The public SFOE register «Electricity production plants» is the only freely accessible, granular data source for photovoltaic systems in Switzerland – but it systematically shows the last twelve months or so too low, because registrations flow in with a time lag. In our own analysis the register total reached roughly 8,635 megawatts, while Swissolar's official year-end figure for 2025 was roughly 9,498.6 megawatts – a capture rate of around 91 per cent. (As of: September 2026)
Key points in brief
- The SFOE/Pronovo register is freely accessible and provides the most granular public data source on Swiss PV systems.
- It systematically under-captures the most recent twelve months or so, because registrations only appear completely with a time lag.
- Register total roughly 8,635 MW vs. Swissolar's 2025 annual figure of roughly 9,498.6 MW – a capture rate of around 91%.
- Structural comparisons (canton, system type) are hardly affected by this effect, provided they remain stable across several time slices.
Which data source is meant – and why does it matter?
Under the designation «Electricity production plants» the SFOE maintains a publicly accessible register that is fed by the registration data each system leaves with Pronovo. It is the only freely accessible source that makes photovoltaic systems traceable down to municipality level.
How large is the under-capture – and how was it checked?
| Source | Value (end of 2025) | Character |
|---|---|---|
| SFOE/Pronovo register (own analysis) | roughly 8,635 MW | granular, with a capture lag for the most recent registrations |
| Swissolar «Solar energy statistics» | roughly 9,498.6 MW | annual, consolidated association figure |
| Calculated capture rate | roughly 91% | ecoEn's own calculation |
As of: analysis with reference year 2025. The capture rate is our own ratio calculation, not an official metric of the SFOE, Pronovo or Swissolar.
How can you check whether a metric is affected by this effect?
A simple robustness test: a structural metric (for example a cantonal ranking) is calculated at several time slices. If the rank stays stable across all time slices, the metric is probably not an artefact of the most recent under-capture. Practical consequence: absolute additions figures for the current year should be checked against a second, consolidated source before they are communicated as the current annual value.
When working with register data for our own analyses, this effect first caught our attention as an apparent "slump" in the most recent year – only the comparison with Swissolar's annual statistics showed that it is a capture lag and not a real decline.
Frequently asked questions
Does that make the SFOE/Pronovo register unusable as a data source?
No. For granular, structural questions it remains by far the most detailed public source. Caution is needed only with absolute addition figures for the most recent twelve months or so.
Why is the under-capture not officially documented?
We are not aware of any official SFOE or Pronovo publication that quantifies this effect explicitly. The classification shown here rests on our own comparison with Swissolar's annual statistics.
How large is the delay typically?
In our own analysis the effect affected mainly the last twelve months or so before the respective analysis reference date.
Can I adopt this methodology for my own work?
Yes, the approach is openly described and can be transferred to other questions using the same data.
Questions about interpreting the data?
For queries about the methodology described here, the ecoEn editorial team is available.
Sources: SFOE/Pronovo register «Electricity production plants» (data.geo.admin.ch), Swissolar «Solar energy statistics».
Last updated: 18 September 2026 · Contact for questions on methodology: ecoEn editorial team
