Noticing your hours creep up before it becomes a problem

24 September 2026 · 7 min read

Working hours rarely jump. They creep: a 7 p.m. finish becomes 7:30 during a busy fortnight, the busy fortnight ends, the 7:30 stays, and a few months later 8 o’clock feels like the normal time to stop. You catch creep by comparing this week with a fixed reference from weeks ago, not with last week, and a stack of week rolls and month rolls is the easiest place to make that comparison.

Why creep never feels like a change

Every step has a reason. A launch, a colleague on leave, a client in another time zone. Each reason is real, and each step is only fifteen or twenty minutes, which is below the level anyone notices in a single day.

The trouble is what you compare against. When you ask whether today was long, you measure it against yesterday, and yesterday was about the same. Nobody remembers that the same week last spring ended at 6:45, because nothing wrote it down.

So creep is not a failure of discipline. It is a measurement problem: the only instrument checking your hours is a memory that resets its zero every morning.

Compare with a month ago, not with last week

The fix is a reference that does not move.

Punchcard sits in the menu bar and notices which app is in front through the day, app names only. At the closing time you set, it prints a receipt: one line per app with its time, a day total and a stamp. On Sunday it also prints a week roll, and at the end of the month a month roll, without being asked. Those two are what make creep visible.

1. Write the reference once. On the first week roll you have, note the week total and the hours you actually meant to work. Something like “41h 10m; meant 40”. Do not update this line later.
2. Keep the rolls in one place. Receipts can be copied as plain text. Paste each Sunday’s week roll into the same running note, newest at the bottom, so the totals line up in a column.
3. Read the column, not the last entry. Week totals of 41, 42.5, 43, 44 and 45.5 hours never differ by more than ninety minutes from the week before. Read as a column, five weeks have quietly added four and a half hours a week.
4. Do the monthly division. When the month roll prints, divide its total by the number of days you worked that month. That typical-day figure is the single best creep detector you have, because one late night barely moves it and a slow drift moves it every month. Month in review covers the rest of what that roll shows.

If you prefer a chart, Export CSV gives you the underlying record for a spreadsheet.

The signs on a daily receipt

The receipt does not print clock times. It will not say “last app at 8:05 p.m.” What it shows is where the time went and how much there was, and creep leaves marks in both.

The paper arrives mid-task. The receipt prints at your closing time. If it keeps arriving while you are still in the middle of something, and you carry on, your real finish has moved past the closing time you chose. Count how many days a week that happens. One is normal. Four is the new finish time.

Lunch has vanished from the total. If you used to step away for forty minutes and now eat at the desk with chat open, the day total grows by roughly that amount and the chat or browser line gets fatter. You will not see a gap disappear, because there is no timeline, but you will see the total grow with no extra work to show for it.

Evening apps turn up on work days. Most people know which apps they open after dinner: mail, chat, sometimes the calendar. When those lines keep growing on days that were otherwise spent in a design tool or an editor, the evening check has become a regular shift.

The late print arrives in the evening. If the Mac was asleep at closing time, the receipt prints late, headed SORRY WE MISSED YOU. When that header greets you the next morning, nothing is wrong. When it tends to greet you at nine at night, it is a record of how often the laptop gets reopened after you had closed it for the day.

Sunday lines appear. A receipt dated Saturday or Sunday, where there used to be none. It usually starts as twenty minutes of “just clearing the inbox” and grows from there. A short one still counts.

You moved the closing time. If you pushed Punchcard’s closing time from 6:00 to 7:00 because the receipt kept arriving while you were busy, the creep is now written into your settings. That is worth noticing on its own.

Overtime you didn’t notice goes through these hiding places within a single day. Creep is the slower version, where every day looks fine and the months do not.

What to do once you see it

1. Put a number on it. Write the gap between your reference and the current typical day in minutes: “8h 45m against 8h 05m, forty minutes.” A number is easier to act on than a feeling of being busy.
2. Decide whether it was chosen. A launch month or covering for someone on leave is a reasonable reason to work later. Give it an end date and write that date next to the number. Longer hours with an end date are a decision; longer hours without one are the new normal arriving unannounced.
3. Find the line that grew. Put this month’s roll next to the first one and compare line by line. If the chat line grew, the extra hours are about responsiveness, and batching messages will help. If your main work app grew, the work itself grew, and that is a workload conversation with a manager or client rather than a habit to fix. The roll gives you a number to bring to that conversation.
4. Walk it back in steps. Pull the finish back fifteen or thirty minutes a week, not all at once, and move the closing time in Punchcard to match each step so the receipt keeps arriving at the time you mean. How to set a closing time for your workday and keep it has the method for making each step hold.
5. Check the next Sunday roll against the reference line, not against the week before.

What the rolls cannot show

Punchcard records the app in front on this Mac and nothing else. Time on the phone does not appear, so creep can quietly migrate: the receipts improve while the evenings fill up with replies on a phone.

It cannot tell work from personal use of the same app. A long browser line after dinner might be a spreadsheet or a recipe, and only you know which.

It has no alerts and no limits; the rolls are something you read. It has no projects or clients, so it can show that the hours grew but not which piece of work grew them. And it has no team features: the record is one SQLite file on your Mac, seen by nobody unless you share a receipt.

Questions

How much creep is worth acting on? A useful test: if the current typical day had been offered to you as the permanent hours when you took the job, would you have agreed? If not, the gap is worth walking back, whatever its size.

My hours grew because the job grew. Will a closing time fix that? Not on its own. A closing time protects the evening, but if the work genuinely increased, the fix is a conversation about scope, deadlines or help. Month rolls from before and after the change are a clear way to show the increase without arguing about impressions.

Do I need months of data before any of this works? No. The first week roll is your reference, and the comparison gets useful by the fourth or fifth week. Tracking never expires and the first seven receipts are free; after that it is $9 once.

Can Punchcard warn me before I go over my hours? No. It prints at the closing time and on Sunday and month end, and it never interrupts you otherwise. If you want a warning, a repeating calendar alarm fifteen minutes before your closing time does that job.