How to improve deep work (what the data says, not just the book)

"Deep work" is a book first and a research topic second. Most pages ranking for the term summarise Cal Newport's 2016 argument and its four scheduling philosophies. This page starts from the data instead: how much uninterrupted time knowledge workers measurably get, what interruptions cost in controlled studies, which of the book's prescriptions the evidence supports, and a weekly plan built from the figures rather than the slogans.

What deep work is and what the book claims

Newport defines deep work as professional activity performed in a state of distraction-free concentration that pushes cognitive capabilities to their limit, and contrasts it with shallow work, the logistical, low-concentration tasks that fill most days ^[1]. The book makes three claims: that deep work is increasingly valuable, that it is increasingly rare, and that it is trainable like a skill. It then offers four ways to schedule it, monastic, bimodal, rhythmic and journalistic, and a set of rules about embracing boredom and quitting social media.

The book is a well-argued essay, not a study, and it does not pretend otherwise. The research literature has something to say about each of its claims, and the verdicts differ.

How much deep time people actually get

Three datasets describe the baseline, each with a different quality.

Source	What it measured	Figure	Grade
RescueTime, 2018, about 50,000 users of a time-tracking app	Daily time in "completely focused work" with no communication-tool use	About 1 h 12 min per day; a commenter notes the chart shows a median where the text says average C[2]	C, vendor data from self-selected users
Microsoft Work Trend Index 2025, 31,000-person survey plus Microsoft 365 telemetry	Interruption frequency	Interrupted about every two minutes, around 275 times across a workday including off-hours activity B[3]	B, vendor research on the general workforce
Atlassian State of Teams 2025, 12,000 knowledge workers	Time spent searching for information	About 25 % of time, roughly ten hours in a forty-hour week by our conversion B[4]	B, vendor survey
Gloria Mark's field logging, UC Irvine, 2004–2020	Average attention to one screen before switching	About 2.5 minutes in 2004, 47 seconds by 2020 B[5]	B, researcher's summary of her field data in an interview

None of these figures is a peer-reviewed measurement of deep work, and they should not be stitched into a single statistic. Together they do support one qualitative claim from the book: uninterrupted concentration is scarce in ordinary knowledge work, and the scarcity is structural rather than a personal failing.

Interruptions: the measured cost

The controlled evidence on interruption is better than the evidence on deep time. In Mark, Gudith and Klocke's laboratory simulation of office tasks, interrupted participants finished faster than uninterrupted ones but reported significantly more stress, frustration, time pressure and effort B^[6]. Interruptions of about three seconds roughly doubled errors in a sequential procedural task and interruptions of about four and a half seconds roughly tripled them B^[7]. Even receiving a notification without checking it impaired a sustained-attention task about as much as actively using the phone B^[8].

In the field, Mark, Gonzalez and Harris observed information workers switching between many "working spheres" in short episodes, with over half of the spheres interrupted before completion C^[9]. How long it takes to recover afterwards is a range, not a number: a few seconds for a well-practised lab task, up to the "almost half an hour" Mark reports from field observation of workers returning to an interrupted project ^[5]. The widely quoted "23 minutes" is from an interview, not a paper. Our page on the cost of context switching lays out the full range with the study context for each bound.

A detail worth keeping: in a five-day logging study of 32 workers at one company, self-reported focus peaked in mid-afternoon and was lowest on Mondays C^[10]. The sample is far too small to generalise, but it is a reminder that "mornings are for deep work" is a preference, not a finding; the better-supported rule is that executive performance peaks at the time that matches your chronotype A^[11].

Four scheduling philosophies compared

Newport's four philosophies are organising devices rather than tested interventions. The table maps each to what the evidence can say about its ingredients.

Philosophy	What it prescribes	What the evidence supports	Practical fit
Monastic	Eliminate shallow obligations almost entirely	Nothing tests this directly; the interruption studies support reducing intrusions [6][7]	Writers, researchers with no coordination role
Bimodal	Alternate multi-day deep stretches with open days	Untested as a package; cutting email for five days halved window switching and lowered stress in a 13-person study C[12]	Project-based roles with control over the calendar
Rhythmic	A fixed daily block at the same time	Best supported by adjacent evidence: implementation intentions ("at 9 I open the draft") have a medium-to-large effect on goal attainment, d = 0.65 across 94 studies A[13]; habits predict behaviour more than inhibition does A[14]	Most people in meeting-heavy jobs
Journalistic	Drop into deep work whenever a gap appears	Weakest fit: switch costs grow with task complexity A[15], and attention residue from the previous task impairs the next B[16]	Experienced practitioners with highly familiar tasks

The rhythmic philosophy wins on evidence for an unglamorous reason: it converts a daily decision into a default. The book presents it as the most practical option; the goal-setting and habit literature explains why.

Environment and tools

• Communication windows. Checking email three times a day instead of freely lowered daily stress in a 124-person crossover trial; output was not measured B^[17].
• Notifications off at device level. The notification study measured the cost of a buzz you do not answer ^[8]; muting one app while the phone still vibrates does not address it.
• Open-plan offices. When two firms moved to open-plan layouts, face-to-face interaction measured by wearable badges fell by about 70 % while email rose B^[18], and across 42,764 occupants of 303 buildings, open-plan workers were less satisfied, with no advantage in ease of interaction B^[19]. If you can choose where to do deep work, do not choose the open floor.
• Music. Across roughly 100 studies, background music had no overall benefit and slightly impaired reading and memory A^[20]; lyrics are worse than instrumental and the harm concentrates in difficult tasks B^[21]. Use it to mask noise, not to improve thinking.
• Breaks. Micro-breaks reliably raise vigour and reduce fatigue, d of about 0.35, but do not reliably improve performance on demanding tasks; longer breaks help performance more A^[22]. The direct Pomodoro evidence is three small student studies: scheduled breaks improved mood and fatigue, and produced no productivity difference compared with self-chosen breaks C^[23][24].
• Sleep. Partial sleep restriction degrades sustained attention and executive function, and the deficit accumulates while you stop noticing it A^[25]. No scheduling philosophy survives a short week.

A weekly deep-work plan built from the data

The plan below uses the measured figures as constraints, not as targets.

1. Set a sleep floor first. Seven hours in bed, fixed wake time. Everything else assumes a rested brain ^[25].
2. Choose two to three blocks of 60–90 minutes, not a four-hour ideal. The book's four-hour ceiling is plausible; the baseline data suggest most people start from around an hour of focused time a day ^[2], so doubling it is a realistic first goal.
3. Place blocks at your chronotype peak, then protect them with an if-then plan. "If it is 9:30, then I close chat and open the draft" is the form with the strongest evidence ^[13].
4. Batch communication into two or three windows. The stress evidence favours it ^[17]; set the windows so colleagues know when to expect replies.
5. Shorten resumption with a parking note. Attention residue falls when a task feels closed ^[16]; one line about where you stopped and what comes next is enough.
6. Take real breaks between blocks. Walk or step away; detachment and physical activity show the most consistent benefits across break studies ^[22].
7. Review the week against your log, not your feeling. Count blocks completed and interruptions that broke them.

Tracking without surveillance

The temptation is to install a tracker that logs every window. The evidence does not require it, and in a team it introduces a trust cost that no study has shown to be repaid. Three counts are enough: blocks planned versus completed, the longest uninterrupted stretch each day, and the source of each interruption that broke a block. Two weeks of those three numbers tells you where your week actually leaks, and they are numbers you can share with a manager without exposing anyone's screen.

What the evidence does not say

• It does not say the brain works in 90-minute ultradian cycles. The claim extrapolates Kleitman's rest–activity hypothesis from sleep into waking performance C^[26]; a direct test with 60 people measured every ten minutes for nine hours found no 90-minute rhythm B^[27]. Choose block lengths by task and calendar, not by physiology.
• It does not say Pomodoro's 25/5 rhythm is optimal. Three small student studies show mood benefits and no productivity advantage ^[23][24].
• It does not say you lose 23 minutes per interruption. The measured range runs from seconds to tens of minutes depending on context ^[5][7].
• It does not say knowledge workers get exactly 72 minutes of focus a day. That is a 2018 vendor statistic from self-selected app users ^[2].
• It does not say deep work is trainable like a muscle. Nothing in the literature tests sustained-concentration training over months in adults at work; what is established is that the state variables, sleep, interruptions and task clarity, move performance now.

What to do next

Build the week before you build the habit. The focus planner takes your meeting hours, wake and bedtimes, energy pattern and interruption sources and returns a day template with deep-work blocks placed where they are most likely to survive, a batched admin window and a shutdown slot. Run it for two weeks, then compare the plan with your three tracking counts.

Related: the science of attention behind these recommendations, what the research says about flow state, and why avoiding a block is usually a procrastination problem of mood rather than time.

Frequently asked questions

What is deep work, in one sentence?

Cal Newport's term for professional work done in distraction-free concentration that pushes your cognitive abilities to their limit, as opposed to shallow work, the logistical tasks that can be done while distracted ^[1]. The research literature does not use the term; it studies sustained attention, interruptions and task switching, which is where the evidence on this page comes from.

How many hours of deep work per day is realistic?

The only large dataset, a 2018 vendor analysis of about 50,000 time-tracking users, put completely focused time at around 1 hour 12 minutes a day ^[2]. It is low-quality evidence, but it suggests two or three protected blocks of 60 to 90 minutes is an ambitious target for most people, and that the book's four-hour ceiling is an upper bound rather than a norm.

Is the Deep Work book worth reading?

As an essay on why concentration matters and how to schedule it, yes; its rhythmic philosophy in particular lines up with the goal-setting and habit evidence ^[13][14]. Read it knowing that its claims about rarity and value are argued rather than measured, and that the interruption figures in popular summaries, including the 23-minute one, are not from peer-reviewed studies.

Which deep work schedule is best?

For most people in meeting-heavy roles, a fixed daily block at the same time. Implementation intentions, if-then plans tied to a time or cue, have a medium-to-large effect on goal attainment across 94 studies ^[13], and habits predict behaviour more strongly than effortful self-control does ^[14]. Dropping into deep work whenever a gap appears is the weakest option because switch costs rise with task complexity ^[15].

Should I work in 90-minute blocks because of ultradian rhythms?

The 90-minute waking rhythm is a hypothesis extended from sleep research ^[26], and a direct test of 60 people over nine hours found no such periodicity in performance or alertness ^[27]. Pick block lengths that fit your task and calendar; 60 to 90 minutes is a practical range, not a biological requirement.

Does the Pomodoro technique improve deep work?

The direct evidence is three small student studies. Scheduled breaks improved mood and reduced fatigue, and there was no difference in productivity compared with taking breaks when you feel like it ^[23][24]. Micro-breaks in general raise vigour but do not reliably improve performance on demanding tasks ^[22]. Use the timer if it helps you start; do not expect it to make the work better.

Sources and evidence

1. Deep Work: Rules for Focused Success in a Distracted World — Newport, C. (2016), book C (retrieved 2026-11-09)
2. The true cost of email and IM: You only have 1 hour and 12 minutes of uninterrupted productive time a day — RescueTime blog (2018), vendor page C blog.rescuetime.com/communication-multitasking/ (retrieved 2026-11-09)
3. 2025 Work Trend Index Annual Report: The year the Frontier Firm is born — Microsoft WorkLab (2025), report B www.microsoft.com/en-us/worklab/work-trend-index/2025-the-year-the-fro (retrieved 2026-11-09)
4. State of Teams 2025 — Atlassian (2025), report B www.atlassian.com/blog/state-of-teams-2025 (retrieved 2026-11-09)
5. How to sharpen your attention and meet your goals (interview with Gloria Mark, UC Irvine) — University of California Newsroom (2024), report B www.universityofcalifornia.edu/news/how-sharpen-your-attention-and-mee (retrieved 2026-11-09)
6. The cost of interrupted work: more speed and stress — Mark, G.; Gudith, D.; Klocke, U. (2008), conference paper B doi.org/10.1145/1357054.1357072 (retrieved 2026-11-09)
7. Momentary interruptions can derail the train of thought — Altmann, E. M.; Trafton, J. G.; Hambrick, D. Z. (2014), RCT B doi.org/10.1037/a0030986 (retrieved 2026-11-09)
8. The attentional cost of receiving a cell phone notification — Stothart, C.; Mitchum, A.; Yehnert, C. (2015), RCT B doi.org/10.1037/xhp0000100 (retrieved 2026-11-09)
9. No task left behind? Examining the nature of fragmented work — Mark, G.; Gonzalez, V. M.; Harris, J. (2005), conference paper C doi.org/10.1145/1054972.1055017 (retrieved 2026-11-09)
10. Bored Mondays and focused afternoons: the rhythm of attention and online activity in the workplace — Mark, G.; Iqbal, S. T.; Czerwinski, M.; Johns, P. (2014), conference paper C doi.org/10.1145/2556288.2557204 (retrieved 2026-11-09)
11. A time to think: Circadian rhythms in human cognition — Schmidt, C.; Collette, F.; Cajochen, C.; Peigneux, P. (2007), review A doi.org/10.1080/02643290701754158 (retrieved 2026-11-09)
12. A pace not dictated by electrons: an empirical study of work without email — Mark, G.; Voida, S.; Cardello, A. (2012), conference paper C doi.org/10.1145/2207676.2207754 (retrieved 2026-11-09)
13. Implementation Intentions and Goal Achievement: A Meta-analysis of Effects and Processes — Gollwitzer, P. M.; Sheeran, P. (2006), meta-analysis A doi.org/10.1016/S0065-2601(06)38002-1 (retrieved 2026-11-09)
14. Taking Stock of Self-Control: A Meta-Analysis of How Trait Self-Control Relates to a Wide Range of Behaviors — de Ridder, D. T. D.; Lensvelt-Mulders, G.; Finkenauer, C.; Stok, F. M.; Baumeister, R. F. (2012), meta-analysis A doi.org/10.1177/1088868311418749 (retrieved 2026-11-09)
15. Executive control of cognitive processes in task switching — Rubinstein, J. S.; Meyer, D. E.; Evans, J. E. (2001), RCT A doi.org/10.1037/0096-1523.27.4.763 (retrieved 2026-11-09)
16. Why is it so hard to do my work? The challenge of attention residue when switching between work tasks — Leroy, S. (2009), RCT B doi.org/10.1016/j.obhdp.2009.04.002 (retrieved 2026-11-09)
17. Checking email less frequently reduces stress — Kushlev, K.; Dunn, E. W. (2015), RCT B doi.org/10.1016/j.chb.2014.11.005 (retrieved 2026-11-09)
18. The impact of the 'open' workspace on human collaboration — Bernstein, E. S.; Turban, S. (2018), observational B doi.org/10.1098/rstb.2017.0239 (retrieved 2026-11-09)
19. Workspace satisfaction: The privacy-communication trade-off in open-plan offices — Kim, J.; de Dear, R. (2013), observational B doi.org/10.1016/j.jenvp.2013.06.007 (retrieved 2026-11-09)
20. The impact of background music on adult listeners: A meta-analysis — Kämpfe, J.; Sedlmeier, P.; Renkewitz, F. (2011), meta-analysis A doi.org/10.1177/0305735610376261 (retrieved 2026-11-09)
21. Background Music and Cognitive Task Performance: A Systematic Review of Task, Music, and Population Impact — Cheah, Y.; Wong, H. K.; Spitzer, M.; Coutinho, E. (2022), systematic review B doi.org/10.1177/20592043221134392 (retrieved 2026-11-09)
22. Give me a break! A systematic review and meta-analysis on the efficacy of micro-breaks for increasing well-being and performance — Albulescu, P.; Macsinga, I.; Rusu, A.; et al. (2022), meta-analysis A doi.org/10.1371/journal.pone.0272460 (retrieved 2026-11-09)
23. Understanding effort regulation: Comparing 'Pomodoro' breaks and self-regulated breaks — Biwer, F.; Wiradhany, W.; oude Egbrink, M. G. A.; de Bruin, A. B. H. (2023), RCT C doi.org/10.1111/bjep.12593 (retrieved 2026-11-09)
24. Investigating the Effectiveness of Self-Regulated, Pomodoro, and Flowtime Break-Taking Techniques Among Students — Smits, E. J. C.; Wenzel, N.; de Bruin, A. (2025), RCT C doi.org/10.3390/bs15070861 (retrieved 2026-11-09)
25. The neurocognitive consequences of sleep restriction: A meta-analytic review — Lowe, C. J.; Safati, A.; Hall, P. A. (2017), meta-analysis A doi.org/10.1016/j.neubiorev.2017.07.010 (retrieved 2026-11-09)
26. Basic Rest-Activity Cycle: 22 Years Later — Kleitman, N. (1982), review C doi.org/10.1093/sleep/5.4.311 (retrieved 2026-11-09)
27. Ultradian rhythms in cognitive performance: no evidence for a 1.5-h rhythm — Neubauer, A. C.; Freudenthaler, H. H. (1995), observational B doi.org/10.1016/0301-0511(95)05121-P (retrieved 2026-11-09)