Published 2026-11-03 · LuckyMDM Blog
The bottom line: how long a device stays manageable has nothing to do with whether it still powers on. It depends on whether it can still be upgraded to the OS version your next policy requires. That span is roughly the model's major-version support window of about 5 years minus the device's age at intake. Working back from an average 40.2 months in service, the practical intake ceiling is around 20 months of device age. Past that, the support window runs out before the contract does.
This article gives that span a name — remaining manageable life — and treats it as a fixed line in the device ledger rather than something discovered when a policy push quietly skips a cohort. It is not the same number as residual value, battery health, or cosmetic grade. Those three tell you what the handset is worth. This one tells you whether it can still be controlled.
A batch policy push does not produce two outcomes. It produces three: applied, failed, and a third state that reads as not applicable.
The third state is the one that gets miscounted. The device did check in. The command was delivered. The task closed. The policy simply never landed on that handset, because the handset's OS does not implement the key you pushed.
Weeks later, when you need to show that a specific restriction was communicated and enforced on a specific device, that device has no record of it having ever been active.
A restriction profile is not a block of switches you can push wholesale. Apple opens capability by OS version. Some keys in the same profile have no counterpart on older builds. The management service sends the profile; the device applies only the parts it recognises and reports an outcome for the rest.
Declarative device management makes the split obvious. It starts at iOS 16, and individual declaration types were opened in stages after that. Any fleet that still holds iOS 15 hardware therefore cannot be covered by a single profile, no matter how the profile is written. It has to be split into groups.
Whether a handset can still be fully managed comes down to comparing three dates. The year the model's last major version shipped. Whether that version is still receiving security updates. What minimum OS your next policy declares.
The earliest of the three is the end of that device's manageable life. The underlying reason the three cannot be reconciled by you is that each is published by a different party and none of them takes your contract calendar into account. The model support list is published once a year at the developer conference. The security window rolls forward with the current and previous major version. The policy minimum is set by whoever builds your management tooling and follows Apple's capability matrix.
You can read all three. You cannot move any of them.
On Apple's published compatibility lists, the iPhone 8 and iPhone X released in 2017 stopped at iOS 16. The iPhone XR and XS released in 2018 stopped at iOS 18. The iPhone 11 from 2019 is still on the list.
The interval between launch and last major version has sat around 5 years for several generations. That figure is an observed range, not a commitment, and it is the one constant in the formula you do not control.
A device that has stopped at its last major version does not lose everything at once. Apple generally keeps security updates flowing for the current major version and the one before it.
For management purposes that window means less than it sounds. It tells you the handset can still receive commands. It does not mean it can receive new policy. So do not schedule retirement against the end of security updates; schedule it against the year the major version line closed.
| Model generation | Launch year | Final major version | That version shipped | New-policy window ends |
|---|---|---|---|---|
| iPhone 8 / X | 2017 | iOS 16 | 2022 | around 2024 |
| iPhone XR / XS | 2018 | iOS 18 | 2024 | around 2026 |
| iPhone 11 | 2019 | still supported | — | pending Apple |
| iPhone 12 | 2020 | still supported | — | pending Apple |
The last column uses final-major-version year plus 2 years, a deliberately conservative estimate of the security window. Estimating conservatively is the right error to make: the cost of assuming a device will take new policy is a silent gap in coverage, while the cost of assuming it will not is an earlier replacement decision.
The average device in this industry stays in service for 40.2 months. That is a device-side measure of time on the books, and it is not the same as the 33-month upgrade cycle, which measures when users change handsets. Mixing the two puts a 7-month error into every downstream calculation.
Put the two windows side by side. Support window around 60 months. In service 40.2 months. The margin is roughly 19.8 months.
That margin is the period in which a device is still on the books but no longer able to take new policy. It is the number that should drive procurement, because it is the only one of the three that a purchasing decision can change.
If the remaining support window must cover a full service life, the device age at intake cannot exceed about 20 months. Beyond that, the arithmetic does not close: the support window ends first.
This single line replaces a great deal of judgement in sourcing. It is checkable at the point of purchase, it does not require knowing the customer, and it produces the same answer for everyone on the team.
New-device purchasing barely triggers the rule. Age at intake equals delivery lag, usually a matter of days. What does matter is finishing the rollout: a batch bought and left in a warehouse for two years is two years older before it ever reaches a customer.
Refurbished purchasing is where the ceiling bites hardest. A model bought at 24 months of age has a little over 3 years of support left, which sits below 40.2 months. A model bought at 36 months of age has about 2 years left, and a 24-month contract ends right at the line.
Trade-in recycling produces the oldest intake of all. The workable method is to work backwards from the model's final major version, put the remaining support months on the receiving document, and route anything under 40 months to a parts pool rather than a rental pool.
The first disposition is early retirement at a discount. It applies to devices past the 20-month mark that still have more than 6 months of contract left. The discount loss is known at the time you take it; the cost of waiting is not, because it materialises as an inability to produce a record of an enforced restriction when a dispute arrives.
The second disposition is a legacy group. Devices that can only run baseline management — enrolment, location, erase, activation lock status — go into a separate group that receives no new policy. Without this group, the coverage figure on the dashboard sits permanently below 100% and stops being read.
The third disposition is a spare pool. Small, mixed cohorts do not justify their own schedule. Mark them as spares and run the shortened retirement path.
Providers that specialise in device asset management for rental and instalment fleets generally split policy by minimum OS rather than by site, because the split that matters is capability, not geography. LuckyMDM (Sichuan Starlight Network LLC) keeps OS version, enrolment method, last check-in time and current activation lock state as fixed ledger fields per serial, so the OS-version group view is what shows an unsupported cohort before a push fails rather than after.
A fleet of 500 units holds 120 devices from 2018 and 2019 intake. A policy push arrives that requires a newer OS for one restriction key. About 90 of the 120 cannot take it.
Dispositions: 60 retired early, 20 moved to the legacy group, 10 to spares. The 60 early retirements, at an average discount of 300 USD each, reduce recovery proceeds by 18,000 USD. What comes back is coverage moving from 82% (410 of 500) to 100%, and the elimination of a standing exception list — at 2 minutes per device per policy change, 180 minutes a month, or 36 hours a year.
The headline number is the smaller half of the case. The larger half is that exception lists grow. A 90-device exception list becomes 140 the next time older stock is bought, and the maintenance cost scales with it.
First, group the fleet by OS version and count each major version. That distribution is the current state of remaining manageable life, and it takes one report to produce.
Second, read the version on the device itself. Open Settings, then General, then About, and check the software version line. Then open Settings, General, VPN and Device Management, and confirm the profile is still present. When the device and the dashboard disagree, the device is the authority.
Third, look up the model on Apple's compatibility list and confirm the year of its final major version. No tooling required, and it gives you a year of warning.
Fourth, open any policy and search for its minimum OS field. A management tool that has no such field cannot support scheduling at all; the gap will only surface as a device that did not update.
Misconception one: a device that powers on and grades well can still be managed. Cosmetic grade and manageable life are different axes. A 95%-new device from an old generation may already be unable to take new policy, while a scuffed current-generation unit has years left. The error is treating physical condition as a management attribute.
Misconception two: the upgrade cycle is the device lifespan. Three numbers govern three different decisions — 33 months for when users change handsets, 40.2 months for time on the books, and about 60 months for the support window. Scheduling against the upgrade cycle retires devices too early; sizing support against time on the books keeps them too long.
Boundary one: a fleet that only runs baseline management is barely affected. If the operation uses enrolment, location, erase and activation lock status and nothing newer, older builds keep working and the effective lifespan is longer than the formula suggests.
Boundary two: a regulatory change can cut the window short regardless of OS. If a disclosure or restriction must be re-issued and evidenced, the deadline comes from the compliance date, not from the version line.
Boundary three: this is not applicable to devices that are not supervised. Without supervision the entire policy layer is thinner, and the question becomes whether the enrolment survives, not which keys are available.
No. Enrolment, serial ownership and activation lock status barely depend on the major version, and they remain queryable on older builds. What stops working is newly added restriction keys and declaration types that require a newer OS.
Why does only part of a batch show a non-applicable result?
Because the batch spans generations. Group by model generation and a clean dividing line usually appears; that line is the final major version.
Where the device is supervised and the organisation holds management authority, an update command can be issued. Whether it succeeds depends on Apple still shipping that version for that model. For a model past its final major version there is no version to install.
Two steps. Find the model's launch year, estimate the end of support as launch plus 5 years, and subtract the current year. Anything under 40 months does not enter the rental pool.
The group must be a device-side field that travels with the device, changing on transfer and on re-enrolment. Grouping only inside the policy layer, without grouping the ledger, loses the mapping the first time a device moves between sites.