Network technician helping fix WiFi mesh restarting issues

Steve H.

Reviewed for technical accuracy

Specialities: home automation trigger checks, routine conditions, app permissions, presence detection, delayed actions, and manual test vs real trigger behavior.

When an automation runs only in manual test, check the trigger, hidden conditions, permissions, and whether the app sees the event in real time

Technician avatar Technician
Let’s start with the safest checks first, because an automation that works only when tested manually is often not broken in the way it seems.

Manual test proves the action itself can run. What it does not prove is that the trigger is arriving, that the conditions are true at that moment, or that the platform still has permission to run it in the background.

What kind of automation is this: time-based, motion/sensor-based, location-based, or something triggered by another device changing state?
Customer
It is based on a door sensor. When I test it manually, the light turns on. But when the door actually opens, nothing happens.
Technician avatar Technician
That is a very useful split. If manual run works but the real event does not, I first check whether the platform is actually seeing the door event the way the automation expects.

Two low-risk checks first:
1) In the app or automation history, do you see the door sensor change from closed to open at the correct time?
2) Did this start recently after an update, battery change, hub restart, or permission prompt?
Customer
I think it started after an app update. The door sensor still shows open and closed in the app, but the routine does not fire by itself anymore.
Technician avatar Technician
That timing matters. This is common after updates, but we should rule out risks before changing anything big.

If the sensor state still updates, then the trigger may be visible in one place but blocked by a condition, a routine type mismatch, a delay window, or a permission setting that changed quietly after the update.

I don’t want you to rebuild the whole automation yet. Let’s narrow the weak point first. Does this routine have any extra conditions such as time of day, home/away status, another device state, or “only when” rules?
Customer
Yes, it should only run at night. But it used to work before, so I did not think that was the problem.
Technician avatar Technician
That could still be the missing piece. Manual test usually jumps straight to the action, but the real automation has to pass through the trigger and then every condition in order.

A routine can look fine in a test and still never run by itself if the “night” window, home mode, timezone, or presence state is no longer matching.

Safe question: when you say “night,” is that based on a clock time, sunset, a sleep mode, or a household state like Home/Away?
Customer
It is based on a time window, and I think it also checks if we are home. I forgot that part was there.
Technician avatar Technician
That is exactly the kind of detail that creates a false alarm. People see the action run in manual test and assume the automation is healthy, but the real trigger path may be blocked quietly by one extra rule.

Let’s check this safely first: when the door opens and nothing happens, does the automation log show that it was triggered and then stopped, or does it show nothing at all?
Customer
I looked earlier and I do not think it shows anything at all. It is like the routine never notices the door opening.
Technician avatar Technician
That points more toward a trigger path problem than an action problem. In plain language, the light action still works, but the event is not reaching the automation the way the automation expects.

A few common reasons fit that pattern:
• the sensor event is delayed or reported differently after an update,
• the automation is listening for the wrong state change,
• the platform no longer has the right permission or background access,
• the routine type changed from personal to household, or vice versa.
Customer
That sounds possible. I also noticed some permission screens after the update, but I skipped through them quickly.
Technician avatar Technician
That is worth taking seriously. I don’t want you to try something that could make it worse, like deleting the automation and rebuilding from memory, until you check the quiet settings first.

Permissions, presence access, background refresh, and notification/event access can all affect whether the trigger is processed automatically even though the manual test still works.

Another question: if you create a very simple version of the automation with only the door trigger and one light action, does that ever run by itself?
Customer
I have not tried that yet. Mine has a few conditions and a short delay because I did not want it firing every time by accident.
Technician avatar Technician
That delay is another important clue. Delays, cooldowns, and “only if still open after…” logic often behave differently from a manual run because the real event has to remain valid during the waiting period.

So now we have several plausible blocks: time window, home status, permission changes, and delay logic. That is actually good news, because it means the failure pattern is explainable and not random.
Customer
I did not realize manual test could skip over so much. I thought if the test worked, the automation itself had to be fine.
Technician avatar Technician
That is the trap with this kind of issue. A manual test proves the destination works. It does not prove the road to the destination is clear.

The safest next step is to inspect the trigger history, simplify one copy of the automation for observation, and verify whether the platform still sees the same door event, home state, and timing rules the automation expects. Once you know which part is being skipped, the rest becomes much easier to sort out.
Customer
That finally makes sense. I was staring at the light action the whole time, but the real problem may be before that. I will check the trigger history and the extra conditions.

Quick answer: the accepted fix

Accepted first approach when an automation works in manual test but never runs on its own:
• Confirm the real trigger event is appearing in the app or automation history at the time you expect
• Review extra conditions such as time windows, home/away state, delays, cooldowns, or “only if” rules that manual testing may not reflect
• Check permissions, background access, presence settings, and any recent update-related changes that could block automatic execution
• Compare the full automation with a very simple trigger-plus-action version to see whether the trigger path or condition path is failing
• Change one thing at a time and keep notes so you can see whether the block is the event, the condition, or the timing


Why an automation can pass a manual test and still never trigger by itself

A manual test often runs the action directly or under simplified conditions. Real automations depend on the trigger being detected correctly, the conditions being true at the same time, and the platform still having the needed permissions and background access. That is why a routine can look healthy in testing but fail quietly in normal use.

What to check first before rebuilding the whole routine

Start with the trigger history, not the action. Check whether the door sensor, motion event, location event, or scheduled time is appearing where the automation expects it. Then look at hidden blockers such as presence mode, delays, quiet hours, cooldowns, and update-related permission changes. Small checks usually reveal more than starting over from scratch.

Tip: don’t delete the automation until you know whether the trigger or condition is failing

A manual success can hide a quiet trigger or permission problem.

If the action works in a test, focus first on what happens before the action: the event, the timing, and the conditions. Many automations stop running by themselves after an update because permissions, home state, or background behavior changed. Keeping one simple test version beside the full version can make the difference much easier to see.

Disclaimer: This information is AI-generated and intended for general guidance only. Steps vary by automation platform, app version, trigger type, and device setup. For exact instructions for your setup, consult the manufacturer’s support site.

Understand what these prompts mean


Getting Started Is Easy

Low-risk diagnosis first

Focus on safe checks such as trigger history, conditions, permissions, and timing before rebuilding the routine.

Identify what changed

Updates, permission prompts, and quiet condition changes often explain why manual testing still works while automatic runs stop.

Avoid “random fixes”

Changing one variable at a time makes it easier to see whether the block is the trigger, the condition, or the timing itself.

FAQ

Why does my automation work in manual test but not automatically?

Manual testing often proves only that the action can run. It may not reflect the real trigger, time window, home state, delay, or permission needed for the live automation. That is why the routine can pass a test and still never run by itself.

Did this happen after an update?

It often can. App or platform updates may change permissions, background behavior, household status, or how a trigger is interpreted. The automation may look unchanged while one hidden requirement no longer matches.

Is this a settings issue or hardware?

If the sensor or device still reports normally in the app, settings are often the first place to look. The trigger path, conditions, and permissions usually matter more than the action itself in this situation. Hardware becomes more likely if the event is no longer reported reliably at all.

Can this be checked safely at home?

Yes. Start by checking event history, trigger timing, conditions, and background permissions. Avoid deleting the working automation until you know which part is being skipped.

Sources

Matt Verified Icon
Matt ● replies in < 1 min
Technician Expert
⭐⭐⭐⭐⭐ Trusted by 27,000+ customers
Avatar
Alex · Expert's Assistant