Methodology and data gaps

Registration data as of: 2026-08-31

How the counting works here

The basis is the original Excel files from auto-schweiz, reference date 31.08.2026, passenger cars for Switzerland and the Principality of Liechtenstein, official imports and parallel imports. Light commercial vehicles are excluded — in the same file they sit in a separate block below the first total row, and evaluations that miss this count them in.

A vehicle counts as chargeable if it has an external charging port. Full and mild hybrids are not included, because they have no plug and never draw electricity from the grid. To tell them apart: a mild hybrid is a petrol or diesel car with a small 48-volt system that merely supports the engine and recovers braking energy — it does not drive a single metre purely electrically. A full hybrid covers short distances, two to five kilometres at city speeds, from a small battery charged by the engine and by braking, and helps above all in town and in traffic jams — likewise without a plug. Both are combustion cars at their core. Why this line matters for the purchase decision is set out at length in the blog; how large their market share is and why it is overstated in the statistics is in the pointer under «Situation».

The essentials in three sentences
  • Only vehicles with a charging port are counted — full and mild hybrids without a plug are in none of the figures on this site.
  • The basis is the original Excel files from auto-schweiz, not secondary reports. Where two sources contradict each other, both values are shown.
  • What this site does not know is in the register below — it is part of what we offer, not a note to ourselves.
How the figures are calculated

Drivetrain codes used by the source

  • E — purely electric (BEV): 37’226 vehicles
  • C — plug-in hybrid with a petrol engine
  • F — plug-in hybrid with a diesel engine, practically only Mercedes
  • R — range extender: 314 vehicles, all from Leapmotor. The drivetrain that Opel brought to Switzerland with the Ampera in 2012 is thus back on the market.

Origin of the brand

The classification follows the group a brand belongs to, not the production site. Brands of the Geely group under European management — Volvo, Polestar, Smart, Lotus — are marked separately, so the choice stays transparent instead of quietly falling into the China total. Both readings can be checked in the filter of the model radar.

Comparison logic

Δ compares Jan–Aug with the same period in 2025, not with the previous month. This removes seasonality and late-reporting effects, which in month-on-month comparisons are regularly misread as a «collapse». Models without registrations in the previous year are marked «new» instead of carrying a misleading percentage.

The buying signals

The signals in the model radar are rules, not judgements. They read: zero registrations in the previous year → «new to the market». Under 50 vehicles → «rare». From 200 vehicles with more than 10 % growth → «widespread & growing», with more than 10 % decline → «widespread, declining». In between «moderately widespread». The thresholds are set, not derived — they are meant to make orders of magnitude visible, not to rate vehicles.

Data gap register what could NOT be substantiated
  1. Swiss-specific breakdown statistics by model Do not exist publicly. The TCS breakdown statistics were an ADAC cooperation; the traceable model-level editions date from 2010–2013. Used instead: ADAC 2026 with a German data basis.
  2. Residual value forecasts for individual models Not freely available. Used-car time series sit behind paywalls at Autovista/Eurotax and AutoScout24. The individual model figures used here are as at early 2024.
  3. Charging prices as at 2026 The TCS comparison is the only systematic Swiss survey that could be found, but it contains stated values from June 2023. A current price comparison of all providers on a comparable methodology could not be substantiated. A practical remedy instead of a table: check the prices in the apps. The section on this sits under «Used or new».
  4. Separating mild and full hybrids Not reliably possible in the available registration data, because the category is derived from the CO₂ figure and manufacturers name their 48-volt systems inconsistently. Both categories are therefore excluded entirely rather than assigned incorrectly.
  5. Number of public charging points Closed as of 11.09.2026: we now count for ourselves, from the open monthly series of the Swiss Federal Office of Energy — locations, charging stations and plugs kept apart, by canton, from November 2020. The figures are in the «Charging» tab. That also disposes of the reason for this gap: the circulating figures of 14’000, 15’200 and 20’000 are secondary sources meaning three different levels. Utilisation remains open — see the entry further down.
  6. Monthly series for individual models auto-schweiz sometimes adds new models as their own row with a delay and books accumulated vehicles into a single month. Annual totals and year-on-year comparisons are unaffected, the monthly distribution of individual models is not. Affected entries are marked with an asterisk in the outlook.
  7. Prices and dates in the outlook For models from 2027 onwards there is almost only German market data in euros. Swiss list prices regularly differ from it and are usually communicated later. Dates for 2027 and beyond are manufacturer announcements, not commitments — alongside 46 confirmed models the ADAC lists 23 uncertain candidates.
  8. Cantonal subsidies and vehicle taxes Partly closed: the outlook now links to EnergieSchweiz/Energiefranken, the TCS comparison and the Comparis calculator. What remains open is that the specific subsidy amounts for Ticino, Vaud and Basel-Stadt come from an overview page and were not verified with the cantons. Several cantons are also changing their rules right now — St. Gallen since 2026, Solothurn planning for 2027.
  9. Market share figures from aggregator sources One source gives a BEV share of 32 % for 2025. The primary statistics from auto-schweiz show 22,8 %. This site follows auto-schweiz throughout.
  10. Experience reports The form under «Experiences» checks submissions in the browser, but there is no server that receives, stores and releases them. The pre-filter is a convenience for the sender, not a security measure, and can be circumvented with the developer tools. As long as no backend is connected, the route is by email or clipboard.
  11. Notification service The sign-up form for notifications under «News» collects channel, contact and topics, but sends nothing as long as no endpoint is set. It deliberately stores nothing locally and does not fake a registration. The actual service with double opt-in, storage under the revised Data Protection Act, unsubscription and dispatch needs a backend and is not yet connected.
  12. Petrol price as an annual series For unleaded 95 there is no freely accessible, machine-readable Swiss annual average series in centimes per litre. Avenergy Suisse maintains one but publishes it only as a chart; at Statista the values sit behind a paywall. The petrol values in the three and six year views are therefore read off that chart and accurate only to about two centimes. The Federal Office of Energy does maintain a monthly series, but as an index (December 2020 = 100), not in centimes per litre. For 2026 TCS spot checks are available, not monthly averages.
  13. Brent figures contradict each other countryeconomy gives a monthly average of 83,76 USD for July 2026, wallstreet-online a spot price of 82,26 USD on 07.08. and a thirty-day change of 15,61 per cent, from which around 71 USD would follow for 10.07. That does not add up. The back-calculated value was therefore removed from the chart; only directly substantiated values remain. Partly closed: since 02.09.2026 we read the daily quotation directly from Avenergy Suisse, a Swiss primary source, and since 20.09.2026 it carries the crude oil line on its own. The FRED series, which drew the history until then, has been taken out of the chart: its gap to Avenergy grew from 4,91 to 17,89 per cent across eleven shared trading days, and it sits above eleven of thirteen comparison points from third sources, by 7,9 per cent on average and by 28,9 on 08.04.2026. Why is not established. Nothing can be retrieved retroactively from Avenergy, so the line starts on 02.09.2026 and for earlier months the contradiction stands.
  14. Forecasts This site does not predict what will happen. The scenario calculator in the «Situation» tab produces figures for coming years from 2026 — as a model calculation with its assumptions and sliders laid open, not as a prediction. For the rest of the text it still holds: where the future is mentioned, it is in named mechanisms — model cycles, price pressure, depreciation curves.
  15. What the scenario calculator does not model Certificate prices are not forecast, only the available budget per tonne. The domestic quota is calculated at the statutory minimum of 20 per cent. Kerosene is excluded — 2.32 billion litres in 2025 that do not run through the road fund. The biogenic share of fuel, 5.3 per cent in 2025, lowers emissions and mineral oil tax at the same time and is not shown separately. Residual value and the electric driving share of plug-in hybrids are set assumptions, not recorded values; for the driving share the range is 25 to 50 per cent. Used-car imports have been measured since 28.08.2026 and are no longer an assumption. Checked against the full years 2021 to 2025: extrapolating from seven months lands between 6.5 per cent below and 0.8 per cent above the measured year. It tends to understate, because the second half of the year is the stronger one. For the driving share there is at least a measurement from the EU: on-board devices (OBFCM) have been mandatory there since 2021 and show plug-in hybrids consuming on average 3.5 times the laboratory value — they are charged less often than assumed. The value is therefore likely to sit at the lower end; ten percentage points less mean around 18 million francs more mineral oil tax in the target year. The assumed consumption of 7 litres per 100 kilometres is not affected: it is calibrated against actual fuel sales and is not a laboratory value, so the gap between test bench and road is already contained in it. Source for both: SFOE consumption catalogue, section on consumption. All lever and trigger amounts marked «assumption» in the controls are calibrated, not measured. Fuel sales from 2015 to 2018 are not recorded and the 2019 to 2022 series is reconstructed — the chart therefore reaches back to 2019 and no further. The intervening years of the Norwegian and Danish series are unverified and not shown. Of the two consultation variants only the distance-based levy is calculated, battery vehicles in full and plug-in hybrids at half the rate. The tax of 22,8 centimes per kilowatt hour of charging electricity is not modelled; depending on consumption it would yield a different coverage. The petrol price affects the levy only through the number of electric cars, not through their mileage: that a single electric car takes over more kilometres when petrol is expensive, because a multi-car household shifts to the cheaper vehicle, is not modelled. The direction is not in dispute, but no documented Swiss figure for it exists.
  16. Electric lorries and the heavy vehicle charge Switzerland has the highest share of heavy electric lorries in Europe; the driver is the exemption from the performance-related heavy vehicle charge. For an electric lorry both diesel tax and the heavy vehicle charge fall away; the charge brings in around 1.9 billion francs a year. The calculator holds consumption outside passenger cars nearly constant and therefore does not model this channel. Without fleet data on electric lorries it cannot be retrofitted responsibly.
  17. No rolling twelve-month figure The monthly view in the scenario calculator shows single months and the year-to-date value, not a rolling twelve-month average. The reason lies in the data: the registration file carries only the current period of both years; August to December of the previous year are missing. Seasonality is therefore removed by comparing matching periods, as with the Δ in the model radar.
  18. The vehicle fleet grows faster in the model than in the data The scenario calculator projects the passenger car fleet at 0.8 per cent per year. The documented series in the data basis grows more slowly: 0.45 per cent on average across 2019 to 2025, 0.07 per cent over the last three years, and from 2024 to 2025 it fell by 0.84 per cent. By 2036 the assumption yields around 5.19 million passenger cars instead of around 4.99 million if the documented average is projected — almost four per cent more vehicles, and therefore around 97 million francs more mineral oil tax per year. The value sits deliberately among the model assumptions and is not updated automatically; it should be reviewed at the next data update.
  19. The path of the ASTRA shortfall calculation between the reference years is our own assumption The ASTRA explanatory report of 26.09.2025 gives four amounts: no shortfall in the 2019 base year, 610 to 860 million francs for 2030 and 1500 to 2300 million for 2040. Since 28.08.2026 the scenario calculator shows two continuous lines derived from these; in between the line is interpolated straight. Only the reference years are documented, and only 2030 falls on the calculator’s time axis — which is why a point is drawn there and the rest is dashed. The corridor is also not a revenue forecast: it counts only the effect of electrification against 2019, without efficiency gains, mileage and fuel tourism. That is why it lies above the documented revenue in the starting year.
  20. Among used-car imports, full hybrids cannot be separated from plug-ins The IVZ used-import dataset lists «petrol / electric» as a single category. Projected, that is 18,468 vehicles per year, a good third of all imports. For mineral oil tax the difference matters: a full hybrid fills up entirely, a plug-in hybrid only for around 60 per cent of its kilometres. Splitting them would require an estimated ratio, which would stand as an invented figure among measured ones. They therefore count entirely towards the combustion stream — the cautious direction, which tends to overstate future mineral oil tax rather than understate it. Purely electric imports are unaffected, as «electric» is a category of its own: 4,841 vehicles per year, counted in the fleet since 28.08.2026.
  21. The SFOE counts the same vehicles differently from auto-schweiz Under Kennzahlen Fahrzeuge, the Swiss Federal Office of Energy also publishes annual figures on alternative drivetrains and charging infrastructure. They come from enforcing the CO₂ emission rules for new vehicles and differ from the auto-schweiz figures this site calculates with, because some system boundaries are drawn differently. Christoph Schreyer, head of the SFOE’s Energy-Efficient Transport section, pointed this out explicitly on 27.08.2026. We carry both sources: anyone who sees two values for the same thing without an explanation assumes one of them is wrong — and that is exactly where public debate about electric cars falls apart.
  22. Consumption, environmental record and running costs of individual vehicles are not shown here The maintained Swiss source for this is the SFOE consumption catalogue, which every model page links to. Its data are not open government data: the catalogue is produced together with the TCS and part of the data is bought commercially. On 28.08.2026 the SFOE declined a request to release it as a dataset. We therefore do not show list price and consumption — hand-entered values for 311 models go stale silently, and a stale figure with a source is worse than none. Since 10.09.2026 the link leads to the individual vehicle wherever we can map it; otherwise still into the vehicle search. This became possible through a list the TCS drew up at the SFOE’s request: 2,422 rows with make, model series, designation and the identifier that appears in the catalogue address. We only map what carries the same name or stands in a named list with a stated reason — two automatic approximations were discarded because they sent the GLC-Klasse to the C-Klasse and the BYD Seal 6 to the Seal. The catalogue lists 2,422 vehicle variants where we list some three hundred model rows; we link to one variant of the series, and the vehicle page shows the others itself, with price, drivetrain, CO₂ and range. The catalogue’s figures remain there only: the list carries names and identifiers, no values. The catalogue’s text search does not change the address, so a pre-filled link is still not possible. On 15.09.2026 the TCS answered two further questions via the BFE. We may not show the values even individually and with attribution, because they are subject to third-party copyright — which settles the question rather than merely refusing it. And: identifiers are not reused. If a vehicle detail changes, the entry disappears together with its identifier and a new one takes its place; a link that points nowhere therefore never leads to a different car. The catalogue is updated monthly on average, our mapping quarterly — in between, a variant that has disappeared falls back to the link into the vehicle search.
  23. The shortfall measures revenue, not cost The calculator compares a scenario’s mineral oil tax with that of 2025. What maintaining the roads costs does not appear in it — the expenditure side is assumed unchanged, without this being stated anywhere. It is well documented: for 2022 the FSO reports public infrastructure costs of 7,7 billion francs, of which 2,9 for national roads, 2,7 for cantonal roads and 2,2 for municipal roads. Against this stand 8,5 billion francs of revenue from motorised road traffic, a cost coverage ratio of 111 per cent. About half of that revenue is mineral oil tax. From this follows a figure the calculator does not show: coverage falls below 100 per cent as soon as the mineral oil tax loses roughly 0,8 billion — just under a fifth. Depending on the settings the calculator reports more than a billion of shortfall for 2036; the tipping point of coverage would therefore lie well before the target year. Mileage adds to this. The model assumes 0,2 per cent growth per vehicle and year with 0,8 per cent fleet growth, together about one per cent more kilometres a year. That matches the official series: for 2024 the FSO reports 130,6 billion person-kilometres, 28 per cent more than in 2000. If kilometres grow faster, maintenance costs grow with them — the mineral oil tax does not, because the additional kilometres are driven electrically. The shortfall shown is therefore a lower bound, not an estimate. Sources: FSO, road infrastructure accounts 2022 and passenger transport performance.
  24. The annual series of new registrations measures differently from this site’s key figures The documented years in the scenario calculator’s share view come from the ASTRA annual holdings (IVZ), not from auto-schweiz. The reason: an annual series needs a source that keeps years, and auto-schweiz supplies us with the current period. The two do not count pure electric cars alike. Measured for January to August 2026: the IVZ extract gives 37’608 electric passenger cars out of 151’522, that is 24,82 per cent; the auto-schweiz series 37’226 out of 151’589, that is 24,56 per cent. The total market therefore differs by 67 vehicles, the electric figure by 382 — 0,4 against 10 per mille. So it is not a matter of coverage but of classification; range extenders do not explain it, both sources keep them separate. The point for the current year still comes from auto-schweiz, like every key figure on this site: at the seam between series and current year there is therefore a step of about 0,26 percentage points which is a change of source and not a market movement. This series shows no plug-in share at all, and the reason is measured: the source’s hybrid code is only partly filled even after 2021. For 2023 the extract counts 5’737 plug-in hybrids while around 22’000 were registered in Switzerland; the rest sits under «unclear», that is combined drive without a marker. The plug-in share calculated that way would be 30,93 per cent for 2025 against 33,9 at auto-schweiz. With pure electric cars this does not happen — the fuel code is unambiguous. Before 2016 the country field is missing as well; those years are not comparably delimited and stay out.
  25. How strongly the residual value acts in the model is set, not measured The slider «EV residual value after four years» carries the badge «assumption». That expressly covers its effect as well: per percentage point of residual value the model adds 0,004 to the saturation of the curve and 0,006 to its steepness. Both numbers are set, not derived from a Swiss series — the same situation as with the factor for an electric car’s mileage. Pulling the slider from 42 to 30 per cent therefore shows a change whose direction is documented and whose size remains our assumption. A dependable residual value for an individual vehicle requires an SoH test anyway, that is a measured remaining battery capacity; without it every figure stays an approximation, including one calculated from selling prices. What we hold in measured battery condition is a single vehicle: 89 per cent after 220’158 kilometres in five years, ADAC endurance test, method Aviloo Premium, tolerance 3 points. That is one vehicle and not a statement about the series. Three Swiss platforms, read off adverts on 4 September 2026, and none shows the condition. AutoScout24 has the fields «battery condition (SoH) (0-100%)» and «battery certificate» — in three adverts the SoH value was empty all three times, including on a 2021 VW ID.3 with 119’900 kilometres. Comparis has no such field, but «battery capacity» as a nominal value and «HV battery warranty». AutoUncle carries neither condition nor nominal capacity, only consumption and estimated range. Where anything about the battery appears, it is the remaining warranty or the factory capacity — neither says how much capacity is left today. Four adverts are not a rate; whether a field exists or is missing is shown by one. The three platforms checked are not the only category: individual importers already show the condition in their own used-car shop, and even among them it is inconsistent. One advert from Emil Frey (Mitsubishi ASX 1.6 PHEV, checked on 18 September 2026) carries an openly visible «battery check» stating «battery condition (SoH) of 98.4%». AMAG, by contrast, advertises an «independent battery certificate with detailed battery diagnostics» as a programme benefit for its electric used cars — but the advert checked (Škoda Epiq Selection, 18 September 2026) carries no figure for it, only the nominal capacity of 39 kWh. The certificate may exist; what is visible on the page is the promise, not the measurement. Whether Emil Frey’s practice holds for every electric used car from that importer is likewise not established — one advert says as little about the whole shop as one does about a model series. The site carries a second residual-value figure, and it is a different one. The tab «Total cost» calculates depreciation with a rule of thumb; after four years it arrives at 58,3 per cent, this slider stands at 42. For the same quantity that is 16,3 percentage points apart. The two figures answer different questions: the slider describes what a lower residual value triggers in the market, the rule of thumb the depreciation of an individual vehicle over the holding period. Both remain assumptions. The one point we have measured on this sits closer to the slider: a VW ID.3 from 10.2020, advertised on Comparis on 4 September 2026 at CHF 18’890 against CHF 44’750 when new, that is 42,2 per cent after four years and ten months; at that same point the rule of thumb stands at 53,4. An asking price for one vehicle confirms none of this, and 42,2 per cent is an upper bound: an advertised price gets negotiated, and what the car finally brings in lies below it. By how much is written nowhere. How far such a price slides by merely sitting there is shown by another electric car on AutoUncle: a VW ID.4 GTX from 03.2022 went from CHF 29’990 to CHF 24’990 in 88 days, 16,7 per cent, without being sold. The figure in the calculator is a guide value; the binding one is the figure in the purchase contract. The gap stands here because it enters both calculations. On the curve see lg26.
  26. The total-cost calculation sets the residual-value curve, it does not measure it The tab «Total cost» calculates depreciation with a rule of thumb: 80 per cent after one year, then 90 per cent of the previous year’s value each year. The curve stands in src/tco_regel.js and applies equally to both drivetrains. That is a decision, not a measurement: there are indications that electric cars depreciate differently from combustion cars, but no Swiss series from which the difference could be quantified — dependable used-car values per model sit behind paywalls at Autovista/Eurotax and AutoScout24 (see lg2). Two different curves would therefore not be more accurate but invented, and the invented figure would sit in the largest cost block of all. The calculator lets the residual value be overridden per vehicle. Insurance, servicing and tyres are not measured values either but entries — we deliberately set no default there. The scenario calculator sets a different figure for the same quantity. Its slider «EV residual value after four years» stands at 42 per cent, this curve arrives at 58,3 after four years, so 16,3 percentage points apart. The slider maps what a lower residual value triggers in the market; here it is the depreciation of an individual vehicle over the holding period. Anyone changing the curve changes lg25 as well, and tools/pruefe_restwert.mjs makes npm test fail as long as one of the two figures is missing from the text. Details in lg25.
  27. The cantonal road tax: seven cantons calculated, 19 open It ranges from full exemption to over 900 francs a year and is therefore larger, in several cantons, than the entire energy-cost difference between an electric and a combustion car. Every canton assesses it differently — by cubic capacity, gross weight, power output or emissions, many with their own rebate or a time-limited exemption for electric cars. The rules change annually and are not available as a dataset but as 26 pieces of legislation. A table copied out once would be wrong after a year and would go on looking right — that objection has not gone away; it has been tied to a condition: a monthly run holds every tariff against the canton’s official page, and a canton without that counter-check does not go in. Grisons, Zurich, Lucerne, Bern, Zug, Aargau and St. Gallen are transcribed; in the tab «Total cost» the calculator computes the tax there from the tariff in force and names the legislation and the tariff date. For the remaining 21 cantons it stays an input field. The clock runs from first registration, not from purchase: a change of holder does not reset a rebate.
  28. The calculator does not know the sales channel What a vehicle finally brings in depends on how it changes hands — and the calculator does not ask. A trade-in at a dealer and a private sale are two different amounts for the same car: the dealer buys in order to resell and carries preparation, time on the forecourt, warranty and his own risk. That spread sits between what he pays and what the car later costs in the window. On the buying side the same spread works with the opposite sign: bought privately a car is usually cheaper than from a dealer, where preparation and warranty are in the price. And both roles exist in both channels — whoever sells privately normally buys again, and whoever trades in at a dealer is selling their car to him. The direction is therefore clear, the size is not. We know of no Swiss series that places trade-in prices and private-sale prices for the same vehicles side by side; what is publicly visible are asking prices from dealers and private sellers, not concluded deals (see lg2 and lg25). The calculator therefore works with a residual value without a channel, and the curve in lg26 does not distinguish one. Anyone trading in tends to land below it; anyone selling themselves tends to land above it and does the work. Whoever knows their own figure enters it in the field — that is what it is there for.
  29. The official electricity price is a population-weighted median. This site calculates with the figure ElCom reports for Switzerland. Alongside it we recalculate the same figure from the tariff filings of the individual grid operators, published by the same body in the same data portal. The two do not always meet. Measured on 09.09.2026 for the reference profile H4: in tariff years 2016 and 2024 they agree to the third decimal place (20,641 and 32,135), in 2027 they lie 0,027 centimes apart, in 2026 it is 0,196, in 2023 minus 0,686 and in 2011 plus 1,254. The exact matches prove nothing: with around 2’300 filings per year the values sit in plateaus, and in 2016 the official figure is 34 ranks away from our median. Something else is measurable: in each of the seventeen tariff years the official figure is an actually filed tariff, but depending on the year it sits between the 36th and the 54th percentile of our distribution — a median over the same set would show 50 every year. ElCom explained why on request on 11.09.2026: it does not report the median across the filings but the value at the middle of the population. It sorts the municipalities by their tariff, assigns them the population figures, and names the tariff that the middle inhabitant of Switzerland pays. Our median treats Zurich like Vals — the percentiles of 36 to 54 are exactly that weighting. Two further points from the same reply: there is no cut-off date; ElCom recalculates as soon as corrected filings arrive, every few days shortly after publication. And the list of filings can have gaps despite the statutory deadline; ElCom does not check the raw data, responsibility for it rests with the grid operators. In principle the figure could now be rebuilt, but we lack the population figure per municipality — our tariff table carries name and canton, no population. We therefore continue to fetch the official figure instead of recalculating it.
  30. How often a Swiss charging point is actually used We know the stock, not the usage. The open series of the Swiss Federal Office of Energy counts locations, charging stations and plugs — it says nothing about how many charging sessions run on them. Utilisation statistics for Switzerland are not publicly available; the operators hold the data but do not publish it. For Germany they exist: alternating-current charging points there see around 0,6 sessions a day, fast charging points more than 2,5 (Cirrantic for Plusminus, via tagesschau.de, 11.09.2026). We do not transfer those figures. What we show for Switzerland is the supply side: for a year the stock has been shifting towards fast charging. Whether that follows demand we cannot measure.
  31. What the power classes of the charging infrastructure count exactly The description file of the SFOE dataset merely repeats the field name for every field; a definition is not there. We took the class boundaries from the SFOE’s own presentation of the same dataset, and the fact that the classes count plugs from the EnergieSchweiz tools page. Both are documented, but neither is in the file. Checked against the data, the five classes make 98,7 per cent of the plug count, which fits the plug reading; the missing 1,3 per cent are plugs with no reported power. Our retrieval script measures this ratio on every run and aborts if it leaves the band.
  32. What a grid operator sets as the refund tariff for electricity fed back Article 18e StromVV requires every grid operator to set a refund tariff for storage facilities with end consumption. ElCom does not collect it — confirmed on request on 15 September 2026, and neither are the feed-back tariffs; the ElCom tariff data therefore carry only the supply tariff. It is published by the utilities themselves, in their machine-readable tariff files, whose addresses the same data series carries in the field «urltr». That is not a series yet: according to ElCom, by no means all utilities have supplied working links and usable files. The 13,04 centimes/kWh we calculate with are therefore assembled from the components of the household tariff — grid usage plus grid surcharge of the official series 2026 — and are an order of magnitude. That they are assembled correctly was confirmed by ElCom the same day: system services, the power reserve and the measures under Articles 15a and 15b StromVG are already inside the grid usage component; only the grid surcharge is added separately. What your utility sets can be above or below it.
  33. How the share drawn from the grid is determined for a vehicle, and who the refund belongs to Article 14a paragraph 4 letter a StromVG refunds the grid usage fee for electricity that is fed back after being drawn from the grid and stored. How that share is to be determined, the ordinance says only for stationary storage: Article 18g paragraph 1 StromVV compares grid draw and storage every quarter hour and takes the smaller value. For mobile storage, paragraph 3 names a point of measurement — the house connection point — and no such rule of calculation. That an electric car falls under that term was confirmed by ElCom on request on 15 September 2026; the rule of calculation is still missing. Anyone charging from both, sun at midday and grid at night, will not find how the shares are separated. That leaves the two-connection case open as well: anyone who charges at work and discharges at home draws at one and feeds in at the other, and with a company car as two people besides. Who then counts as the operator, the ordinance does not define for mobile storage. We claim nothing on this and carry the case as open.
  34. The Swiss plug-in share cannot be compared across Europe In the dataset «New passenger cars by type of motor energy», Eurostat lists 3,0 per cent plug-in hybrids for Switzerland in 2025. auto-schweiz measures 10,7 per cent for the same period — more than threefold. The error is not Eurostat’s: for Germany the same series matches the official German statistics. In the Swiss return, vehicles with a plug evidently land for the most part in the collective item for hybrids without a plug, which comes to 37,9 per cent for Switzerland and 22,7 for Germany. Where along the reporting path this happens, we have not asked. As long as that holds, this site states no europe-comparable plug-in share for Switzerland: the 25,3 per cent from the Eurostat figures would stand next to 30,0 for Germany and look like a lag that would in greater part be a reporting gap. The fully electric share is not affected — there both sources agree, and that is the only figure used in the country comparison.
  35. The reader poll on purchase intent measures our readers, not Switzerland The box in the «Situation» tab asks whether buying an electric car is planned and shows the distribution across four time windows. It is the only figure on this site that we collect ourselves — and the weakest. Whoever answers chose to do so and is reading a site about electric cars anyway. That is not a sample. The value is extrapolated nowhere, it enters no calculation and appears in no cost curve. Counting is per browser: the browser store holds a random identifier and the choice, and a second click changes the answer instead of adding to it. Anyone returning in a private window, on a second device or after clearing site data can vote again. Against deliberate abuse there is an hourly brake — twelve votes per hour and origin, without the IP being stored — and without a login more is not available. The language of each vote is recorded; what is shown is the sum across all four: the same question, whichever language it is read in. What the figure can do is compare itself over time — and in three, six and twelve months compare itself against the registrations reported by then.
Sources
  • auto-schweiz — registration statistics for passenger cars by brand and model, from the ASTRA register (IVZ), reference date 31.08.2026. Primary source for all registration figures.
  • TCS — fuel prices (unleaded 95 CHF 2.14, as of 19.09.2026) and comparison of fast charging stations in Switzerland.
  • ElCom — official Swiss household tariff, tariff year 2026: 25,68 centimes/kWh for the profile «house with an EV». Taken from the cube electricityprice-swiss via LINDAS, directly and without an intermediary. Across municipalities the range runs from 9,17 to 41,80 centimes/kWh.
  • AutoScout24 — market barometer Q1 2026 and annual data 2025 (partly reproduced via motoro.ch).
  • NZZ, 13.04.2026 — used electric cars, time on the lot and demand.
  • Blick / AutoScout24 — depreciation of individual electric models, survey January 2023 to February 2024.
  • ADAC — breakdown statistics 2026, data year 2025, 158 vehicle series from 27 manufacturers.
  • Geotab — battery condition study 2026, over 22’700 vehicles, 21 brands and models.
  • SFOE — key data on vehicles ↗, alternative drivetrains and charging infrastructure. Note, different figures from ours: The SFOE collects its data while enforcing the CO₂ emission rules for new vehicles and draws some of the system boundaries differently from auto-schweiz. Both series are correct, they measure marginally different things. wattradar uses auto-schweiz throughout; anyone holding one of our figures against an SFOE one is therefore not comparing like with like. The SFOE itself pointed out the divergence and its cause to us.
  • SFOE consumption catalogue — vehicle search ↗. Consumption, range, environmental record and running costs of individual vehicles are published by the federal government itself — the mandatory register under Article 11 of the Energy Efficiency Ordinance, run together with the TCS and auto-schweiz. We do not read that data in: it is not open government data, which the SFOE confirmed on request on 28.08.2026. All 1244 model and 200 brand pages link there — the model pages since 10.09.2026 straight to the vehicle wherever we can map it, otherwise into the vehicle search. This became possible through a vehicle list the TCS drew up at the SFOE’s request; it carries names and identifiers, no figures. The catalogue has no English edition; the link leads to the German one.
  • machelo — daily pump prices for unleaded 95, unleaded 98 and diesel. The figures are reported by users through the TCS app and the machelo app; they are shown only once several people have confirmed them. machelo.ch ↗. We use them for a daily comparison against the TCS measuring points. A different population from the TCS table: reported prices from the pump, not a survey.
  • Avenergy Suisse — commodity quotations, including the Brent closing price. They appear in the energy price index next to petrol and electricity. Exchange quotations, not pump prices: Brent in dollars per barrel, unleaded 95 in dollars per tonne ex Rotterdam. Useless for the price at the pump, useful for the commodity trend.
  • FOCBS — mineral oil tax rates and fuel sales, the basis of the scenario calculator. The tax is a fixed amount per litre, not a share of the price; all assumptions of the calculator lie open in steuerbasis.json ↗.
  • ASTRA/IVZ, new registrations — the «new registrations, monthly» dataset from the vehicle information system, retrieved daily. We read it directly rather than through a publication: it tells us whether a reported figure holds, and how far the current month has been recorded. Different figures from q1, and both are correct: auto-schweiz draws the boundaries marginally differently; the raw extract from the register deviates by a few per mille. Every model and brand figure on this site comes from q1; this dataset sits beside it, not beneath it. Open government data ↗
  • ASTRA/IVZ, changes of keeper — the AHOI dataset, month by month since January 2022. From it we count how many changes of keeper involve electric vehicles and how old those vehicles are. From it, the «Used or new» tab shows five individual years, each covering the same months. Only counts and vehicle age are evaluated: no analysis by age, sex or place of residence of vehicle keepers, even where the source would allow it.
  • Eurostat — the dataset «New passenger cars by type of motor energy». It supplies the European country comparison: the share of purely electric registrations in all new registrations of a year, by country. Annual figures, not monthly — monthly shares are not available officially across Europe, and for context the year is the more robust measure anyway: large fleet orders and model launches distort individual months. Until 05.09.2026 a media report stood there. Dataset ↗
  • SFOE, electricity price series — two series: the electricity price index for end consumers, monthly within the consumer price index, from the SFOE energy dashboard; and the wholesale price «Day Ahead Base» in euros per megawatt hour on trading days, from the open dataset ogd106 (ogd106_preise_strom_boerse.csv, catalogue entry BFE-DS-0087, original source ENTSO-E). We collect both and do not yet show them. The wholesale price does not belong next to a household tariff: that one applies for a calendar year and includes grid, levies and metering. It says something about the following year’s tariffs because the utilities buy ahead — ElCom names procurement costs as the main reason for the 2027 reduction. energiedashboard.ch ↗
  • SFOE, public charging infrastructure — the dataset «Key figures about the charging infrastructure for electrical vehicles». It is derived from the live data of ich-tanke-strom.ch, the federal platform: collected daily, averaged per month, series from November 2020, by canton and for Switzerland. It supplies the locations, charging stations, plugs and the five power classes in the «Charging» tab. A monthly value is an average across the days, not the state on the last day of the month, and the series lags the registration figures by up to a month. Licence: free use, attribution required. Dataset ↗
  • Charging tariffs of the providers — not a data series but a dated snapshot. There is no open source for charging prices in Switzerland; several operators do not publish their prices at all and point to their app. What appears in the «Charging» tab is transcribed from provider pages: the network prices from Swisscharge, the flat tariff of Z Volt at Zurich Switzerland, the range from the TCS. Every line carries its own date and retrieval date, openly readable in ladeanbieter.json ↗. What we have not verified is marked there as «not checked» and not given as a figure.
  • SFOE, charging points for electric cars — the individual locations with coordinates, plug type, charging power and operator, from the federal platform ich-tanke-strom.ch. It supplies the overview map in the «Charging» tab and the calculation of how far it is from a charging location to the nearest fast charger. Unlike the monthly series in q18 this is a snapshot, retrieved daily, not an average. Terms of use, verbatim: «Open use. Must provide the source. Use for commercial purposes requires permission of the data owner.» wattradar uses nothing commercially; what remains is the attribution. 55 of 8’921 points were not in Switzerland on the first run — coordinate errors by the reporting operators; they are filtered out and the number is stated on the page. Dataset ↗
  • PVGIS v5.3, Joint Research Centre of the EU — the solar yields parts 3 and 4 of the electricity series calculate from. Queried for Zurich, south facing, 14 per cent system losses: the long-term mean 2005 to 2023 per tilt angle, plus one year of hourly values. The tilt is the lever: 1’151 kWh per kWp at 35 degrees, 1’145 at 30, 801 vertical on a railing. A single year is not the mean — 2020 was six per cent above it, so the hourly series is scaled to the long-term mean. What becomes self-consumption depends on an assumed load profile; real quarter-hour values are not available, and the field lastprofil_ist_annahme in the file says so. We collect the series and do not yet show it. PVGIS ↗
  • Solar devices on the market — not a data series but a dated snapshot, as with the charging tariffs. There is no open source for plug-in solar prices. What we calculate is read off retailer pages — LANDI, swissbatt24 and Helion — per device the price, output, inverter limit, storage, public address and date read, openly visible in pvgeraete.json ↗. Collected by hand, without a fetch script — a script against an online shop would be scraping and would need that shop’s terms of use. The figures age fast: on 14.09.2026 all thirteen offers of one retailer carried a struck-through price, five could not be ordered, and one set had left the range three days after the first reading. We collect and do not yet show.
  • Eurostat, electricity prices and prosperity — three further series alongside q16, all for the European comparison: the household electricity price per country (nrg_pc_204, half-year values, consumption band 2’500–4’999 kWh a year), gross domestic product per head (nama_10_pc) and the euro exchange rate (ert_bil_eur_m). Switzerland reports no electricity price to Eurostat — its value comes from q3 and is converted for the comparison, at the rate of the same months. The comparison runs excluding value added tax, because that is how ElCom states it; Eurostat’s standard figure includes it and would make Switzerland 8,1 per cent cheaper than it is. All openly available in strompreise_eu.json ↗. Dataset ↗
  • ASTRA/IVZ, vehicle stock — the BEST dataset: one line per registered vehicle, a good two gigabytes. From it we count how many vehicles of a model year are still on the register today — by drivetrain and by model — and hold that figure against the new registrations of the same year. Swiss first registrations and imported used cars are counted separately: just under nine per cent of the lines carry a used-import code, and those vehicles were never a Swiss new registration of their year — counted together they would produce a survival rate above one hundred per cent. The rate measures what is on the register, not what is driving: deregistration, storage, export and scrapping look identical in the source. And it cannot be read without a reference, which is why every figure for electric cars sits next to the same figure for combustion cars. A rate exists only from model year 2016; the archive files before that lack the country column, and before 2011 there is no file. Openly readable in fahrzeugverbleib.json ↗.
  • Manufacturers’ battery warranties — not a data series but a dated survey, as with the charging tariffs. There is no open source collecting warranty terms. What stands here was read off the Swiss manufacturer pages — per brand the term, mileage limit, capacity threshold, address, date read and the source’s sentence verbatim, openly visible in batteriegarantie.json ↗. Four of the fifteen largest brands answer no automated request and stand there as a gap, not as a figure. Volkswagen, Seat, Cupra and Škoda share a single set of terms from AMAG as importer. We collect and do not yet show.

Who does this

wattradar is published by Oncobeam AG in Zug; the imprint carries the full details. There is no editorial team in the usual sense — the analyses are produced in one place. That is precisely why the calculation is laid open here and the gap register sits beside it: what can be checked is not who says something, but how it was calculated. Questions about figures, calculations or reuse of the data go to call@wattradar.ch. Anyone wanting to cite the figures will find a data extract as CSV and JSON with each market report, with a citation note in the file header.

Methodology — How we count and calculate · as of 2026-08-31
Source: auto-schweiz, own analysis of the original Excel files · as of 2026-08-31

Where to next