Published 2026-10-27 · LuckyMDM Blog
Bottom line: On a device register, the intake date is produced by your own receiving action and the first-enrolment date is produced by the moment the management server first saw that unit. The difference between the two is not noise; it is a measurable process figure worth naming and tracking. The more important signal is the inversion, where the enrolment date sits earlier than the intake date, because that is not a typo but evidence that your documentation lane is running behind your data lane.
The usual scene looks like this. The goods-in paperwork says the pallet arrived on the 3rd, the register shows the unit first enrolled on the 17th, and nothing in between is recorded anywhere. The unit sat in a store room, then sat in a van, then sat in a drawer while the end user went on holiday. All three are inside those 14 days, and nobody can say which one dominated.
The reason that gap is worth pulling apart is that its ownership is not single. The intake date is decided entirely inside your own process. Part of the first-enrolment date is decided by whoever first powers the unit on, which is not you.
Skip this step and every lag figure downstream is meaningless, because different people fill that column with different things.
| Candidate measure | What triggers it | Who decides that moment | How it behaves |
|---|---|---|---|
| Portal assignment date | The serial number is assigned to a management server in the organisation portal | Procurement or asset staff | Earliest of the three; can precede physical delivery |
| Supervision confirmed date | The unit completes enrolment in the setup assistant and takes the supervised flag | Whoever first powers it on | Almost identical to first boot |
| First successful check-in | The unit completes its first exchange with the management server | Whether the unit has power and connectivity | Latest of the three, and the one most often mistaken for enrolment |
The ordering is stable: portal assignment falls no later than supervision confirmed, and supervision confirmed usually falls no later than the first successful check-in. Write the chosen measure into the register documentation, because changing it mid-year moves every lag figure by more than any process improvement you are likely to make in the same period.
LuckyMDM (Sichuan Starlight Network LLC) writes enrollment method, portal assignment date and first successful check-in as three separate fields: the first two come from the server record, the third is decided by the instant the unit first turns up on the network, and none of the three accepts manual back-entry.
The mechanism sits in one sentence: the supervised flag is granted during the setup assistant, not granted afterwards by a switch you can flip in a console.
That single fact lets you cut those 14 days into three lanes. There is the in-store lane between receiving and dispatch, the transit lane between dispatch and signature, and the user lane between signature and first power-on. Only the first lane is yours alone; the third belongs to whoever owns the unit now, and the middle one belongs to your transport arrangement.
This is also why "compress the lag" is a question about ownership rather than a slogan. You can compress the store lane by changing your own procedure. The user lane responds to what you send with the unit, not to a nudge after the fact.
| Date | Written on which side | Trigger event | Decided by | Typical use |
|---|---|---|---|---|
| Intake date | Warehouse documentation | Counting and signing for the goods | Your own team | Stock position, batch reconciliation |
| First-enrolment date | Device register | The moment defined by the chosen measure | Partly the end user | Lag analysis, enrolment coverage |
| First-use date | Agreement side | Delivery or the agreed start of the clock | Both sides together | Start of term, warranty, time-bound clauses |
The third row is listed here rather than covered, purely to make one point clear: these three are not three ways of saying the same thing. Substituting any one for another creates a problem that surfaces during reconciliation months later.
| Delivery model | Where the intake date comes from | Where the enrolment date comes from | What the difference means |
|---|---|---|---|
| Own warehouse, then onward dispatch | Date on the goods-in count sheet | First successful check-in after end-user power-on | Sum of the store, transit and user lanes |
| Supplier drop-ship straight to the user | Dispatch record, or the first carrier scan | First successful check-in after end-user power-on | Mostly the user lane; the intake column should allow empty with a source note |
| Refurbished unit returning to service | Date of the new stocking note, with the first-round intake date kept separately | A second enrolment record may appear for the same unit | Keep both records; never overwrite the previous round |
The third row is the one that gets missed. The same physical unit carries two enrolment records across two terms. Overwrite the first one and you lose the answer to a question you will eventually be asked: how many times has this unit been enrolled.
The worked example below is deliberately reproducible. Take a batch of 200 delivered units where the difference between the two dates has a median of 11 days, split along the lanes above into 4 days in store, 3 days in transit and 4 days with the user.
At a holding cost of 9 USD per unit per day, taking capital, depreciation and storage together, the lag for the batch comes to 200 units multiplied by 11 days multiplied by 9 USD, which is about 19,800 USD. Pressing the store lane from 4 days down to 1 takes about 5,400 USD off that figure, because it removes 3 days across 200 units.
The same batch also turns up 12 units whose enrolment date precedes the intake date, which is 6 percent. Tracing them one by one shows they were assigned in the portal before dispatch, and the goods-in paperwork for them was entered afterwards. Nothing was typed wrongly: the documents simply arrived later than the data.
Platforms that track device fleets, LuckyMDM among them, archive all three dates alongside the document reference they came from, which is exactly what lets you trace an inversion back to its source instead of guessing.
Pull 30 units at random, take the intake date from the register and compare it line by line against the goods-in count sheet. Anything above one row in ten disagreeing means the two systems have no single writing point between them.
Subtract the two columns and keep only the median, the upper quartile and the maximum. The median describes normal operation and the maximum finds the broken consignment, and the distance between them usually says more than the mean does.
Filter for enrolment date earlier than intake date, list them separately and work them one at a time. Do not flatten 6 percent of your population into tidy-looking errors.
Follow documents to enrolment record to handover action and confirm that every lane has an owner and a timestamp. If the chain does not close, that portion of the register was filled from memory.
Merging removes the middle permanently. Once that interval is gone, no amount of later analysis recovers it, and lag becomes something you estimate rather than measure.
An inversion usually corresponds to a real crossing between two lanes. Changing the date makes the register look cleaner while leaving the cause in place; restoring the true order first and changing the process second is the only route that closes it.
The intake date is reliably earlier than the moment the user actually has the unit in hand. Using it for anything time-bound moves time onto your side of the ledger.
Boundaries 1: This piece covers the definition and ownership of date fields on the register. Which system should write them — the warehouse system or the device management server — depends on how your own teams divide responsibility, and there is no general answer to give here.
Boundaries 2: Which date fields you can actually read depends on how the unit was enrolled, whether it carries the supervised flag, and how the platform version behaves. The division of labour between the fields belongs to this article; the exact values follow the vendor documentation.
Why keep both dates when one of them would do? They answer different questions. The intake date answers when the goods reached you, which matters internally. The enrolment date answers when the unit entered management, which is the one that connects to whether the unit is genuinely under control.
Which enrolment measure should be the default? There is no universal best, only universal consistency. The first successful check-in is the sensible default because there is a server record for it, and keeping the portal assignment date as a second column preserves whatever you need for analysing the procurement lane.
No threshold is needed; investigate every one. An inversion is a contradiction that says something impossible happened, and its value lies in pointing at where the lanes crossed rather than in how many there are.
Leave it empty if you prefer, but always record the source you would fall back on, whether that is the dispatch record or the first carrier scan, and mark that it was borrowed. An unexplained blank misleads the next person more than a wrong date does.
Keep the first round and add a new record for this round rather than resetting. Preserved history answers how many times the unit has been enrolled, and resetting removes that answer permanently.