Modeled or Measured?
Across 150 electric vehicles, model-based scores overstated directly-measured battery scores by 6.2 points
| Cohort | # of vehicles | Direct-measured score | Model-based score | Avg. overstatement |
|---|---|---|---|---|
| ≤ 2024 | 121 | 88.9 | 95.8 | +6.9 |
| ≤ 2023 | 88 | 87.1 | 94.8 | +7.8 |
| ≤ 2022 | 53 | 85.1 | 93.7 | +8.6 |
| ≤ 2021 | 32 | 83.9 | 93.2 | +9.3 |
| All | 150 | 90.2 | 96.3 | +6.2 |
Summary
Dealers relying on model-based scores are vouching for a guess.
The battery is the most expensive component in an EV, which makes battery health the deciding factor when buying used. Dealers increasingly publish an independent third-party score on their Vehicle Detail Pages — but the integrity of that number varies enormously.
Many dealers depend on model-based estimates: statistical predictions of battery health derived from aggregated data about similar vehicles, without ever measuring the specific vehicle in front of them.
Lyteflo analyzed battery degradation across 150 vehicles using its certification test, which takes a direct onboard diagnostic measurement, and compared the results against the same vehicles' "range score" from a well-known model-based product.
The findings
The model-based approach overstates scores on every cohort.
Across all 150 vehicles, model-based estimates exceeded direct measurements by 6.2 points on average. They read higher on 92% of vehicles — 138 of 150. The largest single spread was 18 points, on a 2022 Tesla Model Y that scored 95 model-based against 77 measured.
+6.2
average points the model-based score overstated, across 150 vehicles
92%
of vehicles where the model-based score read higher
+18
largest single-vehicle spread
What the score represents
A range estimate, read as a health score.
Model-based estimates are published as range scores. A direct measurement reports state of health — the actual degradation of the battery's cells. Dealerships and shoppers routinely read the former as the latter, treating a range prediction built from external assumptions as a statement about the condition of the pack.
Cell-level condition is invisible to a model.
Battery problems usually start at the cell level — a single weak cell, or a growing imbalance — long before they show up as lost range. Cell voltages and cell-to-cell variance can only be read directly from the battery management system. No model can derive them from mileage, age or climate data, which leaves it blind to exactly the defects that cause post-sale failures.
The bottom line
An optimistic guess becomes the dealer's exposure.
Model-based scores are predictions, not measurements. Across 150 vehicles they ran 6.2 points optimistic, with individual gaps as wide as 18 points — enough to move what a vehicle is priced at and what a buyer is told about it.
Disclosure & liability
What the record says about battery condition at the point of sale is the dealer's responsibility. An optimistic third-party estimate does not transfer that liability if the buyer later disputes it.
Appraisal & acquisition
An inflated score overstates battery health at trade or auction, and the overpayment risk lands on whoever acquired the vehicle.
Cell faults & comebacks
A model cannot detect a failing or imbalanced cell. Those defects surface after the sale as buyer complaints — precisely the risk direct measurement removes.
Certify what you can measure.
Lyteflo measures battery health directly from the vehicle, so you can sell on facts instead of estimates.
This report reflects vehicles tested by Lyteflo and is not a nationally representative sample; results may not generalize to all vehicles, makes, or model years. References to model-based scoring describe a general methodology, not any specific company or product. Provided for informational purposes only; not legal, financial, or appraisal advice.