The Two-Minute Rule Is Overrated. Here's What Works Instead

The rule comes from David Allen's Getting Things Done, and in its original setting it is sound. Allen proposed it as part of a processing step — a defined window in which you empty an inbox of collected items and decide what each one is. Within that window, if an item can be finished faster than it can be filed, finishing it is cheaper.

That context has almost entirely fallen away from how the rule gets repeated. It now circulates as a standing instruction: whenever something small appears, anywhere, at any moment, do it immediately. Which is a different rule, and a much worse one.

What goes wrong

Applied continuously, the two-minute rule has three failure modes that compound.

It ignores the cost of switching

A two-minute task does not cost two minutes. It costs two minutes plus the time to rebuild whatever mental state you abandoned to do it. For shallow work that surcharge is negligible. For anything requiring you to hold a structure in your head — writing, debugging, analysis, design — it is regularly ten to twenty minutes, and sometimes the thread does not come back at all.

Six two-minute tasks scattered through a morning are not twelve minutes of overhead. They can comfortably consume the entire morning's capacity for deep work while showing twelve minutes on any timesheet.

It rewards whoever interrupts you most

Two-minute tasks are overwhelmingly generated by other people. Requests, quick questions, forms, approvals, "can you just." Your own important work almost never arrives in two-minute units, because meaningful work is by nature larger than that.

So a rule that prioritises by size, and only by size, systematically prioritises other people's agendas over your own. Not because their work matters more, but because it happens to arrive pre-cut into pieces the rule likes.

Nobody can estimate two minutes

"Quickly reply to that email" becomes eleven minutes once you re-read the thread, check a figure, and decide how to phrase the awkward part. The rule depends on an up-front estimate that people are reliably bad at making, and the error only ever runs one direction.

A better filter

Keep the insight — small things should not consume disproportionate overhead — and drop the always-on trigger. Three changes do most of the work.

1. Batch instead of interleaving

Collect small tasks in one list rather than doing them as they land. Then clear the list in one or two dedicated passes a day. You pay the context-switch cost once for the batch instead of once per item, and the tasks themselves go faster because you stay in the same shallow mode throughout.

Two windows works well for most people: one mid-morning, after your first block of real work, and one late afternoon when your capacity for hard thinking has already gone.

2. Ask about consequence, not duration

Replace "can I do this in two minutes?" with "what happens if this waits until my next batch?" Almost always: nothing. Occasionally something real — a colleague is blocked, a deadline is today. Those are the genuine exceptions, and there are far fewer of them than the interrupt-driven day implies.

This filter has the property the original lacks: it sorts by importance rather than by size.

3. Protect one block before anything else touches it

Give your most demanding work a defined block, ideally early, and let the small tasks queue behind it. The queue does not grow indefinitely — that is the fear, and it is unfounded. It grows until the first batch window and then empties.

Being responsive is a real virtue. It is just not the same virtue as being effective, and the two-minute rule quietly trades the second for the first.

Where the rule still earns its place

Two situations, both narrow.

Inside a processing session. Its original home. When you are deliberately working through an inbox, finishing a small item beats filing it. You are already in shallow mode; there is no state to lose.

For physical friction. Hanging up the coat, rinsing the cup, putting the tool back. These accumulate into visible mess, the interruption cost is near zero, and doing them immediately is simply correct.

What the rule should not be is a default posture toward incoming work. Batch the small things, judge them by consequence rather than length, and defend one block of the day from all of it. That keeps the useful part and discards the part that hands your calendar to whoever types fastest.