testing. Every number on this site is drawn from a cited public source.
The pipeline
Four steps between a buried PDF and a score
STEP 01 · COLLECTION
We pull the public record
For cities: the utility's latest Consumer Confidence Report plus EPA UCMR monitoring results, straight from the utility and EPA data files. For bottled brands: brand-published water quality reports, FDA retail surveys, and state testing programs. For filters: NSF/ANSI certification listings and manufacturer performance data sheets. Every datum enters our records with its source, its date, and a link.
STEP 02 · VERIFICATION
Checked, normalized, dated
Reports arrive in a mess of units, detection limits, and formats. We convert everything to consistent units, record each source's reporting limit next to each result, and keep the sample or publication date attached. A "not detected" is only as meaningful as the detection limit behind it, so we show that limit rather than letting n.d. imply zero.
STEP 03 · SCORING
Health guidance, not legal minimums
Every published result is compared to health-based guidance levels, which are usually far stricter than enforceable limits. Legal water is the floor, not the bar. Choosing the stricter yardstick and applying it identically to every brand and city is the editorial work this site exists to do. Weights are published below.
STEP 04 · REVIEW & REFRESH
Signed off, then kept current
Our editorial team reviews methodology changes and anomalous readings before anything publishes. City pages refresh as each year's utility report lands, and EPA monitoring data updates on its own cycle. Every page carries the date of the report behind it, because water is only as good as its last published test.
The panel
112 analytes tracked, coverage stated
These are the analyte groups we track across all sources, with the reporting limits typical of the reports we collect. Coverage varies by source, and that's part of the point: every card shows which analytes its report actually measured, and a report that skips a group can't score points for it.
| Group | Analytes | Typical reporting limit | Examples |
|---|---|---|---|
| Heavy metals | 24 | 0.1 ppb | Lead, arsenic, chromium-6, uranium |
| PFAS | 18 | 1 ppt | PFOA, PFOS, GenX, PFBS |
| Disinfection byproducts | 16 | 0.5 ppb | TTHMs, haloacetic acids, bromate |
| Pesticides & herbicides | 32 | 5 ppt | Atrazine, glyphosate, 2,4-D |
| Volatile organics | 12 | 0.5 ppb | Benzene, PCE, TCE |
| Radiologicals | 6 | 1 pCi/L | Radium-226/228, gross alpha |
| Microbial | 3 | 1 CFU/100mL | E. coli, total coliform |
| Microplastics | 1 | 1 µm particle size | Particle count per liter |
Score weights
What moves a score, and by how much
Two scoring models, because a bottle and a filter answer different questions. A bottle is judged on what its published data shows is in it, and on how much it publishes. A filter is judged on what it is independently certified to remove.
Bottled water & tap supplies
| Reported contaminants vs. health guidance | 60% |
| Disclosure: completeness, recency, detection limits | 25% |
| Panel breadth of the source report | 15% |
Filters
| Certified reductions vs. common contaminants | 70% |
| Certification scope: which standards, how many analytes | 15% |
| Materials safety certification of the unit | 15% |
Where the city figures come from
Two tiers, and every card says which it used.
EPA national data. Lead 90th-percentile results from SDWIS Lead and Copper Rule monitoring, and PFAS results from UCMR 5, both from EPA's 2026 Q2 snapshot. This is structured, machine-readable, and consistent across every system in the country.
Utility reports. Source water names, TTHM and HAA5 running annual averages, nitrate, and arsenic come from each utility's own published annual report, because no national table carries them consistently.
Two rules govern how we combine them. Where EPA's structured record and a utility's prose disagree about PFAS, the UCMR 5 record decides, because it is a defined monitoring program rather than a sentence in a PDF; any utility-reported detection outside UCMR 5 appears as a note on the card instead. And a system with no UCMR 5 record is never labelled “no PFAS detected,” because absence of testing is not absence of PFAS.
The filter score, in full
Filters are scored on what an accredited certifier has independently verified, never on what the manufacturer claims. Certification is model-specific and publicly searchable, which makes it the only part of this category that can be checked rather than believed.
| Input | What earns points | Weight |
|---|---|---|
| PFAS certification | Total PFAS (7 compounds) scores full; PFOA/PFOS only scores partial; none scores zero | 30% |
| Lead certification | NSF/ANSI 53 listed for lead, verified by model number | 20% |
| Certification breadth | How many health contaminants under 53, or 58 for reverse osmosis | 15% |
| Materials safety | NSF/ANSI 372 lead-free content; 61 for components | 10% |
| Cost per gallon | Cartridge price divided by rated capacity, verified against listed capacity | 15% |
| Owner-reported reliability | Leaks, flow loss, fitment and cartridge life, from aggregated buyer reviews | 10% |
Note what buyer reviews are and are not used for. They carry weight on reliability and value, because an owner genuinely knows whether a housing leaked or a cartridge died early. They carry no weight on contaminant removal, because nobody can tell from using a filter whether it removed PFAS. Any site that blends the two is guessing.
The city score, in full
Each measured input earns a subscore: 100 at or below the health benchmark, 0 at the legal limit, straight-line in between. The city score is the weighted average of whichever inputs the utility actually published, so a utility is never penalised for measuring more than its neighbours. A score appears only when at least three inputs are available.
| Input | Health benchmark | Legal limit | Weight |
|---|---|---|---|
| Lead (90th percentile) | 1 ppb | 15 ppb | 30% |
| PFAS (UCMR 5) | non-detect | any detection | 25% |
| TTHMs (highest LRAA) | 15 ppb | 80 ppb | 20% |
| HAA5 (highest LRAA) | 15 ppb | 60 ppb | 10% |
| Arsenic | 1 ppb | 10 ppb | 10% |
| Nitrate | 5 ppm | 10 ppm | 5% |
Health benchmarks reflect health-based guidance rather than enforceable minimums, which is why a fully compliant city can still score in the 60s. Lead sits at 1 ppb because no level of lead exposure is considered safe; 15 ppb is an action threshold for utilities, not a health target.
What we don't do
No sponsorships. No paid placements. No paid reviews. No free samples in exchange for coverage. No lab testing of our own yet, and no pretending otherwise. And no scoring from a marketing page: if a product has no published certification, it says so instead of getting a guess. We do earn affiliate commission when you buy through our links, at the same rate regardless of score, paid by the retailer and never by the brand. That is the whole business model on one line.
The method is public. So is every source.
Open any card and you'll find the report behind it: who published it, when, at what detection limits, and what it did and didn't measure.