Skip to main content
Every operator’s tariffs reach this database by a different road: the operator’s own machine-readable sheet, a reviewer, or a model reading a PDF. Provenance tells you which road a figure took. Accuracy tells you whether the result is right. ElCom, the federal regulator, publishes for every operator and year what a standard customer pays: a household using 4’500 kWh a year (category H4), 2’500 kWh (H2), and businesses from 8’000 kWh (C1) to 7.5 GWh at medium voltage (C7). It splits that figure into energy, grid usage, community fees and the national charges. Every night this service composes the same customer’s bill from the tariffs it holds and compares the two.
The response lists all nine categories, H4 and H2 and C1 to C7; two are shown here.
Business bills (C1–C7) are harder than household bills: their grid tariffs carry a price per kW of peak demand, and many operators file business products in separate documents. About half are within ±5 % today, with a median miss of 4 to 5 %, against 71 % and 2.2 % for households. Check a business bill’s own elcom_check before relying on it (business bills).
The same numbers, with the method in full, are on the accuracy page.

What a miss means

A bill can miss ElCom’s figure for reasons that are not an error in the data. ElCom’s figure is for its standard customer, on the operator’s standard product; an operator with several household products or a dynamic tariff can be billed correctly and still differ. A miss is a place to look, not a verdict.

Operators whose fee differs by municipality

ElCom’s figure for an operator is an average over the municipalities it supplies, and within one operator only the community fee differs between them: Romande Energie’s 262 municipalities range from 27.3 to 29.8 Rp/kWh for H4. Where the operator’s sheet states the fee per municipality, the bill can be exact for each, so the measurement prices every municipality ElCom has a figure for and compares the one with the median miss against ElCom’s figure for that municipality. Not the best match, which would flatter the result: the typical one. Where the sheet has one fee for the whole area, the operator’s average stays the reference. To get an exact bill for one municipality, pass its BFS number as bfs to /api/bundle. When a bill misses, the measurement names the part that misses most: energy, grid usage, metering, community fees, the national charges, last year’s prices still on file, or a part with no price at all. Those causes drive what the collector does next; they are not part of the public response.

What the measurement is not

ElCom’s household figure (H4) is also what some corrections are learned from: which product is an operator’s standard one, which grid tariff a household is on, whether a grid price already contains the national charges. For those operators H4 is no longer an independent check. H2 never feeds any learning, so it is the cleaner test.
Where an operator has published no tariffs at all, the bill falls back to ElCom’s own figures, marked as such. Those bills are excluded: a figure compared with itself proves nothing.
ElCom publishes standard categories, not every product. A dynamic tariff, a heat-pump product or a business on a negotiated contract is not measured.

How far one bill can be trusted

Every composed bill (/api/bundle) carries a confidence block: the strongest evidence behind that bill, so you can use the strong ones directly and treat the rest with care.
source is where the lines came from, whatever ElCom says about them. estimated_parts names parts the sheet left empty and ElCom’s figure filled (the meter, the community fee), and a dynamic price estimated from its published quarter-hour prices (energy (dynamic)). Such a line carries dynamic_estimate, with the period, the number of prices and the method: the mean over the last year, once a week or more is on file. elcom_check is the latest nightly check in the ElCom category nearest your consumption. It applies only to the bill as the database opens it, within 25 % of that category’s consumption. Once you choose another product or tariff, applies is false and why_not says why.