TL;DR. Interstitial journaling means writing a timestamp and two or three lines every time you switch tasks: where you left the last thing, what you are starting, and how you feel. It helps because switching away from an unfinished task leaves your attention stuck on it, and writing down where you left off gives that residue somewhere to go. The same page doubles as a time log and as a parking lot for distractions. It breaks on meeting-stacked days, and you should never interrupt deep flow to write in it.
There is a specific twenty minutes I used to lose several times a week.
A message arrives while I am inside a problem. I answer it in thirty seconds, feeling efficient. Then I go back to the code and read the same four lines over and over, unable to reload what I was holding in my head, and the thirty second interruption charges me twenty minutes of interest.
The fix that worked for me is embarrassingly small. Before I answer, I write one line: the time, where I was, and what I was about to do next.
That is interstitial journaling. That is the entire method.
What it is
Every time you change tasks, you write a timestamp and two or three lines in one continuously open document.
You note what you just stopped doing and where exactly you left it. You note what you are starting and what finishing it would look like. Optionally you note your state: tired, wired, annoyed, second coffee not working.
The name comes from the interstitial spaces, the gaps between tasks, which is where the writing happens. The practice was popularised by Tony Stubblebine, who framed it as journaling and time tracking collapsed into one habit. The idea of a timestamped working log is much older than that. Lab notebooks and legal time sheets have worked this way for a century. What the modern version adds is the two lines about how it went and what comes next, which is the part that actually does the work.
Why the switch is the expensive moment
Sophie Leroy, a researcher at the University of Washington, named the mechanism attention residue. Her work found that when people switch from one task to another with the first still unfinished, part of their attention stays behind on it, and their performance on the next task suffers while that residue lingers. The unfinished task keeps running in the background whether you consented to it or not.
A later strand of that research is the useful one here: when people are given the chance to make a concrete plan for the unfinished task before switching, the residue effect is reduced. Not the task finished. Just a plan for finishing it.
I want to be careful about what that does and does not license. Leroy studied attention residue. She did not study interstitial journaling, and nobody has. Writing "left off at the timezone conversion, next step is to check the check-date cast" is a plausible way to produce the same ready-to-resume plan, and that is the honest claim. Anyone quoting you a percentage improvement from this practice made the number up.
What I can say from my own use is narrower and less impressive: on fragmented days it removes the reload cost, and on clean days it does nothing, because on clean days there was no residue to manage.
The format, and a page from a real day
Timestamp, one line back, one line forward, optional state.
08:40 Starting: sync bug from the Friday report. Guess is the
timezone cast on the check date. Done = a failing test
that reproduces it.
09:25 Reproduced it. Left off: failing test written, no fix.
Next: client call. Slightly annoyed to be stopping here.
10:10 Call done. Owe them a scope note by Thursday, wrote it
on the parking lot. Back to the bug, resuming at the
failing test.
10:55 Fixed. Left off: fix works, needs a second test for the
DST edge. Next: 20 min of email, then lunch.
13:30 Email took 50 min, not 20. Note that.
Next: the DST test, then the blog draft.
15:05 DST test green. Tired, dropping to shallow work.
Next: invoices and the two admin things.
21:15 Evening block. Loggd. Starting the seeder for the
reflection post. Done = draft in the file, not polished.
Seven entries, maybe two minutes of typing across a whole day. Read it back and three things fall out of it that I did not deliberately record.
First, I know exactly where every task was left, which is the whole point. Second, I have an accidental time log. Third, the 13:30 entry contains a real estimation error I would otherwise have forgotten by Thursday.
It is also a time log you did not have to keep
Nobody keeps a time log for long. It is dull, it demands accuracy you do not have, and its output is a spreadsheet you never read. An interstitial journal produces roughly the same data as a side effect of doing something you had a reason to do anyway.
The resolution is rough. It records the shape of the day, not exact minutes, and it will always miss the switches you made without noticing. That is fine, because the shape is usually the surprising part. My 20-minute email block being 50 minutes is a fact I have now written down four times without ever fixing.
If you want the sharper version of this, run a deliberate week: the time audit guide covers the method and what to do with the result. And if you want to price the switching itself, the context switching calculator has two modes. The calculator takes your interruptions per day and your annual salary and turns them into hours and money lost. The counter is the one I would actually use: you tap a button when you get interrupted, choose from seven types like Slack, email, coworker tap or self-distraction, and it timestamps each one and charts your last seven days.
Read the money figure as an illustration, not a measurement. It multiplies your interruption count by a flat recovery estimate drawn from Gloria Mark's interruption research, which is an average applied bluntly to a day where some interruptions cost you nothing and one cost you the afternoon. The counter tab is the honest half of that tool, because it counts things that actually happened.
It is also a parking lot
The second job the page does is catching the things you refuse to chase.
Halfway through a task you remember the insurance renewal. Under the old system you either handle it now, which is a switch, or you hold it in working memory, which is a tax you pay for the next hour. Under this one you write it on the page and it stops asking.
The rule that makes this work: the parking lot is write-only during the working block. You do not sort it, you do not act on it, you do not open a browser tab to check one small thing. You write the line and you go back.
Then you clear it later. If the list has grown past a dozen items, or if it has started collecting worries rather than tasks, empty it somewhere with more room. The brain dump tool is built for exactly this: a free-form page, an optional 5, 10 or 15 minute timer if you want a bounded session, and then a Categorize pass where you tag each line yourself as a task, an idea, a worry or a reminder and copy the result out as a checklist. The tagging is manual, which sounds like a flaw and is not. Deciding which of those four buckets a line belongs to is most of the value.
Where it breaks
Meeting-stacked days. On a day of six meetings the log degenerates into a list of meeting names with nothing between them. There are no interstitial spaces because the whole day is interstitial. On those days I do not bother, and if that is most of your days, the problem is the calendar rather than the journal. What meetings really cost is the more useful read.
Deep flow. Never interrupt real flow to journal about flow. If you are two hours into something and it is going well, the correct entry is no entry. Write one at the end, backdated, guessing at the time. The method exists to protect the good hours, not to instrument them.
Days you are already avoiding the work. On a bad day the log becomes a very tidy way to feel busy. Seven honest entries about not starting the hard thing is procrastination with excellent documentation.
What I actually do
Not every day. I use it on fragmented days, on days where I am switching between contract work and my own project, and on days where I already know I am going to be interrupted. Maybe two or three days a week.
It also pairs badly with nothing else, which is the reason I keep coming back to it. The end of the log is most of a done list already written, because the "left off" lines are a factual record of what happened. I copy the three that mattered and close the file.
A timestamp and two lines. Twenty seconds. The switch was going to cost you anyway, and this is the cheapest way I know to stop paying interest on it.
Last updated: November 2026.
Written by Eusebiu, the solo founder building Loggd in public. My days split between contract work and this project, with an evening block that starts after a small daughter finally goes to sleep, so I switch contexts more than I would like. I share what the aggregate data keeps showing on Threads.
See the shape of your week without keeping a spreadsheet. Loggd turns what you actually did into a forgiving contribution grid, so a fragmented Tuesday still counts for something. Start free.
