Smart lock app says locked, but the door is unlocked: here’s why

The app says locked, complete with a green checkmark, and the door still swings open under your hand. Or it runs the other way: the app insists the door is standing open while you can see the deadbolt thrown from across the room. A smart lock app that says locked when the door is unlocked, or the reverse, is one of the most common complaints on lock support forums, and in nearly every case nothing mechanical has failed. Two separate systems are reporting two separate things: a small sensor that watches the door itself, and a lock mechanism that only knows its own motor position. When they disagree, the app has to show one of them, and it isn’t always the one you’d trust.

Two different kinds of “wrong,” and why they look the same

Every mismatch complaint lands in the app looking identical: a status that doesn’t match reality. Underneath, there are two genuinely different causes, and they call for different fixes.

• A lock with a dedicated door-position sensor, often sold under a name like DoorSense, can be flat wrong about whether the door is open or closed even though the bolt itself moved exactly as commanded.
• A lock with no separate sensor at all only ever knows where its own bolt motor stopped. It never knew what the door was doing in the first place, so there’s nothing to recalibrate: the report was correct by its own narrow definition the whole time.
• A lock that reports over a mesh radio or a weak connection can simply be slow: the app is showing the last status it received, not a live one, and the two catch up a few seconds or minutes later.

The first cause is the one most people mean when they say the app and the door disagree, so it gets the most space here. The other two are worth ruling out first, because neither one has anything to calibrate.

Locks without a separate door sensor only know where the bolt is

Not every smart deadbolt has a door-position sensor. Plenty of keypad and app-controlled locks, especially simpler or older models, only carry a motor encoder that tracks how far the bolt itself has turned. That number gets translated into “locked” or “unlocked” in the app, and that’s the entire picture the lock has. It has no idea whether the door and frame are actually mated, whether the door is propped, or whether the strike plate caught the bolt cleanly.

That matters because a bolt can throw its full travel into a misaligned strike, or into a door that’s swollen and sitting slightly proud of the frame, and the motor will report success without contradiction. Nothing in the app flags this, because nothing in the lock is watching the door, only the motor. A smart lock showing a clean “locked” status is not, by itself, proof that a door is actually secure on a lock built this way. If the bolt itself is binding, stalling, or refusing to fully extend, that’s a mechanical problem with its own causes, and the alignment and binding issues that stall a deadbolt motor cover it in more detail than fits here.

If your lock has always behaved this way and you’ve never seen a “door ajar” alert or a separate door-open notification, this is probably your lock, and the rest of this article’s sensor-specific fixes won’t apply. Skip ahead to the section on network lag, or to the checklist near the end.

How a door-position sensor like DoorSense decides open or closed

Locks that do carry a separate door sensor use a small housing mounted on the door frame, next to the lock body, with a magnet inside the part that faces the door. August calls its version DoorSense, and Yale ships a sensor under the same name on several of its own lock lines. Both work the same way: the housing reads the strength and position of the magnetic field, and as the door swings, the magnet’s position relative to the sensor changes enough for the lock to tell open from closed. August’s own support documentation puts it plainly: “DoorSense’s technology uses magnetic fields to help determine the state of your door as open or closed.”

That status then gets layered on top of the bolt’s own motor position. A lock can throw its bolt correctly and still show “door open” if the sensor’s magnetic read disagrees, and that disagreement is what produces the exact complaint this article opened with. The sensor, not the lock mechanism, is almost always the part that’s actually wrong.

Why the reading drifts: distance, alignment and magnetic interference

Three things push a door sensor’s reading out of step with the real door, and all three show up in the manufacturers’ own installation guidance rather than in anything specific to one bad unit.

• Distance from the lock. August’s installation guidance is specific: the sensor should sit “within 1.5 inches of the frame edge” and of the extended deadbolt, and warns that a deadbolt sitting more than 1.5 inches from the frame can make the sensor hard to calibrate at all. Yale’s own installation article gives the same 1.5 inch figure for a surface-mounted sensor, and 1.5 to 3.5 inches from the top of the strike plate for a flush-mounted one.
• Alignment. Both brands’ instructions call for lining the housing up squarely with the lock’s sensing element, what August’s guide calls aligning “the pill shape vertically.” A housing mounted slightly crooked, or on an uneven frame, reads the field at an angle instead of straight on.
• Outside magnetic interference. August’s support material states directly that “any changes to the magnetic field surrounding the lock may affect the accuracy of the DoorSense readings,” and names removing or swapping the lock’s own faceplate as one trigger. A second magnet placed near the lock, a metal security plate added to the door later, or even a magnetic phone mount near the frame can do the same thing.

None of these is a defect. They’re installation variables that can drift weeks or months after a lock was set up correctly, especially if a door has swollen with the season or a frame has shifted slightly, which moves the whole housing relative to the magnet without anyone touching it on purpose.

Recalibrating the sensor, step by step

Both August and Yale route this through the same kind of menu, usually under lock settings, and the steps barely differ between them.

1. Open the lock’s settings in the app and find Calibrate Lock.
2. Choose the option that calibrates the lock and the door sensor together, rather than the lock alone.
3. Follow the on-screen prompts, which will ask you to open and close the door at specific points.
4. Once it finishes, open and close the door again and watch the app to confirm the status actually matches.
5. If the status is still wrong, or the calibration fails outright, move the sensor housing a little closer to the lock body and run the whole routine again.

One detail worth keeping: August’s own guide recommends leaving the housing’s mounting screws loose until calibration succeeds, then tightening them, so the act of screwing it down doesn’t nudge the alignment you just confirmed. If readings drift again weeks later without anyone touching the hardware, both brands’ support material treats that as normal and points back at the same calibration routine rather than at a faulty part.

When it’s the network, not the sensor, that’s behind

Not every mismatch is a sensor problem, and the fix above does nothing for this second cause because there’s nothing to recalibrate. A lock that reaches the app over Wi-Fi with its own radio generally pushes a status update the moment it locks, since it has a direct line to the internet. A Bluetooth-only lock, or one riding a Z-Wave or Zigbee mesh through a hub, works differently: it reports in on its own schedule, or whenever the hub next polls it, which can be seconds or occasionally minutes after the keypad or thumbturn actually moved. Which radio a lock was built around decides which pattern you’re dealing with, and it’s worth knowing before you assume anything is broken.

The practical difference is how long the disagreement lasts. A sensor-calibration problem stays wrong indefinitely, showing the same incorrect state no matter how many times you lock and unlock the door. A network lag clears itself within a short, fairly consistent window and then matches. If the gap keeps growing, or the app stops updating at all rather than just running a few seconds behind, that’s a connectivity fault rather than a status mismatch, and working out what actually went offline is the more useful place to look next.

What to check right now, in order

1. Stand at the door and test it by hand. Confirm what the bolt and the door are actually doing before trusting either reading.
2. If the lock has a door sensor and the mismatch is consistent, run the lock-and-sensor calibration described above before anything else.
3. If the physical door matches reality but the app took a few seconds to catch up, that’s very likely the network, not a fault, especially on a Bluetooth or hub-connected lock.
4. If the lock’s own activity log shows the status changing with nobody at the door at all, rather than just lagging behind a real event, that’s a different problem with its own set of causes.
5. Check whether the manufacturer has a published firmware update before resetting anything. Door-sensor drift is common enough that some of it ships with a fix already.
6. A factory reset belongs at the very end of this list. It clears the sensor’s calibration data along with everything else, which is exactly the evidence you’d want if the first attempt at recalibrating didn’t hold. Where each brand hides its reset control, and what it actually erases is worth reading before pressing anything.

Common questions

Does a wrong status reading mean the lock isn’t secure?

Not by itself. A mismatched display is a reporting problem, separate from whether the bolt actually threw. It’s still worth testing the door by hand once, since a bolt that won’t fully seat is a real, separate issue rather than a sensor glitch.

Why did this start right after I changed the batteries?

Opening the battery cover can nudge the lock body or the sensor housing just enough to change the magnetic field the lock calibrated against, which is exactly the kind of change the manufacturers’ own support material names as a cause of drift. Running the calibration routine again after a battery swap clears it in most cases.

Is this different for Z-Wave or Zigbee locks than Wi-Fi ones?

Yes. A lock on a hub-based mesh radio only reports in on its own schedule or when the hub polls it, so a short lag after locking is ordinary and isn’t the same kind of fault a dedicated door sensor produces.

Can I just ignore it if the bolt still throws fine?

For a brief, consistent delay, yes. A gap that never closes, or a status that’s wrong at the door itself rather than just slow to arrive, is worth clearing with the calibration routine once, since an alert nobody trusts stops being useful.

The short version

An app that disagrees with the door is almost always one of two things: a door-position sensor like DoorSense reading the wrong magnetic state because of distance, misalignment or interference, or a lock on a slower radio simply showing the last status it received rather than a live one. Test the door by hand first. If the lock has a dedicated sensor, run the combined lock-and-sensor calibration in the app, and move the housing closer to the lock if it still disagrees afterward. If the lock has no separate sensor at all, a short lag between the keypad and the app is normal rather than a fault. Save the factory reset for last, since it erases the very calibration data that would have shown what went wrong.