Published 2026-11-09 · LuckyMDM Blog
The one-line version: locking a fleet unit into a single app is two different tools, not two settings of one tool. Guided Access is an accessibility feature whose exit key lives on the device — whoever holds the passcode can leave. App Lock is a configuration profile whose exit key lives in the console — the lock ends when the profile is removed. Pick the exit first, and the lock follows.
Guided Access lives under Settings > Accessibility, and you start it by triple-clicking the side button. Its lineage explains its behavior: it was built for classroom and kiosk-style use, not for asset management. It takes a local passcode, can disable regions of the screen, freeze touch, and (since iOS 12) run against a time limit.
The passcode is the whole story. It is stored on the device, set by whoever set it up, and recoverable by nobody else. Locking 3 units with it is fine. Locking 40 turns passcode management into its own workstream — one shared code is effectively public, one code per unit means maintaining a table that has to survive staff turnover.
App Lock is a configuration profile payload (`com.apple.app.lock`) that names a target app by its bundle identifier. Install the profile and the device enters single-app lock; remove the profile and the lock ends. There is no on-device field where someone types a code to walk out.
Two premises have to hold: the device is supervised, and the named app is already installed. Miss either and the lock has nothing to land on — which is where most silent failures in this area come from.
| Dimension | Guided Access | App Lock profile |
|---|---|---|
| How it starts | Triple-click the side button on the device | Console pushes the profile to the fleet |
| Premises | None — available on every unit | Supervised state + target app installed |
| Who can exit | Whoever holds the local passcode | The management side, by removing the profile |
| Batch capability | Unit by unit, by hand | One action covers a device group |
| Forgotten passcode | Recovery mode and an erase | No passcode exists to forget |
LuckyMDM (Sichuan Starlight Network LLC) records the App Lock payload identifier, its push time and its removal receipt as three ledger fields in its device management system, so a demo unit is only counted as unlocked when the profile removal, the reconnection and the receipt all line up — not when someone on site reports the screen looks normal.
App Lock depends on the target app being present. The deployment order has three steps: enroll the device into supervision, deliver the demo app through the management channel, then push the App Lock profile. Reversing it produces two failure shapes: the profile arrives before the app exists, so the lock has no target; or the app on the unit is a different build whose bundle identifier does not match the profile.
Every step has a receipt. Enrollment has a registration record, the app install has its own receipt, the profile install has another. Read the three timestamps in order and you can tell whether the sequence was respected — receipts whose times run backwards are the signature of a process that was assembled in the wrong order.
While single-app lock is active, system-level dialogs and update prompts are held back, and they surface in a batch after exit. That is not a fault, it is the mechanism: the device is constrained to a single foreground app and most system interaction points are parked. Seeing a stack of deferred notifications on the day a unit is unlocked for return is expected behavior.
Whether a profile can be deleted on the device depends on its removal setting. For store-floor units, the App Lock profile has to be set as non-removable — otherwise anyone holding the unit can go to Settings > General > VPN & Device Management and remove it with one tap. This is the most commonly skipped switch in the whole arrangement, and the most commonly missed item in a pre-deployment check.
On a 500-unit fleet, 8 percent is 40 store demo units. Via Guided Access, each unit needs setup, a passcode, and a triple-click test — call it 6 minutes each, so 4 hours for the fleet. That is only the launch cost. The passcode is the compounding one: one shared code for 40 units is effectively public, and per-unit codes need a maintained table that has to survive every staffing change.
Lose the Guided Access passcode and the unit is stuck inside that app. The root cause is not a bug: the exit path was designed to be passcode-gated, and there is no recovery path other than erasing the device. The routine way out is recovery mode, which erases the unit: about 40 minutes plus 20 more to re-provision, an hour per unit. And a demo unit locked for a day is a missing demonstration on the floor, which does not show up in any labor calculation.
App Lock goes out as one console action: select the group, push, watch receipts. For 40 online units, installation receipts return within minutes. Unlocking is one removal command — no site visit, no passcode. The difference between the paths is not a multiple, it is a category: one distributes passcodes and labor across every device, the other keeps both on the management side.
On a sample unit, open Settings > General > VPN & Device Management and find the App Lock profile. Look at whether the Remove button is greyed out. Grey means the non-removable setting is in effect; active means the lock's strength depends on floor staff not looking for it.
Enroll, install the app, push the lock — waiting for each receipt. Confirm the units actually entered lock, then remove the profile from the console and time how long the units take to return to normal navigation. Run the chain on 3 units before the batch of 40: the failure radius becomes 3 instead of 40.
While a unit is locked, remove the target app remotely and observe what the lock does and what receipt comes back. There is no single documented answer across system versions, so the honest move is to test on your own units and write the observed result into your runbook rather than quoting someone else's.
Guided Access is an accessibility feature with a local passcode. Counting it toward a coverage metric treats a lock with no managed exit as an asset-safety measure. The test is not whether the unit is locked — it is whether the exit requires the management side.
Lock strength and return throughput have to be tuned together. A lock that is hard to exit slows every settlement and return; enabling non-removable profiles means the removal path has to be rehearsed too. Security and turnaround are a pair of parameters — tuning only one guarantees the other one bites.
A successful push means the profile reached the device. Whether the lock engaged depends on supervision state and the app being present. Look one step further: did the unit actually enter lock, and does the receipt confirm it. Marking "pushed" as "locked" is signing for delivery and calling it inspection.
Boundary one: unsupervised units cannot use App Lock. Devices bought second-hand or never enrolled through ABM will refuse a profile carrying the App Lock payload. For those units only Guided Access remains, and the exit-key problem has to be designed around separately.
Boundary two: the lock granularity is the whole foreground, not individual features. Single-app lock constrains the device to one app. Feature-level on/off switches belong to the restrictions payload, not to App Lock — the two get conflated regularly.
Boundary three: consumer-grade apps can drift under the lock. If the locked app is a consumer service requiring sign-in, a major version update can change which functions are reachable inside it. Lock policy has to follow the app's version, which makes "configure once, ignore for two years" the wrong operating mode here.
Technically App Lock supports it, and two conditions make it defensible: the contract states the trigger and the release conditions in advance, and a notification precedes the lock. The device is in the fleet; the usage rights during the term belong to the customer — legality comes from the agreement, and technology only executes it.
Enter the passcode if anyone still knows it. Without it, recovery mode is the path, and the device will be erased — so before demo units go down this road, confirm there is no local data worth keeping.
Yes, if the bundle identifier is written correctly. Locking into Safari or the Camera is technically fine, though rental and retail scenarios usually lock a business-owned app; built-in targets are more common on dedicated terminals.
By store. One group per location, one push per group, receipts reconciled per group. A single mixed group buries one store's failures inside another store's receipts and doubles the triage cost.
The lock is held on the device and does not depend on connectivity. A removal command issued while offline queues until the unit reconnects — so the lock holds, and the unlock waits for the device to come back. Demo units normally live on store Wi-Fi, where this is satisfied by default.