Every electric truck reports a state-of-health percentage, and fleets reasonably treat it as the number to manage by. A 2026 study that monitored 1,114 electric vehicles across five manufacturers for roughly 375 days suggests that confidence is misplaced. The correlation between reported state of health and actual measured capacity ranged from non-existent to weak across every platform examined, and among the worst-performing vehicles in the sample, 93% were not identified as worst by their own battery management system. If that pattern holds for your platform, the packs you most need to find are precisely the ones your dashboard is least likely to show you — and the fix costs nothing but a habit. See capacity trended per truck instead.
Managing Battery Health by a Number the Vehicle Made Up
For fleet managers, EV programme leads and workshop supervisors: why reported state of health is a weak signal, what to measure instead using data you already collect, and which parts of received wisdom about charging the evidence actually supports.
The Gap Between Reported and Real
The study audited BMS-reported state of health against measured capacity. The correlations are the finding, and they are not encouraging.
Reported correlations between actual capacity and BMS state of health were ρ=0.10 for one E-GMP platform (not statistically significant), 0.17 for Niro and Kona, and 0.24 for commercial vehicles. One platform reached 0.62, but only under strict filtering — unfiltered, that figure reversed to −0.18. On the MEB platform, state of health was not exposed through OBD-II at all. The study also found real capacity differences of 12–25% between identical models, with the largest gap, 24.7%, among commercial vehicles.
Treat this as one study rather than settled science — it is a 2026 preprint, and its platforms are light-duty and commercial van rather than Class 8. But the direction is consistent across five manufacturers, and the practical response costs nothing: measure capacity yourself and use the reported figure as a cross-check rather than as the answer.
What to Measure Instead
All five of these use data a fleet already generates. None require OEM cooperation, a diagnostic subscription or a laboratory.
- Energy consumed per mile, per unitThe most honest proxy you own. It needs no OEM cooperation, it uses data you already have, and a pack losing capacity shows up in it before any warning appears.
- Usable energy on a full chargeMeasured, not reported. Charge to the same point, run a known route, and record what it took. Repeat quarterly and you have a real curve.
- Range achieved against range requiredThe only comparison that decides anything operationally. A pack at 88% health is fine on one route and unusable on another.
- Charge acceptance over timeHow long a full charge takes, at the same charger, from the same starting point. A pack that is slowing down is telling you something before capacity does.
- Variance across identical unitsThe study found 12–25% real capacity differences between identical models, and 24.7% among commercial vehicles. Comparing your own units against each other exposes outliers a single reading never would.
The last one is the most powerful and the most overlooked. A single reading has nothing to be compared against, but a fleet of identical trucks running identical routes is a built-in control group — and a 24.7% spread between supposedly identical commercial vehicles means the outliers are there to be found if anybody looks.
What the Evidence Says About Charging Habits
Two findings here confirm common advice and two contradict it. Reporting only the convenient half would be the easy thing to do and the wrong one.
Associated with 10–13% worse energy efficiency and 2–5% lower capacity than low-SOC driving
Stop charging to full overnight unless the next route genuinely needs it30–50% was associated with better health outcomes across every platform in the study
Where a pack sits between shifts is a decision you control and mostly do not makeNo significant correlation with health in this dataset
Contradicts common advice — modern packs are actively managed, which likely explains itWeak or inconsistent associations
Also contradicts common advice. Worth watching rather than designing your whole operation aroundThe state-of-charge finding is the actionable one, because depot charging is entirely within your control. A truck plugged in at 18:00 and sitting at 100% until 05:00 spends eleven hours at the state of charge associated with the worst outcomes, for no operational benefit whatsoever. Charging to the level the next route actually needs, finishing shortly before departure, costs nothing and is the single easiest change available.
Do you know which of your packs is worst?
If the answer comes from a reported percentage, the evidence suggests you probably do not. In thirty minutes we will set up energy-per-mile and usable-energy tracking on your units so the comparison is measured, and so the outlier shows itself rather than hiding behind a healthy-looking number.
How FleetRabbit Handles This
Four things, and the second is the one that turns a fleet into its own diagnostic instrument.
Energy per mile and usable energy stored as values per unit, so you are trending what the pack actually does rather than what it claims.
Identical trucks on identical routes should perform identically. When one drifts, the fleet is the control group that reveals it.
Health only means something against the energy a route needs. Holding both makes reassignment a decision rather than a discovery.
A claim needs a documented history, not a screenshot. Readings captured on a schedule are the file you will wish you had.
The Maintenance That Actually Exists
Battery packs are frequently described as maintenance-free, which is true only in the sense that there is nothing to lubricate. There are real scheduled tasks, and most of them are safety-critical rather than performance-related. Swipe on mobile.
| Task | What it involves and why | When |
|---|---|---|
| High-voltage cable and connector integrity | Visual inspection for chafing, heat discoloration, corrosion and secure seating, with the system de-energised by a qualified person. Connectors carry very high current and a poor connection makes heat exactly where you cannot see it. | At scheduled PM, by a qualified technician |
| Insulation resistance | A measured value with a pass threshold, checking that the high-voltage system remains isolated from the chassis. It is the single most important electrical safety check on the vehicle and it produces a number you can trend. | At scheduled PM, and after any suspected fault |
| Battery thermal loop | Coolant level, condition and leaks on the pack's own cooling circuit, plus pump and fan function. It has its own fluid and its own service interval, separate from anything on the drivetrain. | Per the manufacturer's interval |
| Pack enclosure and underbody | Impact damage, corrosion and seal integrity on the enclosure. Refuse and municipal duty puts the underbody through debris, kerbs and corrosive runoff, and a breached seal is a water-ingress problem in a sealed high-voltage box. | At every PM, visually |
| Mounting and torque | Pack mounts carry significant mass through every pothole. Torque checks on a schedule, to the manufacturer's figures, recorded as done. | Per the manufacturer's interval |
| Cooling for the charger side too | On-board charger and DC-DC converter cooling is often shared with the pack loop and often forgotten, because it looks like part of the electrical system rather than something with a fluid in it. | With the thermal loop |
Every item above involving the high-voltage system needs a qualified technician, the right insulated tooling and a documented de-energising procedure — that is not a formality, it is the difference between a routine inspection and a fatal one. Follow the manufacturer's service information for your specific vehicle rather than a generic schedule, and record who did the work as well as what they found.
Warranty, End of Life and What Comes After
Battery warranties are written against capacity thresholds, which makes everything above directly financial rather than merely interesting.
A claim requires evidence that capacity has fallen below the threshold in your agreement, and that evidence has to be credible. If the manufacturer's own reported figure is the only record you hold, and that figure is a weak proxy for real capacity, you are in the position of arguing a warranty case using the counterparty's instrument. A documented history of measured capacity, captured on a schedule from the beginning of service, is a materially stronger position — and it costs nothing to build if you start when the trucks arrive.
At the other end, the DOE notes that an electric vehicle battery "could have at least 70% of its initial capacity left at the end of its useful lives" and may serve "another 10 years or more" in secondary applications such as stationary grid and backup power. End of vehicle life is not end of battery life, and a pack with a documented history is worth more to a second-life buyer than an identical pack with none.
Check the specific thresholds, terms and duration in your own warranty rather than assuming a standard. Heavy-duty commercial terms vary considerably between manufacturers and are frequently negotiated, which is another reason to hold your own measurements from day one.
Questions EV Fleets Ask
Can we trust the state of health our trucks report?
Use it, but do not rely on it alone. In a 2026 study of 1,114 vehicles across five manufacturers, the correlation between reported state of health and measured capacity ranged from non-significant to weak, and 93% of the worst-performing vehicles were not flagged as worst by their own system. Measuring energy per mile yourself costs nothing and gives you a second opinion. Start logging capacity free.
Should we stop charging to 100%?
Where the route allows it, yes. The same study associated time at high state of charge with 10–13% worse energy efficiency and 2–5% lower capacity, and time at 30–50% with better outcomes. A truck sitting at full charge overnight gains nothing operationally and spends those hours in the least favourable state. Charge to what the next route needs, finishing near departure. Plan your charging windows.
Does fast charging damage the battery?
The honest answer from this dataset is less than commonly assumed — fast-charging exposure showed weak or inconsistent associations with health. That is one study and not a licence to fast-charge everything, but it does suggest that building an entire operation around avoiding it may be solving the wrong problem while high resting state of charge goes unaddressed. Track both and see.
What about cold or hot weather?
Ambient temperature showed no significant correlation with health in this dataset, which likely reflects that modern packs are actively thermally managed. Temperature still affects available range on the day, which is an operational matter rather than a degradation one — plan routes for winter range, but do not assume climate is silently destroying the packs. Review your winter planning.
How do we know when a pack needs replacing?
When it can no longer do the work, which is a different threshold from the warranty one and usually arrives first. Compare measured usable energy against the energy your routes actually require, with a margin. A pack can be comfortably inside warranty and unable to finish a long route, and that is an operational failure before it is a commercial one. Compare health to route demand.
Is our data good enough to do this?
Almost certainly. Energy consumed and distance travelled are enough to start, and most fleets already collect both through telematics. The precision matters far less than the consistency — a slightly rough measurement taken the same way every quarter beats an accurate one taken once. Check what you already have.
Does this change the rest of our maintenance?
No — brakes, tires, suspension, body and every inspection obligation continue unchanged, and battery health sits alongside them rather than replacing anything. What it does change is planning, because capacity fade is the one degradation curve on the vehicle that determines which routes it can still run. It belongs in the maintenance programme as a recorded measurement like any other, sitting next to the readings you already take rather than in a separate EV spreadsheet nobody opens. Add it to your schedule.
Measure the Pack. Do Not Take Its Word For It.
FleetRabbit trends measured energy per mile and usable capacity per unit, compares identical trucks against each other so outliers surface, and holds pack health beside the route demand that decides whether it still matters.
