Skip to content
Tariff data · 2026-08-26

CC0 · no sign-up · snapshot 2026-08-26

Download the whole thing

Everything behind this site as one CSV: 583 residential rate plans from 135 utilities in 45 states, with the rate range, time-of-use structure, fixed charge, and a modelled annual bill for three households. Free to use for anything, including commercially, with no attribution required.

What is in it

ColumnWhat it holds
urdb_labelThe source record ID. Resolves at apps.openei.org/USURDB/rate/view/{label}
utility, eiaid, stateWho files the tariff and where
rate_name, zoneThe plan, and the rate zone it applies to if the utility has them
rate_kindtime-of-use, standard, or demand-based
is_tiered, is_seasonal, has_demand_chargeStructure flags
service_classtypical, not-a-home, or unusual-fixed — see the caveat below
min_cents_kwh, max_cents_kwhRange across all periods and tiers, adjustments included
peak_offpeak_ratioHighest ÷ lowest, for plans that price by hour
fixed_charge_annual_usdStanding charge, converted to a year from daily or monthly
billing_scope, effective_dateWhat the tariff covers, and when it took effect
annual_usd_apartment / _house / _house_with_evModelled annual bill — see below
tariff_url, page_urlThe filed document, and our page for the plan

The three modelled bills

The last three columns are the part that is not in the source data. Each is the result of running a defined household through that tariff hour by hour for a full year — including the way air conditioning loads summer afternoons, electric heating loads winter mornings, and weekend schedules differ from weekdays.

ProfileAssumesUses about
ApartmentOne or two people, no central air, no electric heat4,900 kWh/yr
Family homeThree or four people with central air conditioning9,500 kWh/yr
Home with an EVSame house, plus an electric car charging overnight13,100 kWh/yr
Full definitions, including the load shapes, are in the households CSV.

These are estimates for comparing plans against each other, not predictions of a specific bill. Because the same household goes through every tariff, most of the modelling error cancels and what remains is the difference in rate structure — which is the thing worth measuring. They exclude taxes, local surcharges, demand charges, and net metering.

What was removed, and why

The source database contains a great deal more than this. A plan is here only if it is currently effective, open to new enrollment, and covers a whole bill rather than the delivery half. Roughly half of the residential records in the source fail one of those tests — mostly because they expired.

  • Expired and closed plans are dropped. A grandfathered rate nobody can join is not an option, and mixing them in makes any “cheapest plan” answer wrong.
  • Delivery-only tariffs are dropped. Their per-kWh prices look spectacularly cheap because they are half a bill. Why that matters.
  • Non-household services are kept but flagged. Irrigation pumping, street lighting, master-metered buildings and large farm service are filed under the residential sector but do not describe a home. The service_class column marks them; we exclude them from our own comparisons and you probably should too.
A caution on fixed charges. A handful of tariffs carry a standing charge that looks implausible — a monthly amount that appears to be tagged as daily in the source, which multiplies it by thirty. We have not corrected these, because we cannot tell an error from a genuinely expensive rural connection without reading the tariff document. Anything over $50 a month is marked unusual-fixed in service_class. Treat those rows as needing a look at tariff_url before use.

License and reuse

The underlying tariffs are public domain via the NREL Utility Rate Database. Our parsing, filtering, classification and modelling are released under CC0 — use it for anything, commercially or not, with no attribution required. A link back is appreciated but not asked for.

If you build something with it we would like to hear about it: [email protected].

Rates move on regulatory timelines measured in months, and this file is a snapshot taken 2026-08-26. Every row carries its own effective_date; cite that rather than the download date if you publish anything derived from it.