A reminder on a schedule is wrong most of the times it fires, and being wrong is what kills it. Nine o'clock arrives whether or not you are scrolling. If the prompt appears during a film you deliberately sat down to watch, it has just taught you that this category of message is noise, and that lesson does not un-teach. A prompt tied to what you are actually doing is never wrong, so it never has to be ignored.
The cost of firing at the wrong time
Habituation is the only real enemy in this category. A signal that repeats without carrying information gets filed as background within about two days, and after that the prompt is not weak, it is invisible.
Scheduled reminders generate wrong firings by design. They cannot see that you are on a call, mid-film, reading, or asleep on the sofa. Every one of those is a false positive, and false positives are far more expensive than missed events, because a missed event costs one occasion while a false positive costs the credibility of every occasion afterwards.
This is the same slope that makes iOS app limits stop working within a fortnight. The limit screen appears in contexts where it is obviously wrong, so the tap that dismisses it becomes automatic, and by the time it appears in a context where it is right you are no longer reading it. The full anatomy is in you tap Ignore Limit before you read it.
What behaviour-triggered means in practice
The alternative is to attach the prompt to the thing rather than the clock. Nothing fires unless a named app is open and has been open for a while. Close the app and the count stops. Open it for forty seconds to reply to somebody and nothing happens at all.
Three properties follow.
It cannot be wrong. If it fired, you were scrolling. There is no case where the prompt appears and you can fairly resent it, and that is the whole reason it stays perceptible after week two.
Short use is free. Checking a message, looking up a train, posting a photo. None of this is the behaviour anyone is trying to change, and a tool that taxes it is being punished for its own imprecision.
The person cannot game the schedule. With a nine o'clock reminder you learn to start at nine fifteen. With a duration trigger there is nothing to route around, because the trigger is the activity itself.
Why most tools do it on a clock anyway
Because it is much easier. A schedule needs no permissions, no monitoring, and no platform support. Behaviour triggering on iOS requires Screen Time authorisation, which means asking for something that looks invasive during onboarding, and a meaningful share of people say no.
So the category defaults to schedules, and then compensates for the imprecision by firing more often, which accelerates exactly the habituation it is trying to avoid. That escalation reflex has the same ending everywhere it appears, traced in strict mode made it worse.
Where a clock is still correct
One place: the daily appointment.
A single reminder at a time you chose is not trying to interrupt anything. It is a fixture, and its job is to exist rather than to catch you. That is a different mechanism with a different failure mode, and it works best when the notification does not persist, for the reasons in a notification that leaves before you read it.
Merging the two is the common mistake. A daily appointment that also tries to catch you scrolling is too frequent to be an appointment and too polite to interrupt, and it ends up doing neither job.
The two dials, set together
Behaviour triggering only answers when the count starts. Two more numbers decide whether the result is useful.
How long the app has to be open before anything happens, which for most people lands near twenty minutes, worked through in how often should a screen time app check in. And what happens when it does fire, which has to be small enough to be paid rather than argued with, covered in what should happen when your phone interrupts you.
Get the trigger right and the other two get easier, because a prompt that is never wrong is allowed to be rarer and gentler than one that is wrong half the time and has to shout to be heard.
This is how reCalm decides when to appear: triggered by the scroll, not the clock, with both dials set together and a first ask small enough to pay. The wrong-moment problem is designed out rather than apologised for.
Frequently Asked Questions
What is the difference between scheduled and usage-based reminders?
A scheduled reminder fires at a time regardless of what you are doing. A usage-based one fires only after a named app has been open for a set stretch. The second cannot produce false positives, which is why it stays noticeable.
Why does my screen time reminder stop working after a week?
Because it fired at times when it was obviously wrong, so dismissing it became automatic. Once that reflex exists, correct firings get dismissed too.
Does behaviour triggering need Screen Time permission?
On iOS, yes. That is the honest cost, and it is the reason many apps stay on a clock instead.
Should short app use trigger anything?
No. Forty seconds to reply to a message is not the behaviour anyone is trying to change, and taxing it is how a tool loses the argument in its first week.