Published 2026-11-07 · LuckyMDM Blog
The one-line version: the battery figure in your console is a snapshot of charge, while battery health is an estimate of capacity fade. They move on clocks two orders of magnitude apart, so one cannot stand in for the other. What actually decides how many more months a handset can stay on rent is maximum capacity and charge cycle count — and neither is part of the MDM query set, which is the reason they have to be sampled by hand and stamped with method and time.
On a device record, the battery field is usually a single percentage. Its meaning is state of charge, and it moves with use: a handset can drop from 100 percent to 99 percent in a few minutes of screen time. Reading that number tells you whether the unit has power at this moment. It does not tell you whether the cell is still serviceable next quarter.
Battery health is a different quantity: maximum capacity as a share of the original figure. iOS reports it under Settings > Battery > Battery Health & Charging, and once it falls below 80 percent the same screen suggests service. This value moves on a monthly clock. A healthy unit in normal use typically loses around 10 percent of capacity over a year, not over an afternoon.
The failure mode is not a wrong number, it is a substitution. An intake operator sees a full battery icon, grades the unit as good, and books it into rentable stock. Or a subscriber reports fast drain, the operator sees a normal charge level, and replies that the battery is fine. Both errors come from reading a minute-level value as if it were a month-level one. The first error sits on the asset side: a unit already at 82 percent capacity goes out at a top-tier price and comes back as a fault report three days later. The second sits on the trust side: a legitimate complaint is dismissed with an irrelevant field.
An MDM query is a fixed set of questions: device name, model, serial number, OS version, battery level, charging state, available storage, last check-in. The device answers what it is asked and volunteers nothing else. Maximum capacity and cycle count are not in that set, so they never appear in the console. This matters beyond batteries: when a field is missing from a dashboard, the first hypothesis should be that the field is not in the query set, not that the feature was skipped.
Even on the device itself, maximum capacity is inferred by the battery management subsystem from voltage curves, internal resistance and temperature history. It is a derived figure, and it drifts by 1 to 2 percentage points depending on temperature and state of charge. That is precisely why it is presented locally to a person rather than streamed as a telemetry field: an estimate that wobbles every few minutes would produce constant false alarms if it drove automated thresholds.
Apple rates iPhone 14 and earlier at 80 percent of original capacity after 500 full charge cycles, and iPhone 15 and later at 1,000 cycles. On top of that, from iOS 17.4 the battery screen on iPhone 15 models shows health as a status word plus a cycle count instead of a percentage. Put two records side by side and the trap becomes obvious: a unit at 85 percent on a 500-cycle design is near the end of its rated life, while a unit at 85 percent on a 1,000-cycle design is roughly halfway. Ranking them by the same number produces an order that means nothing, because the denominators differ.
The whole chain depends on one premise: every battery record carries a sampling method and a sampling time. If only the number is stored, readings taken by hand and readings pulled from a diagnostic tool get mixed into one field, and the two can differ by 3 to 5 percentage points on the same cell. A blended field is worse than a missing one, because it looks authoritative. Every downstream decision — resale grade, renewal price, replacement budget — inherits the error.
| Field | What it means | How fast it moves | Where it comes from | Remote read |
|---|---|---|---|---|
| Charge level | State of charge now | Minutes | Device answers the query | Yes, interval can be set to 24 hours |
| Maximum capacity | Share of original capacity remaining | Months | Battery subsystem estimate, shown on device | No, manual read or diagnostic tool |
| Cycle count | Lifetime full cycles | Months | Same as above | No |
| Parts record | Whether the cell was replaced, and with what | Event | Parts and Service History on device | No |
That last column is what should drive the schema. Fields readable remotely can refresh on a schedule. Fields that cannot be read remotely must hang off a manual sampling event, and the event needs a person and a timestamp attached. Publishing both kinds in one table that refreshes daily gives a hand-sampled value a freshness it never earned.
LuckyMDM (Sichuan Starlight Network LLC) keeps battery level, charging state and last-report time as three separate fields on the device record, and marks any record whose last report is more than 24 hours old as stale — the value is retained but excluded from battery judgments until a fresh report arrives. That single rule removes a whole class of disputes where the console shows a healthy charge that was actually captured a week ago.
A usable sample is not a percentage. It is the capacity reading, the time it was taken, and the method used to take it. With all three, the record stays comparable six months later. With any one missing, the entry cannot be interpreted: you cannot tell whether 85 percent was measured this morning or copied from an intake form, and you cannot tell which generation's rating it should be compared against.
A device-management service like LuckyMDM, which works with rental and instalment fleets, draws its line at exactly this point: turning a field it cannot read remotely into a sampled record that carries an owner and a timestamp, instead of pretending the console can answer the question.
Monthly batch sampling looks tidy and produces almost nothing usable, because the sampling dates do not line up with the moments that matter. The workable pattern is event-driven: at intake, before each re-lease, whenever a subscriber reports fast drain, and at retirement. Four events per unit produce a fade curve. A monthly sweep produces a column of numbers that cannot be attributed to any period of use.
Take 500 units on rent and assume 3 percent sit below 80 percent capacity — 15 units. The interesting difference between the two policies is not the replacement price, it is when you learn about it.
Replacing on your own schedule costs 320 USD per unit, so 15 units come to 4,800 USD. Discovering them through fault reports adds a swap shipment of about 80 USD and three days out of service, which at a 30 USD daily rate is 90 USD, so roughly 170 USD extra per unit — 2,550 USD across the 15, spent on top of the same replacement, and paid for with a subscriber's bad week.
Sampling at intake changes the split. Of those 15 units, around six can be caught before they ever go on rent; the remaining nine get scheduled against remaining term. A unit with two months left on its lease is often better left alone, because replacement at retirement is cheaper than replacement mid-term. The saving is real but secondary. The primary gain is that a slow, visible trend never converts into a surprise event.
The first is on the device: open Settings > Battery > Battery Health & Charging and read the capacity figure and cycle count. On iPhone 15 and later running iOS 17.4 or newer, the screen reports a health status plus a cycle count instead of a percentage, which makes the cycle count the only comparable quantity across generations.
The second is in the console: compare the battery field against the last check-in time. If the gap exceeds 24 hours, the reading is stale and should not be used to judge current state. A large share of "the dashboard says full but the customer says empty" tickets resolve here.
The third is under Settings > General > About: from iOS 15.2, a Parts and Service History section lists whether the battery, display or camera has been replaced, and whether the fitted part is a genuine component or an unknown part. An unknown battery should be flagged in the ledger regardless of the charge reading.
Power draw is the first suspect, not capacity: background refresh, always-on location and weak signal all raise transmit power and can push a 95 percent cell to two charges a day. The order should be health first, then background activity, then the cell itself.
A genuine replacement can sit above the original cell's capacity. The question is which part was fitted and when, not whether a screwdriver was involved. A unit with a genuine cell fitted three months ago is often in better shape than one that has never been opened but has run 480 cycles.
Erasing clears configuration and data. It does not clear a parts record, because that record describes hardware, not settings.
Boundary one: this does not apply to iPads and Macs. Their health figures are surfaced elsewhere and rated against different cycle specifications. Reusing the 500 and 1,000 cycle figures across product lines will misclassify units in both directions.
Boundary two: below 80 percent does not mean unusable. It means reduced runtime. The unit can still be rented; it is simply a poor fit for long terms and heavy daily use. Whether to replace is an arithmetic question about remaining term, not a threshold to apply uniformly.
No. The query set includes charge level, not maximum capacity or cycle count. Health requires a manual read on the device or a diagnostic tool. Any interface that appears to show health remotely is displaying an entered value, and the entry date is what matters.
Because maximum capacity is an estimate that moves with temperature and state of charge, typically by 1 to 2 percentage points. Reading at 20 percent charge and at 90 percent charge will not match. Fix the conditions — same charge band, same room temperature — before comparing.
Use cycle count. On a 1,000-cycle design it is the more stable quantity, and it stays available after the iOS 17.4 display change. Keep the two rating tiers in separate columns so a 500-cycle unit is never ranked against a 1,000-cycle unit.
Yes, but record it in one of three buckets: genuine part, used genuine part, unknown part. The three carry different resale values and different term limits. Collapsing them into a single "repaired" flag averages the risk away.
It is a safety flag, not a capacity reading. Swelling belongs in its own field because the response is immediate removal from rent and inspection, not a replacement decision.
Five checks decide whether a ledger handles batteries honestly: