How long should a task take?
The short answer
There is no correct duration for a task, and anyone offering you one does not know what the task is.
The useful question is different: what is the estimate for? It is not a prediction and it is not a commitment. It is the only mechanism that converts a list into a capacity check. Three items with no times attached will always look like they might fit the day. Three items at ninety minutes each are four and a half hours, and that either fits or it visibly does not.
An estimate that turns out wrong has still done its job if it made the arithmetic visible before the day started rather than at six o'clock.
Why accuracy is not the point
Estimates are bad. Everyone's are, consistently and in the same direction, and getting better at them is slow.
This matters less than it appears, because the estimate is doing three things that do not require it to be right:
It forces the task to be specific. You cannot estimate "sort out the website". The act of putting a number on something reveals that it is actually four things, and that discovery is worth more than the number.
It makes overcommitment visible in advance. The sum is what you need. If the sum exceeds the hours available, that is true regardless of whether the individual figures are good, and it is the single most useful fact about a day.
It gives you something to compare against afterwards. Planned versus actual is the only way estimates improve, and you cannot have it without a planned figure — even a poor one. This is also the only honest feedback loop about how your days actually go, as opposed to how they felt.
A practical way to do it
Estimate in blocks, not minutes. Thirty minutes, an hour, two hours, half a day. Finer resolution is false precision and takes longer to decide.
Estimate the focused time, not the elapsed time. Two hours of work does not fit in a two-hour window containing a meeting. Estimate the work, then look separately at whether the day has that much uninterrupted space in it — usually it has much less than it looks.
If something is over half a day, it is not one task. Break it until the pieces are estimable. An unbreakable large estimate almost always means the task is not understood yet, and "understand it" is the real first item.
Write the number down before you start. An estimate recalled afterwards is not an estimate; it is a memory adjusted by what happened.
What to do when you go over
Add the time and carry on. That is the whole procedure.
Overrunning is information about the estimate, not a verdict on the work or the person doing it. A method that marks an overrun as a failure will simply teach you to pad estimates until they are meaningless, which destroys the capacity check that was the only reason to estimate at all.
The useful move is at the end of the week, not in the moment: look at where planned and actual diverged most. Those are the categories where you do not yet understand your own work, and they are usually a small and specific set — not everything, and not random.
The limits of this, honestly
Some work genuinely cannot be estimated, and pretending otherwise is worse than admitting it. Research, debugging, and anything whose duration depends on a thing you have not found yet are all in this category — the answer is not a better estimate, it is a timebox: decide how much time the question is worth, spend it, and then decide again with what you learned.
That is a different instrument from an estimate and it is worth knowing which one you are using. An estimate predicts; a timebox limits. For unknowable work the limit is the honest tool.
And no arrangement of estimates fixes a day that holds more work than hours. That is a capacity problem, and the estimate's job there is only to prove it early enough to tell someone.
Where to go next
- The Top 3 method — three lines, each with a duration
- What to do when you have too much to do — when the sum does not fit
- How to prioritise your tasks — which method today is asking for
Using this in Decide One
Every line in Decide One takes a duration, because a line without a time is an intention rather than a plan. Over weeks it shows planned against actual — where the time went, and whether the estimates hold.
Running over adds the time and says nothing about it. A line you did not finish is unfinished, it is not marked against you, and a day you did not open is not counted.
Open Decide One — free to use, with no account. Nothing you write leaves your device.
Decide One is a priority instrument. Bring the day into view, choose a method, give the first thing real time.