TL;DR. Flowtime swaps the Pomodoro bell for a stopwatch. Pick one task, start the watch, work until your attention breaks on its own, then stop and write down the start time, the stop time and how many times you got pulled away. The break is sized from the block: roughly a fifth of it, five minutes minimum. It suits people who resent being cut off mid-thought and whose days do not divide into tidy 25 minute pieces. It fails badly for two groups: people who need the timer to make them start, and people who never stop until they crash. There is no research on Flowtime specifically, and I have not found a head to head trial where fixed intervals beat self-paced ones or the reverse.
The bell goes off. You are two sentences from the end of an argument you have been holding in your head for twenty minutes, and a kitchen timer would like you to go and stretch.
Both options are bad. Ignore it, and you have taught yourself that the timer is a suggestion. Obey it, and you pay to load the whole structure back in afterwards.
That moment is why a lot of people quietly stop using Pomodoro. Not because 25 minutes is a wrong number, but because a fixed interval has no idea what you are in the middle of.
Flowtime is the obvious inversion, written down properly.
What Flowtime actually is
Five steps, and the fifth one is what makes it a method.
- Pick one task and write it down. One. Not a category, not "work on the app".
- Note the time and start a stopwatch.
- Work until your attention breaks on its own.
- Stop the watch. Write the stop time, the minutes worked, and how many times you got pulled away.
- Take a break sized to the block you just did, then start the next row.
It is usually credited to Zoe Read-Bivens, who published it as a Pomodoro alternative for people whose work does not divide neatly into 25 minute pieces.
The difference in one line: Pomodoro decides when you stop and you find out how much you did. Flowtime decides nothing and you find out when you stopped.
Which is why the log is the actual product. Take it away and you are just working with a stopwatch running, which people do while sincerely believing they have adopted a system.
The log
Six columns. Resist adding more.
| Task | Start | Stop | Pulled away | Worked | Break |
|---|---|---|---|---|---|
| Draft this post | 20:05 | 20:41 | 2 | 36 min | 8 min |
| Draft this post | 20:49 | 21:38 | 0 | 49 min | 10 min |
| Fix a seeder bug | 21:48 | 22:03 | 3 | 15 min | 5 min |
| Inbox and invoices | 22:08 | 22:26 | 1 | 18 min | done |
That is a real shape of evening for me. I write code for clients during the day and build this in the evenings around a small daughter, so the working window starts when the house goes quiet and ends when I stop being useful.
Read it back and three things fall out that a Pomodoro count of "four sessions" would have hidden.
Row two is the one that mattered. Forty nine uninterrupted minutes on a task that was already warm. Nothing about my day predicted that block would be the good one, and no fixed interval would have let it run.
Row three is not a focus problem. Fifteen minutes with three interruptions is not a short attention span, it is a bad environment. The fix is the phone, not more discipline.
Row four should probably not have happened. By 22:08 I am doing admin because it is the only thing left in me, which is fine, but it is worth knowing that is what it was.
Sizing the break
The table that circulates with the method:
| Block length | Break |
|---|---|
| Under 25 minutes | 5 minutes |
| 25 to 50 minutes | 8 minutes |
| 50 to 90 minutes | 10 minutes |
| Over 90 minutes | 15 minutes |
Nobody derived those numbers. They land near one part break to five parts work, which is roughly the ratio the Pomodoro tradition already uses, and the arithmetic is discussed in how long a pomodoro should be.
My own version is shorter: a fifth of the block, never under 5 minutes, and past 90 minutes you leave the desk. That last clause is the one I would defend hardest. A break in the same posture reading the same kind of text is a change of subject, not recovery.
Who it suits
People who resent the bell. If your recurring experience of Pomodoro is being interrupted at exactly the wrong moment, that is not a character flaw, that is the method being wrong for your work.
Work with a high warm-up cost. Debugging, writing something that has to hold together, reading in a field you barely know. These need a long runway, and a fixed 25 minutes spends a meaningful slice of every block getting back to where you were.
Skill work, with a caveat. Long uninterrupted blocks suit practice sessions well, as long as the block has a target rather than just a duration. Structuring that part is deliberate practice, and it is what turns a long session into an improving one rather than a comfortable one.
Days you cannot predict. Mine are like this. A block that ends when a child wakes up is a real block, and Flowtime records it as one rather than as a failed pomodoro.
People who already start easily. Load-bearing, and the reverse of it opens the next section.
Where it fails
If the timer is what makes you start. Flowtime has no starting mechanism at all. None. A countdown works partly because it is a small, finite promise: 25 minutes and then I am allowed to stop. Take that away and a person who struggles with initiation is left with an open-ended invitation to begin something hard, which is exactly what they were avoiding. If that is you, the honest advice is to keep a fixed interval and go shorter, not longer, and to pick the task by dread rather than importance, which is the whole argument in eat the frog.
If you never stop on your own. Some people do not have a natural break point. They work until the quality is gone and the evening is gone, then call the whole thing flow. Flowtime hands them the keys. If you recognise yourself, put a ceiling on the block, 90 minutes, and treat the ceiling as an actual rule. The reasons a long block stops paying somewhere in its second hour are covered in the ultradian rhythm post, which also explains why the popular 90 minute claim is shakier than the internet suggests.
If you will not keep the log. Three days is enough to tell. If the rows stop appearing, you have not adopted Flowtime, you have stopped using a timer.
Running it with a stopwatch
You need a clock that counts up and gets out of the way. The aesthetic stopwatch is the one I built for this, and here is exactly what it does, no more.
It counts up with millisecond precision and shows the elapsed time in the browser tab, so a glance at the tab strip tells you where you are without breaking the page you are on. Space starts and pauses. L records a lap, which stores the split since the last lap and the running total, and the fastest and slowest splits get highlighted once you have a few. R resets, F goes fullscreen, M mutes. There are four themes and an optional chime at 1, 5, 10, 30 and 60 minutes, which is worth turning on for Flowtime specifically: it is ambient awareness rather than a bell, since nothing happens when it sounds. The session survives a reload for up to 24 hours and comes back paused rather than restarting itself.
The useful trick is the lap button. Press L every time something pulls you away. You end the block with your interruption count already recorded, and the split lengths show you how long the good stretch actually was before it broke.
Now the limits, because they matter for a method built on a log. Laps are numbered, not named. Nothing is exported and nothing leaves your browser, which is the privacy trade working for you and the convenience trade working against you. So you still write the row yourself, in a notebook or a note or a spreadsheet.
The evidence, honestly
There is no Flowtime research. The method comes from a blog post, it has been passed around by timer apps and a couple of university study-skills pages, and as far as I can find nobody has tested it.
What does exist nearby: work on attention residue showing that part of your attention stays with the previous task after a switch, which argues for fewer and longer blocks. And a broad literature on breaks that says taking them beats not taking them. Both support Flowtime. Both equally support Pomodoro.
The comparison people actually want, fixed intervals against self-paced ones, is the one I cannot find a clean answer for, and the articles claiming a winner do not cite a trial.
Here is the thing I am confident about instead, because it comes from the data I do have. Across 7,800 tracked habits on this site, roughly two in three died at or before the second check-in. Almost nothing about a practice matters as much as whether it happens twice. If Flowtime is the version you keep doing, Flowtime is the better method for you, and that sentence is doing more work than any interval comparison.
Trying it for a week
- Three days of logging with no changes. Do not optimise anything. You are collecting rows.
- Take the breaks properly. This is where it falls apart first, because a self-terminated block ends at a natural moment and skipping straight to the next one feels great for about two days.
- On day four, read the interruption column. One recurring source usually accounts for most of it. Fix that one thing.
- Look at your longest three blocks and note the time of day and the task type. That is the pairing worth protecting.
- At the end of the week, ask whether you started more often or less often than you did with a countdown. That is the whole test.
If your blocks got longer but there were fewer of them, you traded starting for flowing, and that trade is only worth it if the total went up. The log will tell you. It is the one thing in this method with no opinion.
Last updated: November 2026.
Written by Eusebiu, the solo founder building Loggd in public. I write code for clients during the day and build this in the evenings around a small daughter, which is why a method that records what actually happened suits me better than one that decides in advance. I share what the aggregate data keeps showing on Threads.
Keep the rows somewhere you will see them again. Loggd tracks focus sessions and habits as a forgiving contribution grid, so a thin week stays a thin week instead of resetting you to zero. Start free.
