Anatomy of a score: how battery health is really measured

Anatomy of a score: how battery health is really measured

It's tempting to think a battery's health is a single number. You ask the car "how are you doing?", the battery's brain — the BMS — replies "I'm at 92%", and that's that. Case closed.

The trouble is that 92% is the patient's own opinion, and a BMS estimates capacity with its own internal models, which can drift over the years. Sometimes it's spot-on. Sometimes it's optimistic. And sometimes it hides a detail that changes everything: one weak cell among hundreds of healthy ones, enough to drag down the whole pack even when the average looks fine.

That's why the SOHpro Score doesn't settle for a single reading. It cross-checks six different signals, each with a concrete physical root. Let's pop the hood.

1. The BMS SOH — the lead witness (43%)

The capacity reported by the battery management system is the starting point, and that's why it carries more weight than any other signal: 43% of the score. It's the car's best estimate of how much usable energy it has left compared to when it left the factory.

Where does that loss come from? Mostly from two processes. The first is loss of lithium inventory: the growth of the SEI layer on the anode keeps trapping lithium that stops circulating — a mechanism the literature points to as dominant in aging (Keil et al., 2016) — it's the "calendar clock" we take apart in the two clocks of aging. The second is loss of active material: with each cycle, the electrode particles crack under the mechanical stress of swelling and contracting, and stop taking part. The BMS SOH is the net result of both — but it's a witness, not a judge. Just as you wouldn't sign for a house only because the owner swears the roof is fine, SOHpro takes that statement and checks it against the other five signals.

2. The voltage spread between cells, the ΔV (22%)

A traction battery isn't one block: it's hundreds of cells wired in series, working as a team. And here an unforgiving physical law applies: in a series chain, the weakest cell limits the whole pack, the way the lowest stave on a barrel sets how high you can fill it with water. It's the "weakest-link effect", and it's exactly why packs carry balancing systems (EL-CELL, technical explainer).

When the cells are healthy and balanced, they all hold an almost identical voltage. When one starts to diverge — from higher self-discharge or lost capacity — it drops off the pack. That gap between the highest and lowest cell, the ΔV, is one of the most recognized indicators of imbalance and of cells aging badly (Journal of Energy Storage, Elsevier), and a pretty average SOH can hide it completely. That's why it weighs 22%.

"The SOH tells you the class average. The ΔV tells you whether there's a student about to flunk and drag everyone down with them. Without that second signal, you're reading half the story." — Alonso Aguilar, Founder & CEO

3. The rate of degradation (20%)

Knowing a battery is at 85% is useful. Knowing how fast it got there is gold. A car that lost 15% over five years is nothing like one that lost it in a single year.

And there's a deeper reason to watch the rate: not every battery travels its long plateau at the same pace. One that suddenly degrades faster than it had been is signaling that something — heat, heavy fast-charging, or a defect, the prime suspects in the region — is stressing it, long before today's number gives it away. (How those factors weigh in Latin America, in tropics, mountains and fast-charging.) When there are at least two scans of the same VIN spaced out in time, SOHpro computes the real slope: how much health is lost per thousand kilometers. That trend weighs 20% and is the one signal you can't fake with a single good measurement day — it needs history. It's also the reason your electric's passport gains value with every visit.

4. Energy efficiency (5%)

When a battery charges and discharges, not all the energy that goes in comes back out: some is lost as heat to internal resistance. Comparing how much energy went in against how much came out — the round-trip efficiency — gives a clue to how healthy the chemistry still is. Internal resistance that climbs over the years is a sign of aging. It's a cross-check signal: it weighs little (5%), but it helps confirm or cast doubt on what the others are saying.

5. The consistency between voltage and charge, the OCV (5%)

Every battery chemistry has a "signature": a predictable relationship between a cell's resting voltage and its charge level. That curve is called OCV (Open Circuit Voltage). When the measured voltage matches the signature expected for that chemistry, all is well; when it doesn't, something is shifting inside the cell.

Here's a fascinating nuance SOHpro accounts for: not all chemistries are equally legible. LFP (lithium iron phosphate) batteries, very common in Chinese EVs like the BYD Blade, have an extremely flat OCV curve across the mid-charge range. That makes them more stable and safer, but also harder to read by voltage — tiny voltage changes correspond to large changes in charge. Reading the OCV correctly demands knowing which chemistry you're looking at, not applying the same yardstick to all of them. (Why that flat curve also changes how you should charge an LFP, in charge to 100%, yes or no?.) Worth 5%.

6. Temperature uniformity (5%)

A healthy battery heats up evenly. When one sensor reads several degrees above the rest, it usually points to a stressed zone — a hot spot. And heat is no innocent bystander: it exponentially accelerates the reactions that age the cell, so a hot spot ages faster right there, worsening the imbalance we talked about in signal #2. The difference between the hottest and coldest sensor enters the score at 5%.

Why the weights rebalance

Here's a detail that matters: not every scan brings all six signals. A quick diagnostic may not include per-cell temperatures, or it may be the car's first scan and therefore have no history to compute degradation.

SOHpro doesn't invent the missing data. Instead, it redistributes the weights among the signals that are available, and tells you how much confidence it had when it computed the score. It also runs a battery of integrity checks — that the cell count is coherent, that temperatures sit within a real physical range, that there are no contradictory signals — and flags anything that doesn't add up. It's the difference between a number that fakes certainty and one that's honest about what it knows and what it doesn't.

"We'd rather say 'we computed this with four of six signals' than pretend to a precision we don't have. Trust isn't earned by inflating the number; it's earned by showing how it was made." — Alonso Aguilar

From number to grade

With the six signals cross-checked, the SOHpro Score lands on a 0-to-100 scale. But a loose number doesn't say much to most people, so that score is translated into a letter grade, from A+ to D — the same logic as a school report card, easy to read at a glance.

The letters aren't arbitrary; they're anchored to two honest references. The lower cutoffs follow the warranties: nearly every brand promises at least 70% health at 8 years or 160,000 km (US federal rule, 40 CFR 86.1815-27), so landing in D means you're below what the manufacturer itself committed to cover. The upper cutoffs follow what's happening on the street: according to fleet data from Geotab, the vast majority of healthy EVs in their first 100,000 km live above 88, so the boundary between A+, A and B+ is the one that separates "almost new" from "well-cared-for used" — exactly what moves resale value, and what an insurer will one day look at. It's a scheme that's half empirical, half practical; there's no paper claiming these exact numbers, and we say so up front.

And here it's worth clearing up a common misunderstanding: the letter is an absolute snapshot of today's capacity, not a verdict on whether your car aged well. A high-mileage used car that lost, say, 10% — but did so evenly and within what's expected — is perfectly healthy for its age, even if its letter comes in below a brand-new car's. Whether that wear is normal for its age and mileage is exactly what signal #3, the rate of degradation, measures. In other words: the letter tells you where you are; the slope, whether you're doing fine. That's why knowing your slope matters so much if you have an EV under warranty: good batteries, big warranties.

Behind that simplicity sits an engine that cites the source of every threshold — chemistry by chemistry — so that a technician, a buyer or, someday, an insurer can audit where each conclusion came from. A single number can fool you. Six cross-checked signals that land on a clear letter, with the physics of each one in plain sight, are much harder to argue with. That's the anatomy of a score you can trust.

---

Curious to see the score of a real EV? Explore a sample report and watch the six signals in action.