Deep work

Deep work describes a way of working more than a kind of job. You take one task that needs all of your thinking, you close everything else, and you stay with it long enough to get somewhere. Writing a report, working through a proof, designing a system, studying a hard chapter: all of them can be done deeply or in scraps.

The name belongs to Cal Newport, who published Deep Work in 2016. His argument, put in our own words, has two halves. The ability to concentrate for long stretches is worth more than it used to be, because hard things are learned and made that way. And the same ability is getting rarer, because so much of a modern workday pulls attention apart. The first part of his book makes that case, and the second part is a set of rules for doing it.

Everything else at the desk gets a plainer name on this page: shallow work. Email, chat, scheduling, status updates and most meetings belong here. Shallow work is not worthless, and many jobs are built from it. The trouble is that a day full of it feels like a day of work and leaves the hard task where it was.

One of the methods Newport recommends is planning the day in blocks, so the deep hours have a place before the day fills up. That method has its own page here: time blocking. This page stays with two questions the method leaves open. How many deep hours can a person really do, and how would you know how many you did?

Interruptions are the normal state of office work, and one field study measured how normal. Gloria Mark and two colleagues followed 24 information workers, meaning people whose jobs are done mostly at a desk and a screen, through their days. An observer noted to the second what each one was working on (Mark, Gonzalez and Harris, 2005). People stayed with one piece of work for about 11 minutes before switching, and 57% of those stretches were interrupted.

Most interrupted work did get picked up again: 77.2% of it on the same day. The return was slow, though. On average it came 25 minutes and 26 seconds later, after the person had worked on 2.26 other things in between.

Switching also leaves something behind. Sophie Leroy gave the leftover a name, attention residue, in a paper built on 2 experiments (Leroy, 2009). People who moved to a new task while the first one was unfinished kept thinking about the first, and did worse on the second. Finishing helped only when people had finished under some time pressure, which seemed to let them close the first task in their heads.

The cost does not always show up on the clock, which is why it is easy to miss. In a lab study, Mark and colleagues had 48 people answer batches of email, once in peace and once while a phone call or a message cut in every 2 minutes (Mark, Gudith and Klocke, 2008). The interrupted batches were finished sooner: 20.31 minutes against 22.77, with no more errors. People made up the time by working faster and writing shorter replies.

What rose was the strain. On a scale that runs to 20, rated stress went from 6.92 to 9.46, and frustration, time pressure and effort rose with it, after only 20 minutes. An email task is shallow work, so the study says little about a hard problem. It does show that people can hide an interrupted day behind a normal output, and pay for it somewhere else.

“More hours at the desk means more deep work.”

The research above points the other way. The strongest violinists did not practice in longer sessions, and the authors raise the possibility of a general limit of around 4 hours a day for the most demanding work. If that holds for desk work too, a longer day past that point adds desk hours, not deep ones.

Output data from a very different kind of work has the same shape. Pencavel (2014) studied British munition workers in the First World War and found that above 49 hours a week, each extra hour added less. Output at 70 hours differed little from output at 56. Those were factory jobs a century ago, so the exact numbers do not carry over. The bend in the curve is the part worth keeping.

“It takes twenty-three minutes to refocus after an interruption.”

The figure travels widely and is usually tied to Gloria Mark’s research. Neither of her two papers cited on this page contains it. The field study reports 25 minutes and 26 seconds, and that number measures something else: how long it took people to come back to interrupted work, with other work filling the gap. Nobody timed how long concentration took to return.

The correction matters for what you do next. If the cost were a fixed number of minutes per interruption, the fix would be to recover faster. What the studies describe is work that gets set aside for other work and a mind that stays partly on the last thing. The fix for that is fewer switches, which means a block of time with a start, an end and one task in it.

More deep hours rarely come from a longer day. They come from deciding where the blocks go and keeping other work out of them, and each of the five steps below leans on a finding from earlier on this page.

1. Put the deep block on the day first. Decide the task, the start and the end before meetings and email claim the hours. The page on time blocking shows how to lay out a whole day this way.
2. Keep a block to about the length of a practice session. The violinists’ sessions averaged 80 minutes, with rest between them. Two blocks of that length come to more deep work than most office days hold.
3. Close the last task before you open the next. Leroy’s experiments found the leftover attention was largest when the earlier task was unfinished. Reach a stopping point, write one line on where you stopped, and then start.
4. Give shallow work its own blocks. Email answered in two or three sittings a day interrupts far less than email answered as it arrives. The messages still get answered, at times you chose.
5. Stop on time. A longer day rarely adds deep hours, and a short night takes them from tomorrow. The page on productivity tips covers stopping, breaks and planning, each with its evidence.

A full day at the desk can hold almost no deep work, and from memory the two are hard to tell apart.

A good way to run this test counts your hours from what you already write or say, so “two hours on the report, door shut” becomes two hours of focused work. No timer to start and then forget. Most journaling apps stop at how the day felt. A day can feel busy from end to end and hold no deep block at all.

One app built for this is Think in Numbers, a smart journaling app with time tracking built in.

You write or say your day in your own words, as it happens or all at once. The app reads what you wrote and counts the hours that went to each part of it: focused work, meetings, email, your phone, and the time you never mentioned (shown as unaccounted). Then it shows those hours over days and weeks. A target for your focused hours holds each week against your real ones.

Write my first day

Download on the App StoreGet it on Google Play

Or scan with your phone's camera.

For iPhone and Android. Try it on your own day before you pay.

Set the weekly number low at first. If your counter above showed two hours on a good day, ten hours a week is an honest target, and twenty is not. The test runs for several weeks because a single quiet week proves little. What you want to see is the number holding when a deadline or a week of meetings arrives.

The example person below is made up: an office professional who began the month working close to sixty hours a week. Nothing in their days was labeled deep or shallow. They wrote down what they did, and the long blocks of work alone can be read straight out of those words.

Week one came to 58h 30m of work. On the Friday of that week, the slides for a Monday meeting were still not done, after weekdays of meetings and inbox. The one long block for them landed on Saturday morning. A week that long still had no room for its hardest task.

From week two the person set two rules: an end to the workday, and no work email after 8:00 PM. The table lists the long blocks of work alone that their words mention, before and after the rules.

Under the rules, the next long block came on a Monday afternoon, inside working hours, and the deck was finished that Friday at the office.

Week four came to 45h 15m, which is 13h 15m less than week one. The deck was finished and the proposal was sent inside those four weeks. Fewer desk hours did not mean less of the hard work. The hard work moved out of the weekend and the late evening and into blocks during the day.

The log’s own summary of week four is below. The line to read is the longest unbroken stretch: a full afternoon of work with nothing cutting into it, on an ordinary weekday.

Every figure on this page links to one of these sources. Each was read for this page, in full or in its abstract.

1. Newport, C. (2016). Deep Work: Rules for Focused Success in a Distracted World. Grand Central Publishing.
2. Ericsson, K. A., Krampe, R. T. and Tesch-Römer, C. (1993). The role of deliberate practice in the acquisition of expert performance. Psychological Review, 100(3), 363 to 406.
3. Mark, G., Gonzalez, V. M. and Harris, J. (2005). No task left behind? Examining the nature of fragmented work. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI 2005), 321 to 330.
4. Leroy, S. (2009). Why is it so hard to do my work? The challenge of attention residue when switching between work tasks. Organizational Behavior and Human Decision Processes, 109(2), 168 to 181.
5. Mark, G., Gudith, D. and Klocke, U. (2008). The cost of interrupted work: More speed and stress. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI 2008), 107 to 110.
6. Pencavel, J. (2014). The Productivity of Working Hours. IZA Discussion Paper No. 8129, Institute of Labor Economics.