Healthy / normal wear / worth checking — built from measurable indicators in your own export, not from the manufacturer's hidden state-of-health value.
For every charging session that gained ≥30% SoC while the car was reporting, the tool integrates battery current × pack voltage (mean cell voltage × 96 series cells — the MEB layout) and divides by the SoC gained. The median across sessions is your pack's measured usable capacity. The integration happens at the battery terminals, so charging overhead and any climate or conditioning load while plugged in never enter the figure. The dashboard shows a session-by-session table — SoC window, duration, energy, current coverage, sample count, AC or DC — so you can judge each measurement's reliability yourself; wider SoC windows and fuller coverage make an estimate stronger.
Some newer portal exports carry no battery-current or cell-voltage history. In those files, the vehicle's reported session energy divided by SoC gained is shown only as a descriptive consistency proxy. Its measurement point is undocumented, and charging losses, auxiliaries and 1% SoC rounding all affect the ratio. It is therefore not used as battery capacity or state of health; the dashboard withholds a health verdict. Sessions above the model's largest pack are flagged as evidence of that uncertainty, but remain in the descriptive data.
The nominal pack for the state-of-health percentage is chosen from the sizes your model actually shipped with, detected from the VIN: ID.3 → 45/52/58/77/79 kWh, ID.4/ID.5 → 52/77/79, ID.7 and ID. Buzz → 77/86 — picking the smallest option the measurement can fit, because packs only lose capacity.
The spread between the highest and lowest cell voltage. A healthy pack stays in the single-digit millivolts; a consistently large or growing spread is an early warning sign worth a service check. The dashboard tracks it over time, by state of charge, and by operating state — divergence under load is the earliest weak-cell signal.
Trip consumption is estimated two independent ways — from SoC deltas and from the same current × voltage integration — and both are shown side by side, so quantization noise on short trips is visible instead of hidden. The integration also splits traction from regenerated energy, giving an honest regen-recovery figure.
When battery-current and cell-voltage history is present, merging several exports over time can turn repeated battery-side capacity measurements into a degradation trend. Charging-energy/SoC proxies from exports without those channels are not suitable for this purpose.
Honesty note: even when battery-side evidence is available, this is a diagnostic estimate, not an official state-of-health certificate. Treat "worth checking" as a prompt to investigate, not a diagnosis.