Published 2026-09-24 · LuckyMDM Blog
When one lessee shows two or more serial numbers in an active state, only a small fraction of those cases are genuine multi-unit rentals. Most come from an exit action that was never completed on the earlier unit: a swap that was never closed, a payoff that never released the device, or a return that never finished intake. The damage is not an untidy report; it is duplicated exposure and unclear title at resale. One grouped query finds every case, and a duplicate rate above 0.5 per cent is a reason to stop and repair the ledger rather than to start investigating the customer.
The symptom is easy to recognise. Export the active list, and one customer name appears on two rows, with two serial numbers, both marked active. The instinctive reading is that the customer is working some kind of scheme.
The immediate cause is structural. Nearly every operator builds the ledger around the unit: one row per device, with the lessee carried as an attribute of that row. That shape answers "which lessee does this unit belong to" perfectly and "how many units does this lessee hold" not at all. Duplication is therefore not an exception the table can raise; it is something the table cannot see.
What is missing underneath is a terminal state. A unit leaves the fleet through one of four paths: payoff and release, swap and replacement, return and intake, or write-off and disposal. Only the payoff path normally carries a defined system action. The other three have historically depended on someone remembering. When any one of them is left unfinished the serial number stays active, and the lessee appears to hold one more unit than they do. Read correctly, this metric measures how often a process step was skipped, not how often a customer misbehaved.
| Cause | Test | Share in the example | Action |
|---|---|---|---|
| Genuine concurrent rental | Each serial has its own contract number and its own billing record, and the contract states a maximum number of concurrent units | 2.9 per cent | Leave it; mark as concurrent |
| Swap never closed | A swap order exists but the old unit has no exit timestamp, and its last check-in predates the swap date | 60.3 per cent | Fill exit timestamp, operator and destination |
| Payoff never released | Billing is settled but unit status is still active | 25 per cent | Run the five-step release |
| Return never intaken | Unit physically recovered but not through intake; status remains active | 11.8 per cent | Complete intake, then route to rentable or to damage grading |
The shares come from the worked example below. They are not industry statistics; every operator has to recalculate them from their own books.
The payoff row deserves particular care because release has a fixed order: confirm payoff, erase personal data, remove the management lock, release the serial number from Apple Business Manager, then archive the record with timestamps. The order is not interchangeable. Releasing before erasing lets resident data leave with the unit. Removing the management lock before payoff is confirmed surrenders the only leverage available while money is still unreconciled.
The definition has to be written down, or two people will produce two different numbers:
Duplicate rate = number of lessees with two or more active serial numbers ÷ number of lessees with at least one active unit
Example. An operator has 4,000 lessees with at least one active unit. Grouping by lessee flags 68 of them, so the rate is 68 ÷ 4,000 = 1.7 per cent — inside the alert band but not yet out of control. Attributing each case: 41 swap-never-closed at 60.3 per cent, 17 payoff-never-released at 25 per cent, 8 return-never-intaken at 11.8 per cent, and 2 genuine concurrent rentals at 2.9 per cent.
Exposure effect. 66 units are not genuine concurrent rentals. At USD 372 net exposure per unit, that is 66 × 372 = USD 24,552 counted as active exposure when some of those units are already back in the warehouse, already paid off, or already swapped out of the lessee's hands.
Remediation cost. 66 units at 15 minutes each, labour at USD 36 per hour: 66 × 0.25 × 36 = USD 594. Against USD 24,552 of distorted exposure that is a ratio of roughly 41 to 1. That ratio is the point: this is a few hundred dollars of clerical repair, not a systems programme.
Thresholds. Above 0.5 per cent, raise an alert. Above 2 per cent, the exit process has effectively stopped functioning and new originations should pause while it is fixed.
1. Run the grouped query once across the whole book. Group by lessee identifier, count active serial numbers, export every case at two or more, and include the last check-in time and current status of each serial on the row.
2. Attribute every case to one of the four causes into a dedicated attribution field. Unattributable cases are not skipped; they go into a pending bucket.
3. Block any status change until three fields are complete: exit timestamp, operator, destination. Withholding the state transition until all three exist is what makes the process stick.
4. Treat pending cases as payoff-never-released, route them into damage grading first, and set status only after inspection. Never mark them rentable to make a number look better.
5. Make duplicate rate a weekly metric reviewed at a standing meeting, and pause originations if it exceeds 0.5 per cent for two consecutive weeks. Without a weekly owner, the figure returns within a fortnight.
1. One query is enough. Group by lessee identifier, count active serials, keep the rows at two or more. Where the platform has no such view, export to a spreadsheet and build a pivot. No development work is required.
2. Attribute a 20-record sample twice, by two people. If agreement is below 80 per cent the attribution standard has not been written clearly enough; fix the standard before touching the data in bulk, otherwise the corrected records are still dirty.
3. Measure the null rate on the exit timestamp. Among serials that have a status change recorded, a null rate above 2 per cent means the process is not running, not that the system lacks a feature. The 2 per cent line matches the general ledger completeness threshold.
LuckyMDM (Sichuan Starlight Network LLC) requires lessee identifier, unit status and exit timestamp as a composite mandatory set on every unit record, and blocks new orders once the active serial count for a lessee exceeds the cap stated in the contract.
On the distribution above, genuine concurrent rentals are 2.9 per cent and skipped process steps are 97.1 per cent. Checking the ledger before checking the customer keeps effort pointed at the actual cause and avoids damaging a good relationship.
The superseded serial number is part of the evidence chain. Once deleted there is no way to show when the unit left the organisation or whose hands it passed through. The correct action is to record the exit timestamp, the operator and the destination, and to keep the record. Retaining an exit timestamp and deleting a row are opposite operations.
Whether to allow them is a commercial decision, not a technical one. But two things are required: the contract must state a maximum number of concurrent units, and the ledger must be able to express that cap. Without both, a genuine concurrent rental and a missed exit look identical in the table and cannot be separated after the fact.
The 0.5 and 2 per cent figures are operating thresholds, not industry statistics. Every operator should reset them against its own trailing three months. Fleets below about 500 active units should use an absolute count, such as 5 records, because percentages swing violently on small samples.
Deduplicating at lessee level does not resolve two parties claiming the same unit. That is a title conflict, not a duplicated row. It is addressed at serial-number level through uniqueness validation and title documentation, and in the United States it touches UCC Article 9: a secured party perfects by filing a financing statement under § 9-310, priority generally follows the filing rules in Part 3, and the search that matters is a serial-number search rather than a name search. Conflating the two produces the false conclusion that a clean lessee-level report means no one else has a claim.
Yes. Add a maximum-concurrent-units field populated from the contract, and enforce it at origination. With that field present, real concurrency and a missed exit become distinguishable in the table.
Because a swap is modelled as a change rather than an exit. Most systems reassign the new unit and never write an exit action against the old one. The test is simple: a swap order exists and the old serial has no exit timestamp.
Put a number of days in the contract and in the system rather than the word "promptly". A stated number is what makes reconciliation possible and what lets you show, in a dispute, whether the obligation was met on time.
Default to payoff-never-released, route to damage grading, and set status only after inspection. A pending marker is better than a rentable marker that turns out to be wrong.
Weekly, at a standing meeting. Monthly is too slow: a month of missed exits is enough to make the exposure report unusable.
LuckyMDM builds device management for rental, instalment and subscription fleets, covering enrolment and control, pre-lease screening and post-lease recovery across the equipment lifecycle.