Skip to main content
The most dangerous thing an API like this can do is return a short list without saying what the list is short of. A reader who asks for dynamic tariffs and gets six results will conclude one of two very different things, and the response has to decide which. So filtered endpoints return the denominator.

The rule

/api/map/tariffs reports three counts on every request: A response where n_matched is far below n_total is a real answer about the Swiss market, not a gap in collection.

What that looks like in practice

Tariff year 2026, grid level NE7, consumption side: Eco products are offered by about a fifth of operators. Dynamic household tariffs, for all the attention they get, are offered by six.

A match without a price is still a match

Look again at the dynamic row: six operators match, one has a rate. That is not a data quality problem — it is what a dynamic tariff is. Its price is a time series, not a number, so there is nothing to put in a rate_rp field. Those five operators are returned with rate_rp: null and priced: false:
Dropping unpriced matches would make the dynamic filter answer “1 of 502” for something six operators offer — the single most misleading number this API could produce. If you re-aggregate this data yourself, count n_matched, and average over n_priced.
To get actual prices for those operators, ask for the time series instead.

Coverage is not uniform across fields

Before building on a field, check how populated it is: GET /api/v1/periods gives the live per-stage, per-year breakdown; GET /api/readiness gives collection progress against the programme’s deadlines.

Years

A tariff year is collected during the preceding year, so the current year is always more complete than the coming one. Ask /api/v1/periods rather than assuming — and when you want “the year that is being collected now”, /api/meta returns it as default_period.