If you are twenty minutes into a session you did not choose, do this now. Stand up. Put the phone in another room, screen down. Say the time out loud. Then, once the session is over rather than during it, write a stop rule for next time, because a rule invented mid-session is not a rule, it is a negotiation you will lose.
Right now, in the next sixty seconds
- **Stand up.** Not later, now, before you finish this paragraph. The posture is holding the session together more than the interest is.
- **Put the phone somewhere with a door.** Another room, or a drawer, screen down. Arm's length is not far enough, because arm's length is one reach.
- **Say the time out loud.** "It's twenty to one." Saying it does something that thinking it doesn't, which is that it makes the hour a fact rather than a background condition.
- **Write one line in a note.** What you came into the app for, and what you actually got. Six words is enough. That line is your material for later.
That is the whole emergency version. You are not trying to feel resolved, you are trying to end a session.
Why "just close it" fails at minute twenty
A swiping session has no natural end. An episode finishes, a chapter finishes, a pint gets empty. A deck of profiles does not. There is no point at which the app tells you that you are done, so the only stopping signal available is a decision you have to keep making, and each one costs a little more than the last.
The economics inside the session are also against you. A swipe costs nothing. The next one might be the one that matters. Twenty minutes in, tired, you are not evaluating people any more, you are clearing a queue, and clearing a queue is a task with momentum.
That is why the interrupt has to be physical and pre-decided. You are not going to out-argue yourself at half past midnight, and you do not need to. You need the phone to be in a different room.
The session you did not decide to start
Worth noticing, because the mid-swipe interrupt is the second-best moment to act. The best one is thirty seconds earlier.
Almost no long session begins with a decision. It begins with a badge, or with an unlock that had a different purpose, or with a thumb that goes to the fourth icon on the second row without consulting anybody. By the time you are aware of being in the app you are already two minutes in, and two minutes in is a different state from not having opened it.
Three things change that.
**Notifications off entirely.** Not important-only, which is a setting that means "still interrupt me". Off. The badge is the thing that starts sessions you did not choose.
**The icon off the home screen.** Search for the app by name instead. Typing four letters is a decision. Tapping a picture in a position your thumb has memorised is not.
**A question attached to the unlock.** Some people write one line in the note that opens on unlock: "what did you pick the phone up for?" It sounds twee and it works more often than it should, because the honest answer is usually something else.
What actually has evidence, and what doesn't
Worth saying plainly, because every other article on this subject is confident about things nobody has shown.
The strongest causal finding in this whole area comes from blocking mobile internet at the network level for two weeks. It ["improved mental health, subjective well-being, and objectively measured ability to sustain attention; 91% of participants improved on at least one of these outcomes"](https://pubmed.ncbi.nlm.nih.gov/39967678/). Texts, calls and desktop internet still worked, so that is friction rather than abstinence. Nobody has run it on dating apps specifically.
Grayscale mode is the other one with real data behind it. In a trial it ["cut measured screen time immediately and significantly"](https://pubmed.ncbi.nlm.nih.gov/36577008/), while producing "no immediate causal effect of reducing usage on subjective well-being and academic performance". Self-set time limits worked in the same trial, but more slowly and less.
Now the honest part. Notification batching, weekly offline days, dumbphones and consumer app blockers have no efficacy trial behind them at all. People say they help. That is not the same as evidence, and you deserve to know which one you are being handed. The workable summary is that friction reliably changes behaviour, and nobody has shown that it changes how you feel. Getting your evening back is worth having on its own.
The stop rule, written before you open it
Do this once, tomorrow, in daylight. Three parts, all with numbers in them.
**A number.** Twenty minutes, or fifteen likes, whichever you hit first. Pick one you would still accept on a Sunday night, not one that sounds impressive.
**A place.** Anywhere except the bed and the sofa you fall asleep on. Kitchen table, standing up, waiting for a kettle. Where you are sitting predicts how long you stay more than intention does.
**An exit.** The specific thing you do next, decided now. Not "go to bed", which is a category. "Put the kettle on and take the bins out" is an exit, because your body knows what to do with it.
Then write it in the note that opens when you unlock the phone. A rule you can read at 11pm beats a resolution you made at noon.
Three interrupts that hold
- **A timer that makes noise in the room.** Not the phone's silent one. A kitchen timer, a smart speaker, anything that goes off out loud, because a sound in the room is external and the phone's timer is just the phone talking to itself.
- **Grayscale.** It is the only cheap tweak with a trial behind it. On iPhone it lives under Accessibility, Display and Text Size, Colour Filters. On Android, look for a greyscale toggle in Digital Wellbeing or Developer options.
- **The pre-committed next thing.** Something at 10pm with another human in it, even a phone call. Sessions end when there is somewhere to be.
If you would rather the phone held the line at 11pm than you did, [Shield](/shield) puts real friction between you and an app. A screen time limit with a passcode a friend holds does much the same for free.
When the rule keeps failing
If you have set a stop rule three or four times and blown through it every time, nothing has gone wrong with your character. You have run a small experiment and it has returned a clear result, which is that moderation is not available to you at the moment. That is worth knowing rather than repeating.
At that point the useful move is not a stricter rule but a different job, which is [deleting properly and replacing what the app was doing](/guides/dating_apps/deleting-the-apps-and-what-to-do-instead). And if the sessions are always starting from the same mood rather than the same hour, [find the mood first](/guides/dating_apps/when-do-you-open-them).
Common questions
**How do I stop swiping when I'm already in a session?** Stand up and physically move the phone out of the room. Any plan that keeps the phone in your hand is asking your worst hour to do the work.
**Do app blockers work for dating apps?** No trial has tested consumer blockers, on dating apps or anything else. What has been tested is heavier friction at network level, which improved wellbeing and attention over two weeks in a randomised study. Treat blockers as friction that might help, not as a solution that is known to.
**Is grayscale actually worth it?** It is the only small tweak with real data behind it, and the data says it cuts usage rather than lifts mood. If getting an hour back is the goal, it earns its place.
**Twenty minutes turns into two hours every time. What now?** Stop treating that as a willpower failure and read it as a result. Repeated failure to hold a limit is the signal that a limit is the wrong tool for you right now.
Sources
- [Blocking mobile internet on smartphones improves sustained attention, mental health, and subjective well-being, PNAS Nexus](https://pubmed.ncbi.nlm.nih.gov/39967678/). The two week network-level block, and the 91% figure.
- [Effects of grayscale mode and self-set time limits on smartphone use](https://pubmed.ncbi.nlm.nih.gov/36577008/). Grayscale cut usage immediately, with no immediate effect on wellbeing.
- [Problematic Online Dating, Journal of Medical Internet Research, 2025](https://pubmed.ncbi.nlm.nih.gov/40608479/). 32 studies, only 3 randomised, and the concept remaining debated.
