An ADHD-friendly way to end the workday on a Mac
· 6 min read
Some people have an accurate internal sense of elapsed time. Others do not, and for them the standard advice about ending the workday, notice when you are tired, decide to stop, is built on a capacity they do not have.
If you routinely surface at 19:40 having intended to stop at 18:00, and are genuinely surprised, the problem is not discipline. It is that the passage of time did not register, and no amount of intending harder addresses that.
What does help is moving the timekeeping outside your head.
Why intention fails here
The plan to stop at six is made at nine in the morning by someone who is not absorbed in anything. It is executed at six by someone who is, and who has no felt sense that eight hours have passed.
Nothing in the environment contradicts that. Screens do not get darker. Work does not run out. The one available signal, a clock in the corner, requires you to look at it and to attach meaning to the number, which is exactly the step that does not happen reliably.
So the fix has to be something that arrives without being looked for, and that requires no in-the-moment judgement.
An external signal, at a fixed time
Set an alert for your closing time. Calendar or Reminders both work. Make it recurring so it is not a decision each day.
The important property is that it arrives rather than being noticed. It interrupts. You do not have to be tracking anything for it to work, which is the whole point.
One alert is often not enough. A warning fifteen minutes before gives you time to reach a stopping point, which matters when stopping mid-task is genuinely difficult rather than merely inconvenient. Being told to stop now, while in the middle of something, tends to be ignored; being told that stopping is fifteen minutes away can be acted on.
A record that says what happened
The second half is a record you did not have to maintain.
The reason this matters specifically: reconstructing a day from memory requires the same time sense that was missing during the day. Asking yourself where the hours went produces a guess, and the guess is usually wrong in a consistent direction, undercounting the absorbing work and overcounting the rest.
An automatic record removes the reconstruction. It says what happened, in numbers, without you having remembered to observe anything. Reading it takes thirty seconds and answers a question that was previously unanswerable.
What people tend to report from this is not guilt but calibration. Three hours in one application does not feel like three hours; seeing it written down gradually attaches a size to the feeling. That is a slow effect and a real one.
Keep the ritual short
Two minutes, maximum, and the same every day.
- Write tomorrow’s first task. One line. This gives the unfinished-work feeling somewhere to go other than your evening.
- Quit the work applications. Command-Q, not the yellow button.
- Read the record of the day once. Do not analyse it.
- Close the lid.
The reason for the length is that a twenty-minute shutdown routine gets skipped on the days it is most needed, and a ritual skipped twice is not a ritual. Two minutes is short enough to survive a bad evening. A shutdown ritual that takes two minutes covers the general version.
Step three is worth protecting from expansion. Reading the record is useful; grading yourself on it is not, and a ritual containing a verdict becomes one you avoid.
What to do with the overruns
They will happen. The useful response is to let them be visible rather than treating each one as a failure.
A record that shows you worked until 19:40 on four days out of five is not a character judgement; it is information about a closing time that does not match your actual pattern. Either the closing time is wrong, or the workload is, and both are addressable in a way that resolving to try harder is not.
Overtime you did not notice covers spotting the pattern, and the “one more thing” trap covers the specific mechanism at the end of a day.
Where Punchcard fits
Punchcard sits in the menu bar, notices which application is in front by name, and prints a receipt of the day at the closing time you set: one line per application, a total, a stamp. You drag down to tear it off.
Two properties matter for this particular use. It requires nothing from you during the day, which means there is no step to forget and no timer to have failed to start. And the receipt arrives at a fixed time rather than waiting to be asked for, which puts the signal in the environment instead of relying on you to seek it out.
It has no timers, projects or tags, and it is not a productivity system. It reports what happened. Time tracking for people who hate time tracking covers why that constraint is the point rather than a limitation.
Questions
Will an alert not just get dismissed? Sometimes, and it still works better than no alert. Pairing it with a physical action, closing the lid, standing up, helps considerably, because the action is what ends the day rather than the notification.
Is seeing the numbers discouraging? It can be at first, particularly the total. It is worth persisting through the first week, because the value is in calibration over time rather than in any single day. Read the receipt; do not grade it.
I work in intense bursts with gaps. The record will show that, which is genuinely useful. A day with four productive hours and long gaps is a real pattern, and knowing its shape is better than assuming it should look like someone else’s day.
Does this replace other strategies? No. It addresses one specific thing: knowing where the time went and having the day end at a fixed point. Everything else is separate.