Battery condition
After 5 to 6 years an electric car still has 94 per cent of its battery capacity
An electric car 5 to 6 years old is a car worth buying in Switzerland. Its battery still holds 94,0 per cent of its original capacity in the median, measured across 543 tests in Swiss garages. Anyone who advises against it because the battery might be worn out is advising against the measurements.
Three figures from the same analysis say why that is and where caution belongs. What follows from it for a purchase is set out at the end in three points — with a table against which the age and mileage of an offer can be checked.
1. The loss amounts to 1,1 percentage points a year
Between the first and the 6th year of operation the remaining capacity falls by 4,3 points, spread over 4 yearly steps. The series falls throughout, across 5 age classes with 86 to 128 tests per class.
The calculation uses the median, not the mean, and here that makes 0,40 points a year: the same cut across the means gives 1,5 instead of 1,1. The reason is in the section below — individual battery tests from warranty cases pull every mean they sit in.
| Age in years | Aviloo flash | OEM diagnosis | OEM − Aviloo | all tests | mean |
|---|---|---|---|---|---|
| 0–1 | 97,8015 | 100,0044 | +2,20 | 100,0061 | 93,57 |
| 1–2 | 98,3045 | 99,0037 | +0,70 | 98,3086 | 97,22 |
| 2–3 | 97,3035 | 98,0061 | +0,70 | 97,70112 | 96,11 |
| 3–4 | 94,1550 | 95,0075 | +0,85 | 94,90128 | 94,07 |
| 4–5 | 93,5056 | 96,0047 | +2,50 | 94,10121 | 92,77 |
| 5–6 | 92,1025 | 95,0061 | +2,90 | 94,0096 | 91,32 |
| 6–7 | 88,705 | 95,5030 | +6,80 | 95,0037 | 91,67 |
| 7–8 | 87,701 | 92,0013 | +4,30 | 92,0018 | 89,25 |
| 8–9 | 85,052 | 85,0010 | −0,05 | 85,0012 | 77,18 |
| 9–10 | none | 74,502 | — | 84,703 | 77,90 |
| 10–20 | none | 70,003 | — | 70,003 | 59,67 |
The class under one year stands at 100 per cent in the median and thus above all the following ones — as it must be. In the mean it lay at 93,6 per cent below the next class, which a battery cannot do. The AGVS looked into the individual records for us: they contain isolated warranty cases, a new vehicle with 2 km recorded at 5 per cent. The calculation nevertheless begins with the second year, because 100 per cent is the top of the scale and a curve starting at the ceiling overstates its first step. With that class included it would be 1,2 instead of 1,1 points a year.
2. One year costs as much capacity as 18’000 kilometres
The same analysis also cuts the tests by mileage. There it is 0,61 points per 10’000 km. Set against 1,1 points a year, one calendar year corresponds to roughly 18’000 kilometres.
A Swiss passenger car covers about 12’000 km a year, so 67 per cent of that. Mileage therefore explains the larger share of the annual loss, and the calendar the rest. On the means it looked different: there a year corresponded to 45’000 km, and the calendar dominated. The reversal comes solely from the median not carrying the outliers.
| Kilometres | Aviloo flash | OEM diagnosis | OEM − Aviloo | all tests | mean |
|---|---|---|---|---|---|
| 1–20’000 | 98,6047 | 100,0074 | +1,40 | 99,00140 | 94,68 |
| 20’000–50’000 | 96,15104 | 98,00144 | +1,85 | 96,85264 | 94,51 |
| 50’000–100’000 | 93,0073 | 95,00139 | +2,00 | 94,15232 | 92,04 |
| 100’000–150’000 | 90,258 | 90,0021 | −0,25 | 90,3033 | 90,15 |
| 150’000–200’000 | 89,702 | 100,004 | +10,30 | 95,006 | 94,90 |
| 200’000–800’000 | none | 15,001 | — | 49,852 | 49,85 |
A vehicle with 100’000 km stands at 94,2 per cent in this analysis, a 6 year old one at 94,0. The two lie 0,15 points apart — between driven a lot and parked for long this analysis does not decide. Anyone looking for a car gains nothing here and has to have the battery measured.
3. The manufacturer diagnosis scatters 4,9 times as widely
Both procedures measure the same batteries, and they do so with differing precision. The scatter — standard deviation divided by the mean — is 3,0 per cent for the Aviloo flash test across the classes, and 14,6 for the OEM diagnosis — measured across 9 and 13 classes respectively with at least 10 tests.
The manufacturer diagnosis also measures more coarsely. Of 17 class values, 15 are whole per cent; for the flash test 1 of 14. Anyone holding a manufacturer diagnosis reading 100,0 per cent knows the remaining capacity to the nearest whole per cent, no more.
Which procedure reads higher depends on which figure you take, and that is no quibble. Below 20’000 km the median of the OEM diagnosis is 100,0 per cent, that of the flash test 98,6 — the manufacturer diagnosis 1,4 points higher. In the mean it reverses: 98,1 against 91,7, the flash test 6,4 points higher. Both figures come from the same 140 tests; the difference comes from the warranty cases sitting in the mean.
This is where the risk of a used purchase lies, not in the cell. An SOH figure without the test procedure is not comparable. «OEM procedure» here covers the test procedures of all manufacturers; according to the AGVS the data volume per brand would be too small for a separate analysis.
The AGVS says itself what it recommends to garages. Markus Peter, Technik & Umwelt, on 22.09.2026: the association recommends using battery test reports and certificates «particularly in the used-car business» as well. Standardisation is challenging because most manufacturers rely on the internal methods of their diagnostic applications and those readings also matter for warranty cases. To compare vehicles of different brands it makes sense «to look for cross-brand reports/certificates such as those from Aviloo».
What to do when buying
Three things, in this order.
- Have the battery tested, and in writing. Aviloo flash test or the manufacturer diagnosis — the result belongs on paper or in a PDF, with the date, the mileage and the name of the procedure. A verbal assurance is no assurance, and a percentage without the procedure beside it is not comparable, per the section above. For comparison across brands the AGVS recommends cross-brand certificates.
- Check age and mileage against the table. Two limits, both generous compared with what this analysis measures: at most 2 percentage points lost per year and at most 20’000 km per year. Whoever keeps to both is within range. Whoever falls below has to be able to explain why.
- With old vehicles, talk about the price. What the analysis offers on this is below the table.
| Age | Mileage up to | Lower bound SOH | measured |
|---|---|---|---|
| 2 years | 40’000 km | 96 % | 98,3 %86 |
| 4 years | 80’000 km | 92 % | 94,9 %128 |
| 6 years | 120’000 km | 88 % | 94,0 %96 |
| 8 years | 160’000 km | 84 % | 92,0 %18 |
| 10 years | 200’000 km | 80 % | 84,7 %3 |
Beyond ten years the value collapses, and we cannot say how far. The class from 10 years on stands at 70 per cent in the median and 59,7 in the mean — on 3 tests. That is not a finding, and it is at the same time the only thing this analysis offers on the subject. Both figures lie far below everything younger. Anyone buying a vehicle of this age is buying an unknown battery and negotiates the price accordingly — or has it measured beforehand, which point 1 demands anyway.
Two things that no figure in this analysis captures:
- The previous owner’s charging habits. Which questions help here and what the cell chemistry changes about it is in our checklist for buying used — LFP cells want the full charge, nickel-based ones do not.
- The remaining warranty. First registration plus eight years, and the mileage against the warranty limit. Whichever comes first counts.
Where this data stops
From 6 years on the classes together carry 73 tests, and with that the statement ends. The drop to 85,0 per cent in the ninth year rests on 12 tests and is not a finding. The battery warranty runs eight years; for the last third of that term and for the time after it there is no robust measurement series in Switzerland. The rate of 1,1 points a year applies to one to 6 years and is not extrapolated here — a battery does not age linearly, and 73 tests establish no curve.
Source and calculation
- Data. Auto Gewerbe Verband Schweiz (AGVS), SOHC-Datenbank, analysis from the database monitoring, received on 16.09.2026, supplemented with median and scatter on 22.09.2026. Period March 2026 to July 2026, 677 recorded traction battery tests.
- The calculation uses the median of the remaining capacity in per cent, per class with the number of tests; the mean is in the last column of the tables. The scatter is the standard deviation divided by the mean, as the AGVS reports it. The class boundaries come from the analysis, «up to» is inclusive. No brand — the data volume per manufacturer would be too small according to the AGVS.
- Both rates are our calculation. For the years: 98,30 per cent at 1 to 2 years against 94,0 at 5 to 6, spread over 4 yearly steps. For the kilometres: 99,00 per cent in the class up to 20’000 km against 94,2 in the class up to 100’000 km. Both apply to the classes they come from and are not a forecast.
- The 12’000 km annual mileage is an order of magnitude for comparing the two rates, not a figure from this analysis.
- The «OEM − Aviloo» column is our calculation. Median of the OEM diagnosis minus median of the flash test, in percentage points. A positive value means the manufacturer diagnosis reports more remaining capacity. On the means it would carry the opposite sign — which is why the column heading says what it refers to.
- The «all tests» column is more than the sum of the two procedures. 234 Aviloo and 383 OEM tests give 617 of 677; the remaining 60 come from further procedures that the analysis does not report separately.
- Faintly set rows carry fewer than 10 tests per procedure and no statement in the text.
