How long a focus block should be (it is not twenty-five minutes for most people)

The twenty-five minutes ran out at the exact moment the work got good.

It has happened often enough now that you have started to suspect the technique, which is fair, but it is worth being precise about what went wrong. It was not your discipline. It was not the tomato. It was the unit.

Twenty-five minutes is a real number with a real history. A university student in Italy in the late 1980s could not get himself to start, so he took a tomato-shaped kitchen timer off the kitchen counter and wound it. His name is Francesco Cirillo, and by his own telling he began with a much shorter interval and worked upward by trial and error until twenty-five felt survivable. That is an honest way to arrive at a number. It is also one man's answer to one man's problem, and it has since been printed on about ten thousand app icons.

It was never a measurement of how long a person can concentrate. It was a measurement of how long that particular student could bear to sit down for.

The honest answer to how long a focus session should be is three times your run-up

Every piece of work costs something to get into. Call it the run-up: the time between sitting down and producing the first thing you keep. For answering email the run-up is close to zero, because there is nothing to reload. For a spreadsheet you last touched on Thursday it might be twelve minutes of working out what the columns meant. For a piece of writing that has a structure in it, twenty minutes is ordinary and is not a character flaw.

Measure yours. Next time you sit down, write the clock time on a bit of paper. Then write down the time you produce the first sentence, line of code or row you do not immediately delete. The gap between the two numbers is your run-up. Do it on three ordinary days and keep the largest of the three, because the largest one is the one that turns up on the day you slept badly.

A block shorter than three run-ups spends more than a third of itself getting in.

That is the whole rule. Your block should be at least three times your run-up. A twelve minute run-up inside a twenty-five minute block leaves thirteen minutes of actual work, and those thirteen minutes arrive in the second half, which is exactly when the bell goes.

The twenty-five minute block is for work you are avoiding, not work you are doing

There is a job the short block does better than anything else, and it is not concentration. It is starting. When the real problem is that you have not opened the file since Tuesday, twenty-five minutes is small enough to agree to, and agreeing is the entire fight.

It also suits any work whose run-up is near zero and which comes in natural seams: replies, invoices, marking, expense claims, a backlog of small tickets, the four phone calls you have been not making. Nothing has to be held in your head between items, so being cut in half costs you nothing.

If you use it, use the actual technique rather than the timer. Cirillo's rule is that a pomodoro is indivisible: interrupted, you either write the interruption down and postpone it, or you abandon the pomodoro and it does not count. Most people keep the twenty-five and throw away that rule, which is like keeping the fence and removing the posts. Five to ten minutes off between blocks, and a longer break of twenty to thirty after four of them.

The failure mode: point twenty-five minutes at writing, code, design or analysis and you will feel busy for a fortnight and ship nothing. That is the feeling that brought most people to this page.

The ninety minute block is for work you have to hold in your head

If the work has a shape you have to rebuild each time you sit down, the block has to be long enough to pay for the rebuild and then use it. In practice that means sixty minutes at the floor and ninety as the sensible default.

You will have seen ninety minutes justified by ultradian rhythms, the idea that the brain runs a ninety minute cycle all day long. Be careful with that one. Nathaniel Kleitman's basic rest-activity cycle is well documented in sleep, in the rhythm of REM and non-REM. His further proposal that the same cycle continues through waking hours has been argued over since the 1970s and there is still no convincing evidence for it in waking. It is a borrowed number wearing a lab coat.

The reason to pick ninety is duller and better. If your run-up is fifteen to twenty minutes, ninety is four or five run-ups, which satisfies the rule with room over. And ninety fits inside a working day next to other people's meetings, which matters more than any rhythm, because a block that cannot be scheduled does not happen.

Take ten to fifteen minutes afterwards, and take it away from the screen. A break spent in the same rectangle is not a break. It is a second task, and it comes with its own bill.

The third path is not a length at all, it is a rule for stopping

Some work should not end on a bell, and forcing it to is the reason the bell feels cruel. The alternative is to end on a condition instead of a clock. Until the draft has an ending. Until the test passes. Until every slide exists, including the ugly ones.

This is the strongest of the three when the run-up is very expensive, and it is the only one that never cuts you off mid-thought. It needs two guards or it turns on you.

1. Write the condition down before you start. A condition invented while you are working will quietly grow. On paper, before the first keystroke, it stays the size it was.
2. Give it an outer bound. A hard stop, a person expecting you at seven, a train. The condition decides when you may stop. The bound decides when you must.

Who it suits: people whose problem is stopping too early. Who it ruins: people whose problem is not stopping. If you already lose evenings this way, take the ninety and the bell.

Pick yours off your own number

Any number given to the minute is somebody's average

People go looking for the best pomodoro length as though it were a fact waiting to be looked up, and the internet obliges. The most quoted answer is fifty-two minutes on and seventeen off. It comes from DeskTime, a time-tracking company, which in 2014 looked at the top ten per cent of its own users and reported that average. It was never a study. It was one product's read of its own database, and it went round the world anyway.

The interesting part is what happened next. The same company, measuring the same way, published 112 minutes on and 26 off for 2021. The number more than doubled in seven years without anybody's brain changing.

Which tells you that it was never a property of human attention. It was a property of who was being tracked, doing what, in which year. Stop shopping for that number. Measure the one that is actually yours.

The last minute of a block is worth more than the first twenty

Sophie Leroy's 2009 paper in Organizational Behavior and Human Decision Processes, "Why is it so hard to do my work? The challenge of attention residue when switching between work tasks", put a name to something you already feel. When people move from one task to the next, part of their attention stays behind on the first, and the effect is worse when that first task was unfinished and under time pressure.

An unfinished block is exactly that. Which makes the last sixty seconds the most valuable in it.

Spend them writing down the next move. Not a summary of what you did. The literal next thing: next, rewrite the second paragraph so it stops repeating the first. It does two jobs at once. It gives the unfinished thing somewhere to sit so it stops following you into dinner, and it hands tomorrow a running start.

That second job is the one worth noticing, because the run-up is not a fixed property of you. It is mostly the cost of finding your place, and finding your place is cheap when somebody left a note. Shrink the run-up and the rule above lets you use a shorter block. The way you end a block sets the length of the next one.

The block you can hold is capped by what you let through the door

Here is where the length argument usually stops, and it should not, because none of the above survives contact with an ordinary desk.

Take a fifteen minute run-up and a ninety minute block, interrupted three times. Count a re-entry as half a run-up, which is generous. That is fifteen at the start plus seven or eight for each return: about thirty-seven minutes of your ninety spent getting in. You did roughly fifty-three minutes of work, in four separate pieces, each one shallower than the last.

Now notice the cruelty in it. You chose a long block precisely because your run-up is expensive. Which means the longer the block you pick, the more each interruption costs you. Block length is not a preference. It is a claim about how well the door is shut, and choosing ninety minutes with the phone face-up beside the keyboard is choosing twenty-five minutes with extra steps.

iOS gives you two real tools here and both have the same hole in them. A Focus silences notifications properly, and it is worth turning on. It can also be switched off from Control Center with a swipe and a tap, by you, mid-block, without typing anything. Screen Time's Downtime is sturdier, but it runs on a clock and a set of weekdays. It has no idea that you just sat down.

That something can be a person. Work in the same room as somebody else and the block holds, reliably, which is why libraries still exist and why body doubling keeps being rediscovered every few years under a new name.

Or it can be something that behaves like one, which is the part where we admit we make an app about it. In Scrappy, a block is a walk: a named session with a length, and a small dog standing in front of the apps for the whole of it. The lengths go 25 minutes, 30, 45, 60, 90, then two hours, four hours and eight. Eight rungs rather than one number, because the argument above is that the length belongs to the work. The one called Deep work is two hours. The one called Sleep on it is eight, and it is for the evening you already know how it ends. The three longest are on the paid tier; the first five are not.

It runs a Live Activity on the lock screen, so the block is a visible thing rather than a private intention. Getting past him during one costs a treat, and treats come from walking, roughly two thousand steps for one, and buy five minutes. They cannot be bought with money and never will be. Per app, you can set how hard he is: negotiable takes the treat, firm makes you sit for ten seconds first, and absolute means he will not take the treat at all.

One honest gap, because this is a desk problem and the desk usually has two screens on it. The Mac app is not out yet. It is coming, it will do this on the laptop, and until it ships the second screen is on you.