In late March, German technical inspection and certification organisation TÜV Nord and vehicle diagnostics company Carly released a report into battery state of health (SoH) across a large sample of EVs over a broad age range. Initially, their analysis of around 50,000 SoH readings delivers a reassuring headline: EV batteries are more durable than public perception suggests, with a median SoH of 96 percentage points and only 9.9% of vehicles falling below the 85 point threshold beyond which performance loss increases and both range and charging behaviour also tend to deteriorate. For an industry still working to counter range anxiety and resale value fears, this is welcome validation. But for developers, the real value lies in the granular trends and the questions they raise, writes Peter Donaldson.

This report has weight because Carly brings access to real-world battery readings from thousands of vehicles from different brands collected via their deployed base of diagnostic tools, while TÜV Nord brings independent, authoritative analysis and validation of the data.

The most striking finding is the degradation tipping point at about 90,000 km. Below this distance, SoH declines at a moderate 0.7 points per 10,000 km; above it, the rate more than triples to 2.3 points per 10,000 km. This non-linear behaviour suggests that cumulative electrochemical stress – lithium depletion, solid-electrolyte interphase growth and rising internal resistance – may be reaching a critical threshold where degradation accelerates. This signals that battery and thermal management strategies optimised for early-life performance may not be sufficient over the full vehicle lifecycle. Adaptive battery management algorithms and more robust cooling architectures could prove decisive in flattening the degradation curve.

Equally notable is the convergence of SoH performance across EV brands in newer model years, with most now clustering between 97 and 100 points. This indicates maturing of the supply chain and thermal management standardisation. However, the significant spread in older models – up to 10 SoH points between manufacturers – serves as a reminder that early design choices, particularly around cell chemistry, thermal management and usable capacity windows, have long-term consequences for vehicle service life. This underscores the value of field data in validating laboratory cycle-life models.

The study’s use of mileage rather than charge–discharge cycles as the primary degradation metric is not ideal, but is understandable because it is familiar to vehicle buyers and dealers alike. Mileage offers a transparent and consistently available variable for market analysis, but it is an imperfect proxy. Two vehicles with identical mileage can have vastly different cycle counts, depth-of-discharge profiles and fast-charging histories, all of which fundamentally affect battery ageing. However, charge–discharge cycles are not consistently logged or accessible across different manufacturers’ electronic control units, especially for older vehicles. Diagnostic tools can read SoH, but full cycle history is often fragmented or unavailable.

For battery system developers in particular, this highlights a persistent gap: the lack of standardised, manufacturer-agnostic access to detailed usage metadata. Without it, population-level studies risk obscuring the very variables that need to be optimised. Finally, the finding that annual mileage was comparable across brands is useful for cross-manufacturer benchmarking, but it should not be misinterpreted as evidence that usage patterns are uniform. Real-world charging behaviour, ambient climate and duty cycles remain critical for understanding battery life.

As e-mobility matures, studies like that of Carly and TÜV Nord provide confidence-inspiring validation. But for engineers tasked with designing the next generation of more durable, safer and more sustainable battery systems, the real work lies in interrogating the data behind the headlines – and building the instrumentation and telemetry frameworks that will allow the industry to move from mileage-based proxies to true cycle-informed lifecycle engineering.