A Condition Grade Is a Reading, Not a Property: Why Last Cycle's Grade Cannot Be Carried Forward

Published 2026-10-29 · LuckyMDM Blog

Bottom line: a condition grade is a reading taken at one moment, not a property bolted to the unit. When a unit comes back from one subscription cycle and goes back on the shelf, only part of the previous grade still holds. The deciding question is not how much that field matters, but whether the underlying variable moved during the cycle and whether it can move back. Anything that degrades monotonically has to be read again, and the new reading has to sit next to the old one rather than replace it.

The same unit, two grades, six months apart

A handset finishes a 12-month subscription and comes back. The person doing intake opens the record and sees what was written when it went out six months earlier: grade A, maximum battery capacity 96 percent. They turn the unit over, see no new marks on the case, and copy the whole row onto the new listing sheet.

Six months later the unit comes back again. The buy-back quote comes in at the grade below. Actual maximum battery capacity by then is 84 percent, and nobody watched it cross the step.

What actually happened

The record stored condition as a single column, and that column held three different kinds of value at once. One of those kinds describes what the unit *is*. The other two describe how hard it has been used. In one column, whichever value gets written first ends up standing for the whole row.

Why the reading moved

The reason the three have to be separated is that each depends on a different driver. Factory attributes are set at manufacture and do not respond to use. Physical wear accumulates in one direction only. Appearance picks up marks during a cycle and can be partly restored afterwards. At root, the test for whether an old reading still holds is reversibility, not importance.

When carrying the old value forward is safe

Carrying a previous reading forward only holds when nothing happened between the two cycles. Once the unit has been shipped out, sent to a third-party repairer, or moved between sites, the previous reading stops describing the unit as it is now. That line is drawn by events, not by the calendar.

Three kinds of variable behind one grade

KindExamplesBehaviour over timeCan the old reading be reused
Factory attributeModel, storage size, colour, part numberFixed at manufacture, unchanged across cyclesYes, unless the board has been replaced
Monotonic readingMaximum battery capacity, charge cycle count, non-genuine part flagsMoves one way only; a wipe does not reset itNo; read it again and keep both versions
Restorable appearanceCase marks, bezel dents, accessory completenessCan get worse in use and better after a repairNo; judge it in person at the moment of handover

Use the table one column at a time. Ask whether the variable moved during this cycle. If the answer is not certain, treat it as a field that has to be read again.

Factory attributes are the only ones you can reuse

Model, storage size, colour and part number are set before the unit reaches anyone, so they are stable across any number of cycles and can be quoted from the previous reading. The one exception is a board swap: the serial number and IMEI move with the board while the outside of the unit does not change, so this group needs a separate flag that marks it as unreadable after a repair.

Monotonic readings have to be read every cycle

Maximum battery capacity degrades with every charge cycle, and a non-genuine part flag, once set, is not cleared by a wipe, a restore, or re-enrolment. These fields share one property: not looking at the latest version means not knowing what the unit looks like now. Apple's own warranty threshold sits at 80 percent of maximum capacity and 500 charge cycles, but that boundary only decides whether a battery is replaced under warranty. In a buy-back pricing model every step down lands on a different price band.

Restorable appearance has to be judged in person

Case marks and bezel dents move in both directions. They get worse during a cycle and better after a replacement housing or a new accessory set. Because they are reversible, the previous reading carries no information at all: whether anything was repaired in between cannot be recovered from the number itself.

Six events that expire the previous reading

Whether an old grade can still be used depends on what happened in between. Any one of the six events below retires the previous reading, and each of them requires a fresh measurement rather than a reuse of the most recent one.

TriggerWhy this one retires the readingWho takes it
Returns inspection passedThe only moment the unit is back in your hands; it is the zero point of the new cycleReturns inspector
Shipped to a new subscriberResponsibility changes hands; without a reading taken at dispatch you cannot evidence the state it left inDispatch operator
Returned from third-party repairRepair means opening the unit, and whether parts were swapped can only be settled on the spotRepair intake clerk
Transfer between sitesThe disputed interval is bracketed by two signatures, so both ends need their own readingBoth sites, once each
Idle more than 90 days on the shelfLong idle stretches are where physical damage clustersWarehouse, during the monthly count
Change of responsible ownerThe new owner inherits no evidence value from the old readingIncoming owner

The absence of a time dimension in that table is deliberate. An untouched unit sitting on a shelf for 90 days differs little from one sitting there for one day, while a third-party repair that takes two days can move the grade a whole band. Events drive expiry; waiting does not.

The arithmetic at 1,000 returns a year

Every parameter is stated here so you can substitute your own. Take 1,000 units returned and re-listed in a year. Assume 30 percent of them reuse the previous grade instead of being re-read, so 300 units. Assume 12 percent of those turn out to be one grade band away from what a fresh reading would give, so 36 units.

A public buy-back price sheet published in October 2026 lists one previous-generation flagship in six bands: as-new 6,950 USD, excellent 6,650 USD, light wear 6,350 USD, heavy wear 4,950 USD, cracked outer 4,800 USD, cracked inner but testable 4,100 USD. The gaps are uneven. Light wear down to heavy wear costs 1,400 USD, a drop of 22.0 percent, while as-new down to excellent costs 300 USD.

Set that against the cost of doing it properly. A full re-read takes 4 minutes per unit, so 1,000 units is 66.7 hours. At 36 USD an hour for inspection labour, that is about 2,400 USD a year.

Missing one unit costs between 4.5 and 21 times what reading all of them costs. The ratio barely responds to fleet size and responds strongly to the width of the price gap, so the first step is not a new process. The first step is finding the most expensive gap in your own price sheet. For most models that gap sits between visible wear and visible damage.

How the record should be stored: one reading, one row

Once the reading rules are settled, what is left is structure. Platforms in this category, including LuckyMDM (Sichuan Starlight Network LLC), a device asset management service built for rental and instalment fleets, tend to split the grade into two groups: factory attributes, whose values survive every cycle and can be quoted, and time-varying readings, which get retaken every cycle because their movement is not reversible.

In structural terms, one condition reading needs at least four columns: the capture timestamp, the capture method, the field name, and the value. Two readings of the same field should produce two rows rather than an update to one row, because that is what makes any two points in time comparable without maintaining a separate history table.

Three checks you can run yourself

The third check is only possible because of how the record is stored. LuckyMDM writes maximum battery capacity, charge cycle count and non-genuine part flags as a new row on every reading, seals the previous row instead of overwriting it, and gives each row its own capture timestamp and capture method.

Where this article divides labour from what is already here: the piece on parts and service history covers how a swapped display is detected at inspection; the piece on battery thresholds covers how one field maps onto price bands; the piece on acceptance sampling covers what share of returns to open. This one is about whether a reading taken in one cycle still describes the unit in the next.

Misconceptions

Misconception one: the model and the storage size are unchanged, so the grade is unchanged.

Model, capacity and colour really are stable, but maximum battery capacity, non-genuine part flags and case marks all move. Factory attributes explain only the small part of the price difference that comes from one configuration of the same model; the large part comes from the readings.

Misconception two: the newest version is enough, so the old row can be deleted.

Two readings exist so that the interval between them can be examined. Without the earlier one you cannot answer whether the damage in this cycle happened in this cycle. When a subscriber says it arrived already damaged, you have nothing to put next to that claim. Keeping both rows is not clutter; it is what makes evidence possible.

Misconception three: the error only runs one way, toward being too generous.

A generous reading undercharges. A harsh one triggers disputes and complaints at the next end of term, which costs more because it also loses the customer and the orders behind them. Grade drift on units that have already run several cycles has to be counted in both directions.

Boundaries

Boundary one. Where a unit never leaves the building, is never unsealed, and moves straight from one agreement to the next with no dispatch step in between, the reuse window can stay open until the next dispatch. Once any one of dispatch, third-party repair, or transfer between sites happens, the window closes with no grace period.

Boundary two. Android has no single system-level reporting item for non-genuine parts. The irreversible readings described here come mainly from the iOS parts and service history screen. On Android these values depend on manual comparison and third-party tooling, so transferring the same thresholds across will not hold; each brand needs its own criteria.

Questions teams ask

Does carrying forward ever save enough time to be worth it?

Only on the factory attribute group, and only where the board has not been swapped. That group is small, quick to verify, and carries none of the price movement.

Which timestamp should the record carry? The moment the reading was produced, which is when the action actually happened, not the moment the paperwork was keyed in. Batch entry across an overnight shift can separate those by a full day, and that distance is where timelines start to disagree.

Are photographs worth storing? Yes, and they should travel with the readings. Two versions of one field paired with two photographs settle disputes faster than numbers alone, provided that shooting angle and lighting are written into the capture standard; otherwise the same unit photographs into two different grades at two different sites.

The checklist

All Articles