The $6,000 EV Residual Value Problem Nobody's Tracking When a fleet writes off the value of a vehicle at end of life, the calculation is usually straightforward. For an internal combustion vehicle, the residual value is a function of mileage, age, model, condition, and market demand. The variables are well-understood, the depreciation curves are well-modeled, and the resale market — auctions, dealer networks, fleet remarketing companies — has decades of price data to draw on. Finance teams plug a number into the TCO model at acquisition, and by the time the vehicle leaves the fleet five to seven years later, the number is usually within a few hundred dollars of correct. EVs break this model in a way most fleets haven't fully reckoned with yet. The single biggest determinant of an EV's residual value isn't mileage or model year. It's battery state of health — the percentage of original capacity the battery retains after years of use. A used EV with a battery at 92% of original capacity is a meaningfully different product from the same EV with a battery at 75%. The lower one charges more slowly, has less range, and signals to a buyer that further degradation is coming sooner. Across the major used-EV remarketing platforms in 2025 and 2026, the price differential between an otherwise identical vehicle at 92% versus 75% state of health is consistently in the range of $5,000 to $7,000 — and the difference grows wider for larger battery packs. Most fleets have no way of knowing where their EVs are landing on this spectrum, in real time, while there's still time to do something about it. By the time the battery state of health number becomes visible, the damage is done and the residual value is set. This article walks through what's actually known about EV battery degradation in fleet use, why the variance is so wide, what the financial exposure looks like at fleet scale, and what it would take to start tracking this proactively rather than discovering it at remarketing.

What we know — and what the data is just starting to show For most of the last decade, fleet operators running EVs were doing so largely on faith. Battery degradation was understood to happen, but the public datasets were small, the manufacturers were cagey about real-world numbers, and the variance between vehicles in identical fleets was treated as anecdotal. That's changing. In January 2026, Geotab published the largest study of EV battery degradation conducted to date — a longitudinal analysis of 22,700 fleet vehicles across multiple manufacturers, vehicle classes, and use patterns. The headline number was a fleet-wide average of 2.3% annual state of health degradation, which is broadly consistent with what manufacturers had been suggesting in their warranty disclosures. The more interesting finding wasn't the average. It was the variance. The Geotab study explicitly noted "enormous variance" in degradation rates across vehicles that were, on paper, comparable. Vehicles operating in the same fleet, on similar duty cycles, in similar climates, were degrading at meaningfully different rates. Some vehicles were tracking close to the 2.3% average. Others were degrading at 1.5%/year. Others were closer to 3.5%/year. Across a five-year operating life, the difference between those rates compounds into wildly different end-state battery health numbers. The study's authors were transparent about why they couldn't fully explain the variance: the explanatory variables they didn't have access to. Telematics platforms can observe a great deal about how a vehicle is driven, but the variables that drive battery degradation — frequency of DC fast charging, time spent dwelling at high state of charge, exposure to temperature extremes during charging specifically, the rate of energy delivery during each session — live in charging session data, not in telematics. The Geotab dataset, despite being the largest of its kind, hit the limit of what telematics alone could explain. For fleet finance leaders reading this, that's a useful technical detail with a very practical implication. The reason your EV residuals are likely to surprise you in the next remarketing cycle is the same reason this enormous study couldn't fully explain the variance it found: the data needed to predict and manage battery degradation is in a different system than the data your fleet is currently looking at.

Why the variance is so wide Battery degradation isn't random, but the inputs are dispersed across multiple data sources. The factors that consistently emerge in the literature as drivers of accelerated degradation include:

DC fast charging frequency. Fast charging delivers high-amperage current to the battery cells, which generates heat and accelerates the chemical processes that cause capacity loss. Vehicles that fast-charge multiple times a week degrade meaningfully faster than vehicles that primarily Level 2 charge overnight. Time at high state of charge. Lithium-ion batteries are stressed when held at or near 100% state of charge for extended periods. A vehicle that's plugged in to "top off" overnight after every shift, sitting at 100% for eight hours, is putting more wear on the battery than one that finishes charging at 80% and dwells there. Time at low state of charge. The opposite end of the spectrum is also harmful. Vehicles consistently allowed to deplete to single-digit state of charge before being plugged in degrade faster than vehicles maintained in the middle of the charge band. Charging temperature. Charging in extreme cold or extreme heat — and especially fast-charging at temperature extremes — is significantly more degrading than charging in moderate conditions. Climate-controlled garage charging is materially better for the battery than charging at a public DC fast charger in a 35°C parking lot in July. Cumulative energy throughput. Independent of any of the above, the total kWh that has flowed in and out of the battery over its lifetime is a degradation factor. A high-mileage vehicle is going to degrade faster than a low-mileage one.

What makes this hard to manage is that none of these variables are available in a single dataset. Charging frequency and rate live in charging network session records. Dwell time at high state of charge requires correlating charging session end times with telematics state-of-charge readings. Temperature at the time of charging requires charger-side weather data or telemetry from the vehicle itself. Energy throughput requires summing across all charging sources — depot, public, home — for each vehicle. In other words, the explanatory data for battery degradation is precisely the kind of data that requires correlation across multiple systems to surface. No single system holds it. And in most fleets today, no system is bringing it together.

The financial exposure, at fleet scale Let's put numbers to this. Take a fleet of 100 EVs, with an average original purchase price of $48,000, an expected operating life of five years, and a planned residual value at end of life of 35% — so $16,800 per vehicle, $1.68 million across the fleet, assuming all vehicles land on the planned residual. Now apply the variance the Geotab study identified. If the fleet average degradation is 2.3% per year, the fleet-average state of health at end of life will be around 88.5%. That's the planning model number. That's what your TCO worksheet assumed at acquisition. But the fleet doesn't degrade as a single average. It degrades across a distribution. If the variance is anything like what the Geotab data suggests, the fleet might break down something like this at end of life:

30 vehicles at 90%+ state of health (close to planning model) 50 vehicles at 82-89% (moderately below planning model) 20 vehicles at 75-82% (significantly below planning model)

Apply realistic remarketing prices to this distribution, and the financial impact is concrete. The 30 high-state-of-health vehicles command roughly the planned residual or slightly above. The 50 mid-range vehicles take a $2,000-$3,000 haircut each. The 20 low-state-of-health vehicles take a $5,000-$7,000 haircut each. Run the math on a 100-vehicle fleet with that distribution and the residual-value shortfall versus the planning model lands somewhere between $200,000 and $300,000. On a fleet that was projecting $1.68 million in recovered residual value, that's a 12-18% miss. Scale this to a 500-vehicle fleet — well within the range of operators reading this — and the exposure is over a million dollars. The further uncomfortable part is that this calculation is conservative. It assumes you remarket the vehicles individually, with full price discovery. In practice, fleets often remarket in bulk to specialist EV remarketers, who price the entire batch off the worst-performing vehicles. A handful of severely degraded vehicles in a remarketing batch can drag down the price recovery on the entire batch.

Why this is invisible until it isn't The reason this exposure isn't on most fleet finance dashboards today is straightforward: nobody is tracking the input data in a form that makes the financial implication visible. Battery state of health, where it's exposed by the OEM at all, is treated as an operational metric, not a financial one. It shows up — when it shows up — on a fleet manager's vehicle detail screen, alongside tire pressure and oil change interval. It's not in a finance report. It's not in a residual value model. It's not in the TCO update the fleet controller produces for the CFO each quarter. Charging behavior data — the explanatory variables that drive degradation — is in a charging network's portal, scattered across multiple networks if the fleet uses multiple, and sitting in the depot's charging management system for private charging. Nobody is correlating this data to state of health and asking which vehicles are heading for an expensive surprise. The result is that a fleet can be twelve months away from a $300,000 residual value miss and have no early warning signal anywhere in its existing reporting. The miss surfaces at remarketing, when the resale price is set, and at that point there is nothing to be done.

What proactive tracking would actually look like If a fleet wanted to bring this exposure into view — to see the trajectory of each vehicle's residual value while there was still time to influence it — what would that require? The answer is a relatively well-defined set of capabilities, none of which is exotic, but which has to be brought together in one place:

Per-vehicle state of health tracking, over time. Not a snapshot — a curve. Each EV's state of health, sampled monthly, plotted against the fleet average. Vehicles diverging meaningfully from the average flagged early.
Per-vehicle charging behavior summary. For each vehicle, what percentage of its charging is DC fast versus Level 2? What's its average state of charge at the start of charging sessions, and at the end? How much time does it spend at >90% state of charge? How does it compare to vehicles in the same fleet?
Correlation between charging behavior and degradation rate. Within the fleet, identify which charging patterns correlate with faster degradation. This is a regression analysis, but a fairly tractable one with enough vehicles and enough data history.
Per-vehicle residual value projection. Translate the projected end-of-life state of health for each vehicle into a projected residual value, using current used-EV market data. Compare to the planning model. Surface the variance per vehicle and in aggregate.
Actionable interventions. For vehicles trending toward the bottom of the distribution, what changes would slow the degradation? Reassigning the vehicle to a duty cycle with less DC fast charging? Adjusting the charging schedule to finish at 80% rather than 100%? Avoiding particular drivers or routes that correlate with worse outcomes? These interventions are only possible if the data surfaces the trajectory while there's still time to change it. None of this requires technology that doesn't exist. It requires data from telematics, data from charging networks, and data from utility bills, brought together with state of health readings and a residual value model on top. The work is structural rather than novel — it's a question of whether anyone is doing it for your fleet.

The CFO conversation this enables The reason this matters for fleet finance leaders specifically is that it changes the conversation with the CFO from one about hope to one about management. Today, when a CFO asks "how confident are we in the EV residual value assumptions in our TCO model?", the honest answer at most fleets is "we used the manufacturer's projection, and we'll find out at remarketing." This is uncomfortable for everyone involved. The number is large enough to matter to the financials. The uncertainty is large enough to make the number unreliable. And the controller has no way to tighten it, because the input data isn't in any system the controller has access to. The conversation that becomes possible with cross-system data in place is meaningfully different. "Of our 100 EVs, 30 are tracking at or above planning-model state of health. 50 are within a normal variance band. 20 are trending toward a degradation rate that would put them well below the planning model at end of life. The aggregate exposure if we don't intervene is approximately $X. Here's what we're doing about the bottom 20." That's a conversation a CFO can engage with. It moves the residual value question from a faith-based input in the TCO model to a managed line item in the operating plan. It also creates the possibility of intervention — reassigning vehicles, changing charging behavior, adjusting depot infrastructure — while there's still time for the intervention to matter.

Closing thought The $6,000 residual value swing between a 92% state-of-health battery and a 75% one isn't a hypothetical. It's a market reality, documented in current used-EV pricing across the major remarketing platforms. The variance isn't a hypothetical either — the largest battery degradation study ever conducted just confirmed it, and explicitly noted that the explanatory variables for the variance live outside the dataset that confirmed it. For fleets running 50, 100, 500 EVs, the aggregate residual value exposure is large enough to be a meaningful financial risk — large enough that it should be on the fleet finance dashboard with the same prominence as fuel cost variance or maintenance cost variance. Today, at most fleets, it isn't. The reason it isn't is the same structural reason this article keeps circling back to: the data needed to surface the exposure lives in multiple systems that don't talk to each other. Bringing those systems together is the precondition for making the exposure visible. Making it visible is the precondition for managing it. The fleets that figure this out first will be the ones whose remarketing cycles five years from now don't surprise their CFOs. The ones that don't will be the ones learning, one vehicle at a time, that the planning model's residual assumption was off by more than the operational savings the EVs delivered.