There is a version of this advice that stops after the first paragraph: turn off notifications, install a blocker, put your phone away. It is correct, it is insufficient, and its insufficiency is the reason people conclude they are the problem.
Distraction enters through four doors. The first is the one everybody writes about. It is also the easiest and the least important.
| Door | What comes through | What closes it |
|---|---|---|
| The device | Notifications, tabs, the phone | Mechanical. One hour of setup, once |
| The room | People, noise, sightlines | Physical. Position, signal, timing |
| The task | Ambiguity about what you are doing | Written. One sentence, before you start |
| The self | The reach for relief when it gets hard | Practised. Months, not minutes |
Close them in that order. Not because the first is most important, but because the first two are finishable, and finishing them buys you the credibility to attempt the last two.
Door one: the device
This is a checklist, and it is boring on purpose. Do it once, properly.
Notifications off at the operating-system level, not per app. Per-app settings decay — every update re-enables something. One system-level focus mode that permits calls from three people and nothing else survives updates.
The phone in another room. Not face-down. Not in a drawer. Another room. A face-down phone still leaves should I check it available to you, and that question costs something every time it is asked, which is more often than you think.
One browser profile for work, signed into nothing. No mail, no chat, no social. The friction of signing in is the entire mechanism, and it works because you will not bother.
Close the tabs. All of them, including the eleven you are keeping for later. Open tabs are a queue of unfinished intentions in your peripheral vision, and peripheral vision is not neutral.
Turn off the badge counts. A red dot is a notification that never expires.
That is the door. It takes about an hour and it stays closed. Anyone who tells you this is the hard part is selling something that closes this door.
Door two: the room
Less controllable, and worth what control you have.
Sightlines. Facing a door or a corridor means every passing person is a micro-interruption you did not consent to. Turning the desk is free.
A visible signal. Headphones, a closed door, a light — anything that means not now without requiring a conversation. The signal has to be legible to your particular colleagues, so ask rather than assume.
Choose the hour other people want less. This is the largest lever in the whole section and it costs nothing. The hour before most people start, or the hour most people are at lunch, is worth two hours of defended mid-afternoon.
Water and a notebook within reach. Not for productivity reasons. Because standing up to get something at minute twenty is how a session ends.
Door three: the task
Here the advice thins out and the failures begin.
An ambiguous task is a distraction with no notification. You sit down to work on the proposal and discover, thirty minutes later, that you have been deciding what work on the proposal means. That deciding is real work, but it is judgement work and it does not belong in a session you meant to spend making something.
The rule: before the session starts, write one sentence describing what finished looks like.
Not the task. The finish line.
- Not work on the proposal. A draft of the strategy section that I would be willing to send.
- Not review the design. A written list of the three things I think are wrong with it.
- Not think about pricing. A one-page argument for one number.
Write it where you will see it. When you drift — and you will — the sentence is what you come back to, and coming back to a specific thing is much faster than coming back to a topic.
If you cannot write the sentence, that is the finding. The task is not ready for a session; it needs five minutes of deciding first, and doing that deciding inside a protected hour is how protected hours get wasted.
Door four: the self
This is the one that matters, and there is no setting for it.
Somewhere between fifteen and forty minutes into genuinely demanding work, there is a moment where the problem stops being pleasant. You reach the part you do not know how to do. And precisely then, checking something feels not lazy but reasonable — a small break, a quick look, you have earned it.
That reach is where sessions die. Not at the beginning, where the blockers are.
Two things are worth knowing about it.
The first is that a substantial share of workplace interruptions are self-initiated. This is an observed finding from Gloria Mark’s group’s observational work with knowledge workers: removing external interruptions reveals how many of the remaining ones are your own. You are not merely being interrupted. You are interrupting yourself, and no amount of device hygiene touches that.
The second is that the reach is not cheap. Sophie Leroy’s studies of attention residue found that when attention has not fully disengaged from one task, performance on the next suffers. Applied to a session, the two-minute check does not cost two minutes; it costs the two minutes plus a diminished return to a problem you had just got difficult enough to be worth solving.
What we are inferring. Leroy's studies measured switching between separate tasks, not the specific case of glancing at a phone mid-session and returning. Applying the finding to that case is inferred — reasonable, and not demonstrated. What is demonstrated is that switching leaves residue at all, which is enough to take the reach seriously.
What actually helps
Not willpower, which is what everyone reaches for and what runs out.
Name the friction out loud. I do not know how to structure this section. Said plainly, the urge to escape usually drops a level, because most of the urge is about not wanting to admit you are stuck.
Stay seated. Not working, necessarily. Seated. The rule is not keep making progress; it is do not leave the chair. Most of the reach is a bid to leave, and refusing the leaving is easier than forcing the progress.
Write the stuck thing down. One line, in the same place as your finish-line sentence. It converts a vague feeling of failure into a specific problem, which is a thing you can work on.
Expect it every time. People treat the difficult moment as evidence the session is going badly. It is evidence the session reached the work.
Boundaries: a start and a real end
A session without an end is a shallow session that lasts a long time.
Decide the length before you begin, and stop when it is over even if it is going well. Two reasons. The first is that a session with a known end is easier to enter, because the commitment is bounded. The second is that stopping mid-flow leaves you with somewhere obvious to start tomorrow, which makes tomorrow’s entry cheaper.
The comparison of session methods goes into what length to pick and why the answer depends on which door keeps opening.
The ritual
Under two minutes. The same every time. No decisions in it.
- Phone to the other room.
- Write the finish-line sentence.
- Set the length and begin.
That is the whole thing. It is not impressive and it is not supposed to be. A ritual with variety in it is another place to hesitate, and hesitation at the threshold is the failure it exists to prevent.
When it fails
It will fail, regularly, and the failure is not informative on its own. What is informative is which door opened.
If the phone got you, door one is not actually closed. If a colleague got you, door two. If you spent the hour deciding what you were doing, door three. If you started, got somewhere difficult, and left — door four, which is the practice rather than the setup.
A session is not a state of mind. It is four closed doors and a sentence about what finished looks like.
If you close all four and still cannot work, the problem is not the session. It is somewhere else, and it has a name.