Published 2026-10-07 · LuckyMDM Blog
The one-line version: a reported device position is a circle with an accuracy radius attached, not a point. The radius comes from whichever source the device could reach at that moment — satellite, Wi-Fi or cell tower — and the difference between the best case and the worst case is not a small correction, it is a factor of 10,000 in the area you would have to search.
Fleet operators in device subscription and rental read the map pin and then decide whether to send someone. That decision is where accuracy turns into money. This piece sets out the three sources, the radius-to-area arithmetic, and the point at which a location stops being actionable and becomes a region to hand over instead.
A device does not report a position with a fixed precision. It reports the best result available from the sources it can reach, and the console usually receives that result together with an accuracy value.
| Source | Typical error | When it is available | What the report means |
|---|---|---|---|
| Satellite (GNSS) | A few metres | Outdoors with a clear sky view | Usable for doorstep verification |
| Wi-Fi positioning | Roughly tens of metres | Indoor or urban, with known networks nearby | Usable for street or building-level work |
| Cell tower | Several hundred metres or more | Anywhere with a mobile signal, no data needed | Usable for region-level triage only |
The underlying reason the number moves so much is that the three sources measure different things. Satellite positioning measures ranges to orbiting transmitters and needs line of sight. Wi-Fi positioning matches nearby network identifiers against a database and depends on how densely that area has been surveyed. Cell positioning uses the serving tower and its timing or signal strength, which is why the error grows with tower spacing.
What makes this awkward in practice: the console often shows a single pin regardless of which source produced it. Two pins that look identical on screen can differ by two orders of magnitude in what they actually tell you.
Search area grows with the square of the radius. That single fact drives the whole cost curve.
Take three radii that arise in normal operation:
A team that treats all three as meaning the same thing will send someone out three times and succeed once. The arithmetic says the decision rule should be written the other way round: set a maximum radius for physical verification, and route everything above it to a different process.
A workable threshold for most fleets is 50 metres. Below it, one field visit can cover the area in about 15 minutes on foot. Above it, the visit becomes a search, and a 0.79 square kilometre area is not something a two-person team walks in an afternoon. Those cases are better handed to the carrier, the platform or law enforcement with the region attached.
Put a fleet number on it. A 1,000-device fleet with a 2 percent annual rate of unreturned units sees 20 units per year in this workflow. If 4 out of those 20 reports arrive at cell-tower accuracy and the team sends a visit on all 20 regardless of radius, the wasted visits are the whole difference between a location-assisted recovery process and a location-shaped hope.
An accuracy radius without a timestamp is not a position, it is an anecdote. The value of a last known location decays quickly, and the decay is not linear for a device that is still powered and moving.
Three reported states are worth separating in the record:
There is one case that changes the stale rule. From iOS 15, an iPhone that has been powered down does not go fully dark: it keeps a low-power Bluetooth state and can be discovered by nearby Apple devices, which relay a position back through the Find My network, provided the network is enabled. The same holds after an erase while Activation Lock is still on. That means a device reported as offline can still produce a position — but it also means the absence of a fresh report is weaker evidence than it used to be, and the timestamp on the last report matters more, not less.
| Report age | Satellite accuracy | Wi-Fi accuracy | Cell accuracy |
|---|---|---|---|
| Under 15 minutes | Dispatch a visit | Visit if the unit value justifies it | Region-level triage only |
| 15 minutes to 24 hours | Confirm before dispatch | Region-level triage | Region-level triage |
| Over 24 hours | Treat as a lead on where it was | Treat as a lead | Treat as a lead |
The matrix is only usable if both dimensions are stored as fields. A record that carries coordinates alone cannot be placed in it at all.
Conversely, three conditions produce no report at all: the device is powered down with the network disabled, it is in airplane mode, or it is out of coverage. In all three the correct reading is no observation, which is not the same as no problem.
If the map shows a pin and the record stores only latitude and longitude, the field you need for the decision is missing. Ask for the horizontal accuracy value to be written to the ledger alongside the coordinates.
Sample 10 location records and plot the radius distribution. On a mixed urban fleet you will usually see two clusters. Knowing the split tells you what fraction of reports are actually actionable, which is the number your field process should be sized against.
Confirm the timestamp is stored separately from the position. Search your own records for a report where coordinates and time arrive in one string. If they are fused, you cannot apply a staleness rule, and every report is treated as current.
It is not the same thing. A pin is a rendering choice; the position is a circle. Any process that does not read the radius is silently assuming the best case.
That does not follow. Reporting depends on network reach, power state and user settings. A device can move a long way between two reports.
Not necessarily. A 5-metre report that is 3 days old is generally less useful than a 50-metre report from 10 minutes ago, because the device moves and the building does not.
Boundary one: this is about recovering organization-owned devices, not tracking people. The two are different in purpose, in consent and in law. A fleet workflow locates an asset that belongs to the organization and for which the enrollment and disclosure terms were set out in the agreement. Building movement histories of the person holding it is a different activity with different requirements.
Boundary two: not applicable where local rules restrict location access. Jurisdictions differ on when an employer or a lessor may query device position, on what notice is required, and on whether consent can be a condition of the agreement. The technical capability and the permission to use it are separate questions, and the second one is not answered by the console.
Why does the same device report different accuracy at different times?
Because the available source changes. Outdoors with a clear view, satellite fixes are available; indoors in a dense area, the device falls back to Wi-Fi or cell positioning, which carry larger error.
Yes, but classify it. It is useful for region-level triage and for handing a case over with a defined area, and it is not a basis for sending someone to a door.
A practical rule is 24 hours for treating a position as current, and 15 minutes for treating it as fresh enough to dispatch. Both thresholds should be written down rather than left to the operator.
Not always on recent Apple devices, while the Find My network is enabled — a powered-down device can still be relayed by nearby devices. The behaviour depends on the setting and on remaining battery.
Three fields at minimum: coordinates, horizontal accuracy radius, and the timestamp. Without the second and third, the first has no decision value.
LuckyMDM (Sichuan Starlight Network LLC) gets asked where the line sits between a usable position and a region. LuckyMDM stores the reported accuracy radius and the timestamp of the last known location as two separate fields in the device ledger, and marks any report older than 24 hours as stale rather than current.
Two fields decide it: how wide the circle is, and how long ago it was drawn.