Most fleet operators who manage electric two-wheeler assets think about battery wear in terms of charge cycles. The pack is rated for 500 cycles, you have completed 300, you have 200 left. It is a simple mental model, and it is not wrong exactly, but it is imprecise in ways that end up costing money or causing operational surprises.
State of Health is the metric that gives you a more accurate signal. SoH is expressed as a percentage representing the pack's current usable capacity relative to its rated capacity when new. A pack with an SoH of 85% delivers 85% of its rated range under the same conditions as a new pack. As SoH declines, range decreases, and at some threshold, the pack starts behaving unpredictably: it may report adequate charge but then drop voltage suddenly under load, a problem called the cliff effect that is particularly unpleasant for a rider in the middle of a delivery corridor.
Why Cycle Count Is Not Enough
Two packs with identical cycle counts can have very different SoH values depending on how they were used and stored. A pack that was repeatedly charged from near-zero to near-full (deep cycling) will degrade faster than one that was kept between 20% and 80%. A pack that sat at full charge in 38-degree Pune heat for extended periods will have accelerated calendar aging even without usage. A pack that was occasionally overloaded when a rider used a vehicle with a higher-draw motor than rated will have localized electrode wear that cycle count cannot capture.
In our Pune pilot through Q4 2025 and Q1 2026, we track SoH per pack through voltage curve analysis during each charge cycle. As a pack charges, the voltage rise rate across the cell reflects how much usable capacity remains. By analyzing this curve against the pack's initial baseline profile, we can calculate a reasonably accurate current SoH without additional specialized testing equipment. This is important for operational scale: we run 12+ stations across Pune's delivery corridors, and we cannot send every pack to a lab for periodic capacity testing.
What SoH Tracking Changes Operationally
When you know a pack's SoH, several fleet management decisions change. First, inventory allocation. A pack at 90% SoH delivers a full-shift range for most delivery routes. A pack at 72% SoH may be fine for a short-radius corridor but will leave a rider short on a longer route. Rather than rotating packs randomly through a station's inventory, we can route higher-SoH packs to high-demand stations and use lower-SoH packs in shorter-range applications or prioritize them for retirement before they become a reliability issue.
Second, rider experience. A rider who swaps a high-SoH pack knows what to expect: a full shift of range, predictable performance. A fleet operation that is not tracking SoH will occasionally issue a degraded pack without knowing it, and the rider will notice mid-shift when they run low earlier than expected. From a fleet management standpoint, this is an invisible reliability problem that shows up as rider complaints or missed deliveries rather than as a line item in a maintenance log.
Third, retirement timing. The decision about when to retire a pack from swap circulation is currently a judgment call at many operations: take it out when riders complain, or when it fails a charge test, or when it reaches some nominal age threshold. SoH tracking lets you set a threshold, say 70% or 75%, and retire packs systematically before they become a field problem. In our pilot, packs approaching the retirement threshold are flagged and routed to shorter-range corridors or held back from circulation, not discarded immediately but managed to lower-risk applications until they reach the retirement threshold.
The Fleet Operator View
For fleet operators using BatteryPool, SoH data flows back via the fleet analytics dashboard. Monthly SoH reports on the Fleet Starter tier, weekly on Fleet Pro with degradation alerts. The degradation alert is the operationally useful piece: if a pack's SoH drops more than 3-4 percentage points in a short period, that is a signal of abnormal use or a cell problem that warrants pulling the pack from inventory sooner than its normal retirement curve would indicate.
We do not share raw cell-level electrochemical data with fleet operators because it is not operationally interpretable without additional context. What we share is the practical signal: usable range estimate, trend over time, alert if the trend is anomalous. That is what a fleet manager can act on.
Where SoH Tracking Has Limits
Voltage curve-based SoH estimation is a practical method, not a laboratory-grade measurement. In our experience, it is accurate to within roughly 3-5 percentage points at mid-life SoH values. At very high SoH (above 95%), the curve differences are subtle enough that estimation accuracy decreases. At very low SoH (below 60%), the cells may be exhibiting behavior that the nominal curve model does not fully capture.
For a swap network where the operational decision is "route this pack to high-demand station or lower-demand station, or retire it," this level of accuracy is sufficient. We are not making cell-by-cell manufacturing quality decisions. We are making inventory routing decisions, and for that purpose the SoH estimate gives us meaningful signal to improve over random rotation or cycle-count-only management.
As our pack fleet grows and we accumulate more per-pack charge event data, the accuracy of the per-pack SoH model improves. A pack with 200 charge events logged in our system will have a more refined SoH estimate than one that joined the network last week. This is one of the structural advantages of operating a pooled network rather than a fleet owning its own packs in isolation: the collective data improves the accuracy of every individual pack's health model.