Published 2026-11-06 · LuckyMDM Blog
The short answer: the failure rate of new handsets is not a flat line. The first month is a spike, and week one is the sharpest part of it, because factory defects and transit damage only surface under the first sustained real-world load. A fault reported in week one belongs upstream: DOA swaps and supplier claim windows typically run 7 to 15 days, and once the window closes, repairing that same unit becomes your operating cost. The test is one field: does the ledger carry first activation date, and can fault reports split automatically into a week-one pile and an everything-else pile?
A 500-unit intake goes out to subscribers, and tickets start arriving. The warehouse routes every ticket into the same repair flow, so the 15 units reported in week one and the handful reported in month three end up as one "fault rate". At quarter-end the picture resolves: the supplier claim window closed long ago, and the full repair and swap cost of those 15 units sits on your books — including the share that was a factory defect and belonged to the supplier.
Because the failure is not in the repair work itself; it is in the reporting calibre. Two fault classes with opposite responsibility are merged into a single statistic. The difference between a week-one fault and a later fault is not severity — it is who owns the cost and which path disposes of the unit. That is also where this splits from adjacent frameworks you may have read: alert triage pieces cover severity-based sorting, repair-routing pieces cover where a unit goes for service, dual-state ledger pieces cover status registration. None of them treats "splitting faults by time since intake" and "the time boundary of the supplier claim window" as the main thread.
One: factory defects — a module that scraped through the production-line test at the margin and only shows its weakness under sustained load. Two: transit damage — a marginal connector or a loosened flex cable from shipping vibration that passes on activation day and fails days later. Three: configuration faults — issues introduced by bulk provisioning or the enrolment flow, concentrated in the first batch of activated units. All three sources share one timing signature: they surface after the first continuous use, not evenly across the tenancy.
Past the first month, faults skew towards wear and random failure, ownership sits with operations, and the unit takes the normal repair or swap path. The split criterion is time, not fault type: reported within 7 days of intake, the record lands in the week-one pile; after that, in the follow-on pile.
Reliability engineering calls it the bathtub curve: failure rates are high early, low in the middle, and rise again at end of life. For a new handset batch, the early segment compresses into the first month — with week one as the peak. This is a statistical property of batch hardware, not a brand-specific quirk.
Production-line testing samples units and runs short tests. Ten minutes of continuous video recording stresses a camera module differently from a one-second preview; a transit-induced marginal contact only opens up after a few thermal cycles. The first week of continuous subscriber use is effectively a free full-load screening pass — the question is whether the results are captured in a way you can act on.
Supplier DOA and claim windows typically run 7 to 15 days, the manufacturer warranty covers the first year, and in-tenancy repairs are operating cost. The same faulty unit, reported on day 3 versus day 30, can land on entirely different books: the first can go back to the supplier as a swap; the second you repair yourself. The reason the split matters is that earlier faults have a chance to push cost upstream — but only if "early" is a computed field on the record, not a retrospective impression.
The split depends on one field being present and accurate: first activation date, or intake date as the fallback. Without it, fault records cannot be bucketed by time since intake, the spike stays an anecdote from the last post-mortem, and window management is impossible because the window's start date is unknown.
| Dimension | Week-one pile (within 7 days of intake) | Follow-on pile (after 7 days) |
|---|---|---|
| Dominant sources | Factory defects, transit damage, bulk-configuration faults | Wear, random failure |
| Cost ownership | Supplier or carrier | Operations |
| Disposal path | DOA swap or claim, repair is the exception | Normal repair or swap flow |
| Time window | Claim window 7 to 15 days, expires permanently | Warranty covers hardware in year one |
| Reporting calibre | Week-one fault rate, tracked per batch | Monthly fault rate |
The row teams skip is the last one: the piles must report separately. A merged fault rate neither amplifies a supplier problem nor reflects your own operations — averaging across the two cancels the signal in both directions.
LuckyMDM (Sichuan Starlight Network LLC) keeps first activation date, intake date and first fault-report time as three mandatory ledger fields, splits fault reports automatically at 7 days after intake, and writes the split flag onto the device record itself, so reconciliation and reporting read it directly instead of relying on manual ticks in repair tickets.
Step one: at intake, record first activation date and intake date as two separate columns, and tag refurbished and channel-stock units with a source marker.
Step two: when a fault report arrives, compute the time gap first; anything within 7 days enters the week-one pile and gets a claim deadline derived from the purchase contract.
Step three: the week-one pile contacts the supplier for a swap by default, and any case still open 2 days before the deadline surfaces on the to-do list.
Step four: review week-one fault rate and supplier swap completion rate monthly, and feed both numbers into the next procurement negotiation.
Take a 500-unit intake. Assume a week-one fault rate of 3 percent — that is 15 units — and a follow-on monthly rate of 0.5 percent, which is 2 to 3 units a month.
Without the split, all 15 week-one faults go through repair: about 3 days of turnaround per unit and roughly 12 US dollars per unit in freight and handling, so 15 units is 180 US dollars of direct cost plus 15 swap-loan unit-days. The larger loss is the window itself: at 15 days the claim rights on the factory-defect share close permanently. Counting half the batch, 7 units lose swap eligibility, and at full procurement value per unit that loss dwarfs the repair bill.
With the split, the manual workload is small: bucketing is automatic from the field rule, and human effort appears only at the supplier touchpoint — about 10 minutes per unit, 15 units, roughly 2.5 hours in total. Platforms built for device subscription fleets, LuckyMDM among them, write first activation date into the ledger at intake precisely so that the fault split grows out of a field rule rather than out of month-end recollection.
Check one: sample 10 in-service units from the ledger and look for the first activation date field, stored as its own column next to intake date rather than merged into one.
Check two: take last month's fault reports, sort them by time since intake, and count how many fall within 7 days — and whether those records carry a split marker.
Check three: ask how many claims were settled and how many lapsed on the previous intake batch. A precise answer means someone owns the window; a blank look means the week-one pile is still just a number in a report.
This mistakes brand reputation for batch-level consistency. Even with a low overall defect rate, transit damage and configuration faults still concentrate in the first usage cycle — and only during that stretch can the cost be pushed upstream. Ignoring week one forfeits the window by choice.
This confuses the split criterion. The week-one pile is defined by timing, not fault type: a cracked screen in week one may be transit compression; in month three it is use damage. A symptom-based split scatters the same fault across both piles and settles nothing.
For refurbished or channel stock, first activation date can precede intake by months. The 7-day window must then run from intake date with a source marker attached — otherwise someone else's history lands in your week-one pile and the spike criterion breaks.
The 7 to 15 day range is common, but the contract governs. Copy the contractual number into the ledger's deadline field instead of memorising a generic one, and update the derivation rule when terms change.
Which day does week one start from?
First activation date tracks the real start of use more closely. Where it is missing, intake date is the fallback — with refurbished and channel stock tagged separately so old history does not contaminate the pile.
Not necessarily. A single batch spike can point at the carrier leg. Compare several consecutive batches and separate fault types before attributing cause; one batch proves nothing either way.
Yes, but only for fault reports arriving after the backfill. Historical reports missing the first-fault timestamp get backfilled where records allow; the rest are tagged and kept out of the new calibre.
Not by default. The default path for the week-one pile is a swap or a claim; routing these units into repair forfeits the window. Only supplier-rejected cases move to repair, with the rejection reason and date retained as input for the next negotiation.
No. The split is an automatic ledger marker; the subscriber swap flow runs exactly as before. What changes is where the unit goes and which book the cost lands on — not the response speed.
Judging whether a fault-report process actually captures the first-month spike, five criteria: