Methodology

Data sources

• Utility registry — EPA Safe Drinking Water Information System (SDWIS), which lists all ~50,000 active community water systems in the US. EPA SDWIS ↗
• Rate schedules — collected directly from each utility's published documents: PDFs, web pages, and official fee schedules.
• Median household income — US Census Bureau American Community Survey (ACS) 2022 5-year estimates, variable B19013_001E, at the census tract and zip code tabulation area (ZCTA) level. ACS 5-year ↗
• Service area boundaries — EPA Water System Service Areas dataset, where available. EPA Service Areas ↗
• Census tract boundaries — Census Bureau TIGER/Line Cartographic Boundary Files, 2022 vintage. Census Boundaries ↗

Data collection and extraction

Rate data is sourced directly from official utility publications. We use a wide variety of tools enabled by large language models to search the web, locate each utility's current rate schedule, and extract it into a consistent schema adapted from the Open Water Rate Specification (OWRS) ↗.

Bill calculation, tier ordering, and unit conversion are all deterministic rule-based code, not model inference — see Data quality below for the checks every extraction passes before publication. The source document, extraction date, and calculation steps are recorded for every entry and shown in the detail on each utility's page.

We aim to eventually cover all 4,500+ large utilities (serving 10,000+ people) on a recurring basis, refreshing each on some regular interval as capacity allows — we haven't locked in a target frequency yet.

Benchmark calculation

All rates are standardized to 6,000 gallons per month for a residential customer with a 5/8" or 3/4" meter — a typical US household (~50 gallons per person per day for a 4-person household). This consumption level is used as a reference in EPA affordability analysis. EPA Water Affordability Needs Assessment ↗

The monthly cost includes:

• Fixed base / service charge
• Volumetric charges across all applicable tiers

It excludes taxes, sewer fees, stormwater charges, and other surcharges, which vary widely and are often billed separately.

Seasonal, zone, and billing-frequency normalization

• Seasonal rates — we report the peak-season rate as the headline figure: a household that can't afford the peak bill faces genuine access risk regardless of what they pay in other months (Teodoro 2018 ↗). Full seasonal breakdown shown per utility.
• Zone-based rates — we use Zone 1 (or the lowest-numbered, lowest-elevation zone) as the standard customer rate. In-district vs. outside-city is not treated as a zone distinction; we always use the in-district rate.
• Billing frequency — bimonthly/quarterly bills are normalized to a monthly equivalent: volume is scaled to the billing period, the bill is computed against that scaled volume, then divided back to a per-month figure.

Water burden

Water burden is calculated as annual water cost as a percentage of median household income:

The EPA's affordability guidance applies a 4% threshold to combined water and sewer costs. Since whatwatercosts.org reports water-only rates, we use 2% as the equivalent benchmark for water service alone. Utilities where water costs exceed 2% of median household income are flagged as a potential affordability concern. EPA Financial Capability Assessment Guidance (1997) ↗

Median household income is matched to each utility in two stages: tract-weighted where a service area polygon exists (EPA Water System Service Areas, intersected with 2022 Census tracts, population-weighted), falling back to zip code (ZCTA) otherwise. Source: ACS 2022 5-year estimates.

Even with tract-level matching, service area boundaries are an approximation. A utility's true customer base may differ from its mapped footprint.

Data quality

Every extraction passes a deterministic validation gate before publication. Status labels describe the outcome — they do not mean the utility has approved the number. All figures are standardized estimates for comparison, not official bills.

Hard checks

Any failure rejects the extraction outright:

• Known unit — gal, kgal, ccf, hcf, cf, or m3.
• Tier boundary integrity — gapless and contiguous, starting at 0 with no upper bound on the last tier.
• Bill sanity ($10–$300) — the computed 6,000-gallon benchmark bill must fall in a plausible range; outside it almost always means a unit, billing-period, or customer-class error.

Scored checks

Pass/fail/not-applicable; publication requires ≥80% of applicable checks to pass. Full results shown in the calculation detail per utility.

• Monotonic pricing — tier prices non-decreasing.
• Price range plausible — per-unit price within known bounds for the stated unit.
• Fixed charge range — base service charge between $0–$150/month.
• No zero tier prices.
• Tier count (1–8).
• Tier boundary scale — consistent with the stated unit.
• Residential customer class.
• Effective date parseable — valid, and not more than a year in the future.
• Plausible for state — within the Tukey fence (Q1 − 1.5×IQR to Q3 + 1.5×IQR) of same-state peers, when ≥10 exist.

Status labels

flagged

Failed a hard check or scored below 80%. Not shown in comparisons, rankings, or averages.

model estimate

Passed all hard checks and ≥80% of scored checks. Internally consistent but not independently verified.

human verified

A person manually checked the estimate against the source document and confirmed or corrected it.

Benchmark accuracy

We continuously evaluate the accuracy of this project. Our latest run showed that the majority of rates were within $1.42 of a curated benchmark dataset. We'll share more on benchmarking here later.

Limitations

Rate schedules change. We do our best to keep data current but cannot guarantee real-time accuracy. Seasonal rates, drought surcharges, and budget-based tiers may not be fully captured. Low-income assistance programs are not reflected in the benchmark figure. Service area boundaries are approximations; a utility's true customer base may not align with its mapped footprint.

If you find an error, please use the feedback form on the utility's page or contact us directly.

How to cite

If you use whatwatercosts.org data in research or reporting, please cite it as:

Raw data and machine-readable descriptions are available at /llms.txt.