Pomodoro, Flowtime and 52/17: three systems, one honest comparison

If you only wanted the tool, the timer runs any of these intervals and you can stop reading here.

What each one actually is

The Pomodoro technique. Twenty-five minutes of work, five minutes off, four times, then a longer break of fifteen to thirty minutes. A “pomodoro” is one indivisible block: if you break it, it does not count and you start again.

Flowtime. No fixed block. You pick one task, note the time you start, and work until you are tired or your attention breaks. Then you note the time and take a break scaled to how long you worked. The version that circulates most widely goes: up to 25 minutes of work earns a 5-minute break, 25 to 50 earns 8 minutes, 50 to 90 earns 10, and over 90 earns 15. It was created by Zoë Read-Bivens, specifically as a response to the way fixed intervals cut into a flow state.

The 52/17 rule. Fifty-two minutes of work, seventeen minutes of break. Presented everywhere as a scientific finding about the ideal work rhythm.

Where each one came from, which matters more than it should

Francesco Cirillo invented the Pomodoro technique in the late 1980s as a student, using a tomato-shaped kitchen timer he had in his kitchen. That is the entire origin story. He picked 25 minutes because it was a length he could face. No experiment produced it, no data selected it, and there is no reason to believe it is optimal for anyone in particular. There are only about four peer-reviewed studies of the technique in existence, and none of them derived the number.

This is fine. A useful default does not have to be an optimum. It only becomes a problem when people treat 25 minutes as a fact about human attention rather than a fact about one Italian student’s kitchen.

Flowtime is a blog-post method with no research behind it either, and it is refreshingly upfront about that. It is a design argument, not a finding: rigid blocks interrupt flow, so stop using rigid blocks.

And then there is 52/17.

The 52/17 rule is not research, and its own authors have changed the number twice

This one deserves a section, because it is repeated as science in hundreds of articles and it is not science.

The number comes from DeskTime, a time-tracking software company. In 2014 they looked at their own user database, took the top 10% of users by their own productivity metric, and reported that those people worked about 52 minutes and broke for about 17. That is the whole thing.

What is missing from that:

• No control condition. Nobody was assigned to work 52/17 and compared against anyone working a different ratio. This is a description of what a group of people happened to do, not a test of whether doing it causes anything.

• A circular definition of productive. “Most productive” meant highest score on DeskTime’s own measure, which is time spent in applications the software classifies as productive. That measures app usage, not work quality.

• A self-selected sample. These are people who chose to install employee time-tracking software and keep it running.

• No peer review and no publication. It is a company blog post that got picked up by the press.

Now the part that should settle it. DeskTime re-ran the analysis and the ratio moved. In a 2021 update the company reported that its most productive users now worked, on average, 112 minutes followed by a 26-minute break, and noted that just before the pandemic the ratio had been 80/17.

So the number their own data produces has been 52/17, then 80/17, then 112/26, depending on which year you ask. That is not what a law of human attention looks like. That is what a shifting user base looks like. If you are going to quote 52/17 as the scientifically ideal work rhythm, you have to explain why the science changed its mind twice in seven years without running a single experiment.

Use 52/17 if you like the length. Do not cite it.

What happened when someone actually tested them

There is one study that put these approaches against each other properly, and it is recent.

Smits and colleagues, 2025, Behavioral Sciences. A randomised controlled trial with 94 students working a two-hour session. Three conditions: Pomodoro at 25/5, Flowtime, and self-regulated breaks where people simply took a break when they wanted one.

The results:

• No significant difference in productivity, task completion or flow between the three.

• The Pomodoro group got tired faster than the self-regulated group, at p=0.025.

• The Pomodoro group lost motivation faster as well.

A separate study a couple of years earlier found a complementary result. Biwer and colleagues, writing in the British Journal of Educational Psychology in 2023, put 87 students through systematic breaks against self-chosen ones. Systematic breaks produced mood benefits, and the students taking self-regulated breaks ended up more fatigued and more distracted. But again: no significant difference in mental effort or in how much got completed.

Read those two together and you get something more useful than either alone. Fixed intervals and self-chosen intervals have opposite failure modes, and neither one wins on output.

• Fixed intervals fail by interrupting you. You get tired faster and lose motivation faster, because the schedule keeps cutting across work that was going fine.

• Self-chosen intervals fail by not interrupting you. You get more distracted and more fatigued, because the decision to stop is being made by the part of you that is currently absorbed and does not want to.

Nobody in this argument is winning on how much work gets done. They are trading one problem for another.

So which should you use

Pick by the shape of the task, not by allegiance to a system.

Use short fixed blocks, 25/5, when:

• Starting is the hard part and you need a finish line you can see from here.

• The work is naturally chunky. Email, admin, flashcards, reading, reviewing pull requests, invoicing.

• You are on hour six and you no longer trust yourself to notice you have stopped working.

Use long fixed blocks, 50/10 or 90/20, when:

• The work needs context loaded before anything moves. Debugging, writing once you have the argument, problem sets, anything with several threads held at once.

• You are already going well and the only thing that could ruin it is a bell.

Use Flowtime when:

• Your work is genuinely unpredictable in length and any fixed number would be wrong.

• You already know from experience that you notice when your attention goes, and you will act on it.

Use 52/17 when: you like a block of just under an hour. It is a perfectly good interval. It is just not a discovery.

There is a small piece of evidence worth adding here. The 2022 PLOS ONE meta-analysis of micro-breaks found that longer breaks helped performance more than shorter ones did. That is a mild argument for the 10-to-17-minute end of the range over the five-minute end, if you have the choice.

The variable everyone leaves out

All three systems are arguments about the number on the timer. None of them says anything about what you do when it goes off, which is strange, because that is the part with actual evidence behind it.

Here is what is known about the break itself.

Short breaks reliably restore energy and reduce fatigue. The 2022 meta-analysis, covering 22 samples and 2,335 people, found micro-breaks under ten minutes significantly raised vigour and lowered fatigue. Movement makes that stronger: a meta-analysis of 158 studies covering 13,101 people found acute aerobic exercise raises energy and vigour at d=0.47 and the effect lasts at least thirty minutes, with low intensity working consistently.

What breaks do not do is make you think better. The same 2022 analysis found the effect of micro-breaks on performance was not statistically significant, and on cognitively demanding tasks it was effectively zero. Three separate trials of movement breaks measured cognition and found nothing. The honest framing, and the one the 2026 study of 11,484 desk workers supports, is that breaks restore energy and do not cost you output. That is a weaker claim than the one on most productivity sites. It is also the true one.

There is a second thing breaks are good for, and it has nothing to do with productivity. A 2018 analysis of 34 cohort studies covering 1.33 million people found all-cause mortality risk starts rising above eight hours of sitting a day. A small Columbia study in 2023 found every break dose it tested lowered systolic blood pressure by 4 to 5 mmHg and reduced fatigue. Whichever ratio you pick, the break is where the sitting stops, and that is worth more than the difference between 25 and 52.

This is what exercise snacks are for. Short, deliberate movement folded into time you were going to spend anyway, rather than a workout you have to schedule. A 2025 meta-analysis in the British Journal of Sports Medicine found that pattern improved cardiorespiratory fitness with 91% adherence, which is a higher stick rate than almost any gym membership. It found no effect on body composition, blood pressure or blood lipids, so it is not a body-composition strategy. It is a fitness-and-consistency strategy.

What we built, and what we did not

PomoFit takes no position on the 25-versus-52 argument. Set whatever interval you want, including a Flowtime-style manual stop. What it does have an opinion about is the break: when the timer goes off, you get one short guided bodyweight exercise on screen, so the five minutes has something in it other than a phone.

You can skip every exercise and use it as a plain timer. Plenty of people do. If you would rather not choose the movement each time, the ready-made plans sequence it for you, and Weekly Desk Warrior or Full Body Strength are the two to start with depending on whether you want variety or progression.

What we will not tell you is that our breaks make you focus better. We have read the papers. They do not say that, and we wrote a whole post about the evidence rather than quietly overclaiming. If you want the practical version of the same argument, what to do during a Pomodoro break is the list.

Pick a number. Any of these three will do. Then stand up when it goes off, which is the part that was actually tested.