TL;DR. No serious athlete trains at maximum every week of the year, and the reason is measured. A meta-analysis of 27 tapering studies found that the most effective deload lasts two weeks and cuts training volume by 41-60% while leaving intensity and frequency alone; a companion review puts the typical gain from a taper at about 3%. Knowledge work runs the opposite experiment. WHO and ILO estimate that 488 million people worked 55 hours or more a week in 2016 and that 745,194 deaths from ischaemic heart disease and stroke were attributable to that exposure. The risk climbs with the hours: in a Lancet meta-analysis of 528,908 people, the relative risk of stroke rose from 1.10 at 41-48 hours to 1.27 at 49-54 and 1.33 at 55 or more, against a 35-40 hour week. A CEOtudent calculation on ILOSTAT data shows how unevenly the permanent sprint is distributed: 48.8% of India’s employed worked 49 hours or more a week (hours actually worked), 25.0% in Türkiye, 11.3% in the United States and 4.2% in Germany. Recovery is real but perishable: vacation improved health and well-being by d = +0.43 across seven studies, and in the four studies that measured fade-out the gain shrank by d = -0.38 between two post-vacation measurements roughly two to four weeks apart. The answer is to run three deliberate modes, Sprint, Maintain and Recover, each with an entry signal, an exit criterion and its own metrics. CEO move: write the end date of every sprint before it starts and book the recovery block in the same calendar entry. Student move: log realized hours and next-morning energy for one quarter and find out which mode the body was actually in.
This article belongs to the Personal Operating System cluster. It sits above the recovery protocol, which sets rest at the hour, day, week and season, and alongside the 40-year career, which prices the sprint over a working life. Those pieces answer “how to recover” and “why it matters”. This one answers “when to push, when to hold, and when to deload”.
Why do athletes periodize while careers run flat out?
Sports science has a word for alternating load and deload on purpose: periodization. V. B. Issurin’s 2010 review in Sports Medicine dates the theory of training to “about five decades ago”, when traditional periodization was proposed as “a division of the entire seasonal programme into smaller periods and training units”. Among the reasons the traditional model ran into trouble, he lists one that reads like a description of office life: “excessive fatigue elicited by prolonged periods of multi-targeted training”. The response, block periodization, sequences blocks of “highly concentrated workloads directed to a minimal number of targeted abilities” instead of improving everything at once.
The joint consensus statement of the European College of Sport Science and the American College of Sports Medicine states the governing rule in one sentence: “Successful training not only must involve overload but also must avoid the combination of excessive overload plus inadequate recovery.” The terminology of the earlier 2006 European College of Sport Science position statement, reproduced in Kreher and Schwartz’s open-access guide, is the most useful vocabulary a knowledge worker can borrow. Functional overreaching is increased training that produces a temporary performance decrement “with improved performance after rest”, lasting “days to weeks”. Nonfunctional overreaching produces a longer decrement “but with full recovery after rest”, with psychological or neuroendocrine symptoms, lasting “weeks to months”. Overtraining syndrome adds a “longer performance decrement (> 2 months)” and more severe symptoms.
Read as a career model, these are three things people call “a busy period”. A Sprint is deliberate functional overreaching: harder than sustainable, bounded, followed by rest, and expected to leave you better than before. A busy period with no end date drifts into the nonfunctional kind. The deload itself has been measured in detail.
Table 1. What sports science has measured about deloading (figures as printed)
| Finding | Value | Source |
|---|---|---|
| Studies included | 27 of 182 potential studies | Bosquet and colleagues, 2007 |
| Optimal taper duration | 2 weeks (overall effect 0.59 +/- 0.33, P < 0.001) | Bosquet and colleagues, 2007 |
| Optimal volume reduction | 41-60%, exponential (overall effect 0.72 +/- 0.36) | Bosquet and colleagues, 2007 |
| Intensity and frequency | No modification (overall effects 0.33 +/- 0.14 and 0.35 +/- 0.17) | Bosquet and colleagues, 2007 |
| Volume, frequency and duration ranges in the literature | Volume cut up to 60-90%; frequency cut no more than 20%; 4 to more than 28 days | Mujika and Padilla, 2003 |
| Typical performance gain | About 3% (usual range 0.5-6.0%) | Mujika and Padilla, 2003 |
Sources: Bosquet, Montpetit, Arvisais and Mujika, Medicine and Science in Sports and Exercise, 2007 (abstract); Mujika and Padilla, same journal, 2003 (abstract).
Two translations matter for anyone who works at a keyboard. A deload is a cut in volume, not in quality: the athletes who gained most kept intensity and frequency and removed roughly half the volume. And the deload is short and scheduled; nobody waits for the athlete to break.
What happens to people who never leave Sprint mode?
The literature on long working hours is, in effect, a study of permanent Sprint, and its reference category is instructive: the major analyses compare long hours with a standard week of 35-40 hours. That band is Maintain mode.
Table 2. The hours ladder: relative risk by weekly hours, compared with 35-40 hours (figures as printed)
| Outcome | 41-48 hours | 49-54 hours | 55 or more hours | Source |
|---|---|---|---|---|
| Stroke incidence, WHO/ILO review | RR 1.04 (95% CI 0.94-1.14) | RR 1.13 (1.00-1.28) | RR 1.35 (1.13-1.61), 7 studies, 162,644 participants | Descatha and colleagues, 2020 |
| Ischaemic heart disease incidence, WHO/ILO review | RR 0.98 (0.91-1.07) | RR 1.05 (0.94-1.17) | RR 1.13 (1.02-1.26), 22 studies, 339,680 participants | Li and colleagues, 2020 |
| Ischaemic heart disease mortality, WHO/ILO review | RR 0.99 (0.88-1.12) | RR 1.01 (0.82-1.25) | RR 1.17 (1.05-1.31), 16 studies, 726,803 participants | Li and colleagues, 2020 |
| Incident stroke, 528,908 people | RR 1.10 (0.94-1.28) | RR 1.27 (1.03-1.56) | RR 1.33 (1.11-1.61) | Kivimäki and colleagues, 2015 |
| Incident coronary heart disease, 603,838 people | Not reported by band | Not reported by band | RR 1.13 (1.02-1.26) | Kivimäki and colleagues, 2015 |
| Onset of depressive symptoms, 189,729 participants, 35 countries | Not reported by band | Not reported by band | 1.14 (1.03-1.25), long hours defined as 55 or more in most cohorts; Asia 1.50, Europe 1.11, North America 0.97 | Virtanen and colleagues, 2018 |
Sources: Descatha et al. and Li et al., Environment International, 2020 (abstracts); Kivimäki et al., The Lancet, 2015 (abstract); Virtanen et al., Scandinavian Journal of Work, Environment and Health, 2018 (abstract). Both WHO/ILO reviews judge the evidence for 55 or more hours “sufficient” for the outcomes shown (stroke incidence; ischaemic heart disease incidence and mortality) and for 41-48 hours “inadequate”.
Three things stand out. The first band above standard hours shows no measurable excess in any dataset; the ladder starts at 49 and steepens at 55. The stroke estimate is the one that holds up best: Kivimäki’s team reports that the excess risk “remained unchanged” under alternative adjustments, with RR estimates from 1.30 to 1.42 and a dose-response trend at p below 0.0001. And the cost is not only cardiovascular: the depressive-symptoms meta-analysis found an overall association of 1.14 between long hours, defined in most cohorts as 55 or more a week, and the onset of depressive symptoms. An earlier meta-analysis of 12 studies by Virtanen and colleagues in 2012 reported that its prospective studies suggest “an approximately 40% excess risk” of coronary heart disease among long-hours workers; the larger 2015 individual-participant analysis came in at 1.13, so the heart-disease signal is real but modest and the stroke signal is the stronger one.
At population scale the arithmetic is not small. Pega and colleagues report in the WHO/ILO joint estimates that 488 million people, or 8.9% of the global population, were exposed to 55 hours or more in 2016, with 745,194 deaths and 23.3 million disability-adjusted life years attributable to that exposure. The full burden table, and who in the EU still works those hours, is in the 40-year career piece and is not rebuilt here.
The Job Demands-Resources model explains why the permanent sprint is self-reinforcing. In the original 2001 study of 374 workers across human services, industry and transport, Demerouti, Bakker, Nachreiner and Schaufeli found that “job demands are primarily related to the exhaustion component of burnout, whereas (lack of) job resources are primarily related to disengagement”. A decade later Demerouti and Bakker summarized the two engines: “job demands are initiators of a health impairment process and job resources are initiators of a motivational process”, and demands become stressors when meeting them requires high effort from which “the employee fails to recover adequately”. Bakker and de Vries’s 2021 extension closes the loop: when strain increases, “employees are less likely to use adaptive self-regulation strategies, such as job stress recovery and job crafting”. The mode that most needs a recovery plan is the mode in which people stop making one. That is why the plan has to be written in Maintain mode. For the structural version of this argument, see burnout as a systems failure.
Who is actually living in permanent Sprint?
ILOSTAT publishes employment by weekly hours actually worked in bands, which allows a direct count of how many employed people worked 49 hours or more, the band where the risks in Table 2 begin.
Table 3. Share of employed people working 49 hours or more a week (hours actually worked), latest year available (CEOtudent analysis of ILOSTAT)
| Country | Year | Share working 49+ hours | Mean weekly hours actually worked | National survey as listed by ILOSTAT |
|---|---|---|---|---|
| India | 2025 | 48.8% | 45.45 | Periodic Labour Force Survey |
| Türkiye | 2025 | 25.0% | 42.06 | Household Labour Force Survey |
| Mexico | 2025 | 24.1% | 42.09 | National Occupation and Employment Survey |
| Viet Nam | 2024 | 24.1% | 41.85 | Labour Force Survey |
| Indonesia | 2024 | 22.3% | 38.20 | National Labour Force Survey |
| Colombia | 2025 | 20.6% | 42.77 | Integrated Household Survey |
| South Africa | 2025 | 17.3% | 42.42 | Quarterly Labour Force Survey |
| Korea, Republic of | 2025 | 15.5% | 38.49 | Economically Active Population Survey |
| United States | 2025 | 11.3% | 37.54 | Current Population Survey (break in series) |
| United Kingdom | 2025 | 8.5% | 34.92 | Labour Force Survey |
| France | 2025 | 7.9% | 35.35 | Employment Survey |
| Germany | 2025 | 4.2% | 33.53 | EU Labour Force Survey |
Sources: ILOSTAT, Employment by sex and weekly hours actually worked (thousands), indicator EMP_TEMP_SEX_HOW_NB_A, and Average weekly hours actually worked per employed person, indicator HOW_TEMP_SEX_ECO_NB_A, both sexes, retrieved 6 November 2026. Shares computed by CEOtudent as the 49+ band divided by total employment; people whose hours are not classified stay in the denominator, so the shares are lower bounds (the unclassified group is 6.4% of the total in Türkiye and 7.0% in Viet Nam). National surveys differ in design, so comparison is indicative.
The share in India is 11.6 times the share in Germany, a gap of 44.6 percentage points. The point for an individual is not the ranking: in several large economies roughly a quarter to a half of all employed people work in the band where the hours ladder starts to climb, and “normal” in one labour market is “permanent Sprint” in the epidemiology.
Two labelled surveys add texture. The International Game Developers Association’s Developer Satisfaction Survey 2021, an industry survey of 803 responses collected from 15 February to 12 April 2021 and analysed at Western University, asks about crunch by name. One third of respondents said their job involved crunch and a further 22% said it required long or extended hours they did not call crunch; adding the two groups, a CEOtudent calculation, puts roughly 55% in some form of recurring sprint. Almost half (47%) felt crunch or long hours was expected as a normal part of the job, 74% among freelancers. During crunch, 26% worked more than 60 hours a week and 29% worked 50-59; 19% of the self-employed reported more than 80. Among employees who had crunched, 58% had done so more than twice in two years, against about 35% of respondents at that rate in 2019. Among employees, one third received no additional compensation for overtime or crunch and only 12% received paid overtime. It is a self-selected sample, but it describes the permanent sprint from the inside: a dated push has become a standing condition.
Gallup’s State of the Global Workplace: 2026 Report, a company-reported survey whose 2025 data were collected from January to December 2025, is the organizational mirror image. Global engagement “dropped to 20% from its peak in 2022 of 23%”; 64% of employees are not engaged and 16% actively disengaged. The steepest fall is among the people who set the mode for everyone else: manager engagement declined by five points between 2024 and 2025, “from 27% to 22%”. Gallup’s construct is proprietary and should be read as that company’s measure, but the direction matches the Job Demands-Resources prediction.
Does Recover mode work, and how long does the effect last?
If Sprint has a measured cost, Recover should have a measured benefit. It does, and the benefit has a half-life.
Table 4. What respite research has measured (figures as printed)
| Study | Design | Result as reported |
|---|---|---|
| Westman and Eden, 1997 | 76 clerks measured twice before, once during and twice after a vacation | Burnout fell during the vacation and returned part way “by 3 days after the vacation and all the way by 3 weeks” |
| de Bloom and colleagues, 2009 | Meta-analysis, 7 studies | Vacation effect “small effect, d=+0.43”; fade-out d = -0.38 as “the overall mean d across 4 studies”, with post-vacation measures “approximately 2 to 4 wk” apart |
| Fritz and Sonnentag, 2006 | 221 university employees, measured 1 week before and 2 days and 2 weeks after vacation | “Vacation effects and partial fade-out effects”; vacation experiences and post-vacation workload predicted some outcomes |
| Grant, Buchanan and Shockley, 2025 (abstract only) | Meta-analysis, 32 studies, 256 effect sizes; activity analysis of 8 studies, 69 effect sizes | Vacation “has a large effect on well-being that does not fade out as quickly as previously thought”; detachment and physical activity “may be the most beneficial” |
Sources: Westman and Eden, Journal of Applied Psychology, 1997 (abstract); de Bloom et al., Journal of Occupational Health, 2009 (full text); Fritz and Sonnentag, Journal of Applied Psychology, 2006 (abstract); Grant et al., Journal of Applied Psychology, 2025 (abstract).
The 2009 and 2025 meta-analyses disagree about size and speed, and the disagreement is informative: seven studies found a small effect, and the four that measured fade-out saw it shrink within weeks; 32 found a larger, slower-fading one. Even the more generous 2025 reading describes a slower fade, not a permanent effect. Recovery is a decaying asset, which is why a single long break cannot carry a career and why Recover has to recur. The de Bloom review also noted that vacation activities “have hardly been studied”, the gap the 2025 analysis begins to fill. How to fill a block so that it restores rather than merely interrupts is the subject of active versus passive recovery; this piece fixes when the block happens and how long it lasts.
The three modes: a CEOtudent editorial framework
The framework below is CEOtudent’s synthesis, not the output of a study. It borrows the structure of periodization and the terminology of the overtraining literature, and uses the evidence above to set the boundaries. Durations are design choices anchored to the sports-science analogues, not clinical thresholds.
Table 5. Sprint, Maintain and Recover (CEOtudent editorial framework)
| Mode | Purpose | Entry signal | Exit criterion | Duration (editorial) | Load relative to Maintain | Non-negotiables |
|---|---|---|---|---|---|---|
| Sprint | A dated outcome with an asymmetric payoff that will not be available later | All three present: a deliverable with a date, a written end date, a Recover block already in the calendar | The end date, or two consecutive weeks of worse sleep plus rising errors, whichever comes first | 2 to 6 weeks; 4 as the default block (the “days to weeks” of functional overreaching) | Hours up, scope narrowed to one target; learning and optional projects paused | A sleep floor, one detached day a week, no second concurrent sprint |
| Maintain | Sustainable output, skill building, and the slack that absorbs surprises | Completion of a Recover block, or any period without a dated sprint | Only into Sprint, and only when the three entry conditions hold | Most of the year | Realized hours inside the 35-40 band the health studies use as reference | Weekly review, protected learning time, a real weekend |
| Recover | Repay the sprint’s debt and reset the baseline | Immediately after a Sprint’s end date, or when two of the three red lines in Table 6 are crossed | Sleep, morning energy and error rate back at the pre-sprint baseline for a full week; minimum one week | 1 to 2 weeks after a 4-week sprint; the taper evidence favours 2 | Volume cut by about half; the few remaining tasks kept at full standard; show up every day | No new commitments, detachment protected, physical activity most days |
The Recover row is the one most people get wrong. The taper evidence says cut volume, keep intensity, keep frequency. A Recover week is therefore not an empty week: the two or three tasks that remain are done to a full standard, everything else is declined or deferred, and the person still appears, replies and ships. A full stop is a vacation, which has its own place in the plan. Recover mode is the working deload between vacations.
What should you measure in each mode?
A mode without a metric is a mood. Each mode gets one leading indicator read weekly, one lagging indicator read at the end of the block, and one red line that forces a change of mode regardless of the plan. The design follows personal KPIs: few numbers, logged the same way every week, read against a baseline rather than a target.
Table 6. Metrics by mode (CEOtudent editorial framework)
| Mode | Leading indicator, weekly | Lagging indicator, block end | Red line that forces a mode change |
|---|---|---|---|
| Sprint | Realized hours, logged not scheduled; sleep against the Maintain baseline | The dated deliverable shipped, judged on a review defined before the sprint | Realized hours above 55 for two consecutive weeks, or sleep below baseline for two: end the sprint and enter Recover |
| Maintain | Share of the week in deep work; one learning session; one fully detached day | A skill or asset added this quarter that outlasts the current role | Three weeks without a detached day, or a quarter without learning: an unlabelled sprint |
| Recover | Morning energy on a 1 to 5 scale against baseline; days with physical activity; new commitments accepted (target zero) | Baseline restored on sleep, energy and errors; next quarter’s plan written | Baseline not restored after two weeks: extend Recover and audit what the sprint cost |
Sleep debt gives the method for the sleep line and energy auditing for the morning-energy line. Neither needs a wearable; the same question at the same time each morning is enough to see a baseline move.
What do you say to a manager or a client when you switch modes?
Mode changes fail at the point of communication, because people announce the push and hide the deload. A manager who hears about four weeks of all-out effort and then sees a quieter fortnight reads the fortnight as slackening. The repair is to announce both halves at once, in operational language, before the sprint starts.
Table 7. Mode-change scripts (CEOtudent editorial framework; dates are placeholders)
| Transition | What to say | What to avoid |
|---|---|---|
| Entering Sprint | “From the 3rd to the 28th the launch takes priority over everything else; here is what is paused. On the 29th a two-week lower-volume period starts, so anything new should be booked for after the 12th.” | Open-ended intensity (“until this is done”); a sprint announced without its recovery block |
| Exiting Sprint into Recover | “The launch shipped on the 28th. As agreed, the next two weeks run at reduced volume: the client review and the weekly report stay at full standard, the rest is deferred.” | Disappearing, or apologizing for a plan agreed in advance |
| Exiting Recover into Maintain | “Baseline is back as of Monday. Capacity for new work opens on the 13th; the next sprint candidate is the Q3 migration, to be dated in the planning meeting.” | Letting the deload drift into an undated low-output period |
| Declining a second concurrent sprint | “That can start in Sprint mode on the 20th, after the current one closes, or now in Maintain mode at a slower pace. Running both at full intensity is how the last two slipped.” | Accepting two sprints and delivering neither on time |
Dates, scope and standard are things a manager or client can plan around; feelings about workload are not. The scripts also give the other side something it rarely gets: a known date on which full capacity returns.
What does a year of alternating modes look like?
The yearly map is a design, not a dataset. Its week counts are editorial choices anchored to two pieces of evidence: the sprint block lasts four weeks, inside the “days to weeks” window of functional overreaching, and the Recover block lasts two weeks, the optimal taper duration in the Bosquet meta-analysis. The shares are recomputed in the accompanying script.
Table 8. A two-peak year (CEOtudent editorial framework; week counts are design choices)
| Weeks | Mode | Purpose |
|---|---|---|
| 1-8 | Maintain | Base season: steady delivery, learning block, planning the first sprint |
| 9-12 | Sprint 1 | First dated outcome of the year |
| 13-14 | Recover | Two-week deload; baseline restored |
| 15-26 | Maintain | Longest steady stretch; the annual review falls here, on complete data |
| 27-28 | Recover | Mid-year break, scheduled before anyone is tired |
| 29-36 | Maintain | Base season; plan the second sprint |
| 37-40 | Sprint 2 | Second dated outcome |
| 41-42 | Recover | Two-week deload |
| 43-50 | Maintain | Close the year’s commitments at sustainable pace |
| 51-52 | Recover | Year-end break |
Totals: Maintain 36 weeks (69.2% of the year), Sprint 8 weeks (15.4%), Recover 8 weeks (15.4%). Every sprint week is matched by a recovery week, and the longest stretch without a Recover block is 12 weeks.
Three design notes. The map has two peaks because one wastes half the year and four leave no base season; Issurin lists the inability “to provide multi-peak performances over the season” among the old model’s failures, and two is the smallest number that fixes it. The spring Maintain block is where the annual review belongs, on closed data rather than in a tired year-end week. And the hours inside each mode are realized hours, not scheduled ones; the four-day week evidence reviews what pilots found when working time was actually cut. A Recover block worked like a sprint is a sprint.
Which mode are you actually in? A self-diagnosis test
Most people cannot name their mode, and the calendar often lies. Answer Column A from the body and the output of the last four weeks; answer Column B from the calendar. Score one point per “yes” in each column.
Table 9. The two-column mode test (CEOtudent editorial framework)
| # | Column A: what the body and the output say | Column B: what the calendar says |
|---|---|---|
| 1 | Realized hours were above 49 in at least two of the last four weeks | A dated deliverable with a written end date sits in the next six weeks |
| 2 | Sleep has been below the usual baseline for two or more weeks | A recovery block of at least one week is booked after the current push |
| 3 | Small errors, rework or missed details have increased | Learning time or an optional project is paused, with a restart date |
| 4 | Morning energy is lower than a month ago on your own 1 to 5 scale | No new commitments accepted in the last two weeks |
| 5 | The last fully detached day was more than two weeks ago | The current week contains at least one fully detached day |
| 6 | Patience with colleagues or clients is shorter than usual | Realized hours in the last month were inside the 35-40 band |
| 7 | The current push has no end date you could state out loud | Two or more pushes are running at the same time |
| 8 | A holiday or long weekend did not restore energy for more than a week | The next recovery block is more than 12 weeks away |
Reading the result: the body says Sprint if four or more of A1 to A7 are “yes”. B1 with B2 means the calendar says Sprint done properly; B1 without B2 means Sprint without a plan; B5 with B6 means Maintain; B2, B3 and B4 together mean Recover.
The diagnosis is the mismatch. Body Sprint and calendar Sprint with a recovery block booked: the system is working, keep going to the date. Body Sprint and calendar Maintain: an unlabelled sprint, the most common state and the one that becomes permanent. Body Sprint and calendar Sprint without a plan: write the end date and book the deload today. Body Recover and calendar Sprint: the sprint is already over physiologically; stop it formally rather than letting it decay. A8 is the most informative single question: a break that did not restore is what the respite research predicts when the load underneath has not changed.
Where does the system fail? Three failure modes
Table 10. Failure modes and repairs (CEOtudent editorial framework)
| Failure mode | What it looks like | What the evidence says about the cost | The repair |
|---|---|---|---|
| Permanent sprint | Every quarter is “the busy one”; realized hours sit at 49 or above; the end date keeps moving | The hours ladder: stroke RR rising from 1.10 at 41-48 hours to 1.33 at 55 or more; the European College of Sport Science terminology describes recovery lengthening from “days to weeks” to “weeks to months” as overreaching stops being functional | Write an end date for the current push within 48 hours and a Recover block after it; if no end date can be written, the problem is scope or role, not stamina |
| Never recovering | Sprints end but Maintain starts at once; holidays are deferred or worked through; the baseline never resets | Vacation effects fade within weeks (d = +0.43, then a fade-out of d = -0.38 between post-vacation measures two to four weeks apart; full fade-out by three weeks in the 1997 respite study), so recovery must recur; strained people use recovery strategies less, so the deficit compounds | Book the year’s Recover blocks now, in Maintain mode, as fixed cost; match every sprint week with a recovery week |
| Recovering without a plan | Time off is taken but filled with errands, unfinished work and screens; energy on return is unchanged | Vacation experiences and post-vacation workload predicted outcomes in the 2006 study; the 2025 meta-analysis points to detachment and physical activity; de Bloom noted activities “have hardly been studied”, so this is the least settled area | Define the block by what it protects (detachment, daily movement, zero new commitments) and by its exit criterion (baseline restored), not by length alone |
The three failures share a root: they treat the modes as something that happens to the person rather than something the person decides. That is the CEO half of the thesis. The Student half is the measurement: the only way to know which mode the body is in is to log it, and the only way to improve the map is to compare the planned year with the realized one. When a sprint breaks the plan anyway, the resilience stack covers how to recover the plan rather than abandon it. Inside any single mode, the 90-minute cycle still applies; periodization sets the season, not the hour.
What this does not mean
- The sports analogy is a frame, not a proof. Afonso and colleagues’ 2017 review of 42 randomized periodization trials found that more than 95% studied the physical factor, mainly strength, and that no study lasted more than nine months. Even in sport, periodization is a practice with mixed empirical backing; applying it to knowledge work is a design choice made explicit here.
- Relative risks are population associations. Table 2 rests on self-reported hours in cohorts. A relative risk of 1.33 does not mean a given person’s risk is a third higher, and the 41-48 band shows no measurable excess.
- The ILOSTAT shares are indicative. National surveys differ in design, unclassified hours remain in the denominator, and the United States series carries a break flag.
- The surveys are what they say they are. The IGDA sample is self-selected game developers; Gallup’s engagement measure is proprietary. Neither is a population estimate.
- The week counts are design choices. Nothing in the literature says a knowledge-work sprint should last four weeks or a deload two. The numbers borrow sports-science durations as starting points to be tested against your own log.
- This piece reports research findings for general information only. It is not medical advice; anyone with symptoms affecting sleep, mood or cardiovascular health should consult a qualified clinician rather than a periodization table.
Frequently asked questions
How long should a sprint last?
The framework defaults to four weeks, within a range of two to six, because the European College of Sport Science terminology places recoverable overreaching in a window of days to weeks. The firmer rule is that the end date is written before the sprint starts and a Recover block is booked after it.
Is working 45 hours a week dangerous?
Not measurably, on the evidence in Table 2. The 41-48 hour band shows no significant excess in the WHO/ILO reviews or the Lancet meta-analysis; the ladder begins at 49 and steepens at 55.
Why does Recover mean “reduce volume” rather than “stop”?
Because that is what the deload evidence supports: the best taper results came when training volume fell 41-60% while intensity and frequency were unchanged. At work: keep the few remaining tasks at full standard and keep showing up, but remove about half the volume.
How often should Recover blocks happen?
Often enough that the benefit has not fully decayed. The 2009 meta-analysis found effects fading between post-vacation measures two to four weeks apart and the 1997 study found complete fade-out by three weeks; the 2025 meta-analysis finds a slower fade but not a permanent one. The yearly map schedules a Recover block at least every 12 weeks.
What if the job does not allow a Recover block?
Then the job is in permanent sprint, and the first step is to say so and measure it. Log realized hours for a month; if they sit at 49 or above with no end date, the conversation is about scope and staffing, using the scripts in Table 7, not about stamina.
Does this apply to freelancers and founders?
More than to anyone. In the IGDA survey, 74% of freelancers felt crunch or long hours were expected as normal and 19% of the self-employed worked more than 80 hours during crunch. Without a manager to set the mode, the calendar has to.
Sources
- Bosquet, L., Montpetit, J., Arvisais, D., and Mujika, I. (2007). Effects of tapering on performance: a meta-analysis. Medicine and Science in Sports and Exercise, 39(8), 1358-1365 (abstract only).
- Mujika, I., and Padilla, S. (2003). Scientific bases for precompetition tapering strategies. Medicine and Science in Sports and Exercise, 35(7), 1182-1187 (abstract only).
- Issurin, V. B. (2010). New horizons for the methodology and physiology of training periodization. Sports Medicine, 40(3), 189-206 (abstract only).
- Meeusen, R., Duclos, M., Foster, C., and colleagues (2013). Prevention, diagnosis, and treatment of the overtraining syndrome: joint consensus statement of the European College of Sport Science and the American College of Sports Medicine. Medicine and Science in Sports and Exercise, 45(1), 186-205 (abstract only).
- Kreher, J. B., and Schwartz, J. B. (2012). Overtraining syndrome: a practical guide. Sports Health, 4(2), 128-138. Open-access full text, including the European College of Sport Science terminology table.
- Afonso, J., Nikolaidis, P. T., Sousa, P., and Mesquita, I. (2017). Is empirical research on periodization trustworthy? A comprehensive review of conceptual and methodological issues. Journal of Sports Science and Medicine, 16(1), 27-34 (abstract only).
- Kivimäki, M., Jokela, M., Nyberg, S. T., and colleagues, IPD-Work Consortium (2015). Long working hours and risk of coronary heart disease and stroke: a systematic review and meta-analysis of published and unpublished data for 603,838 individuals. The Lancet, 386(10005), 1739-1746 (abstract only).
- Descatha, A., Sembajwe, G., Pega, F., and colleagues (2020). The effect of exposure to long working hours on stroke: a systematic review and meta-analysis from the WHO/ILO Joint Estimates of the Work-related Burden of Disease and Injury. Environment International, 142, 105746 (abstract only).
- Li, J., Pega, F., Ujita, Y., and colleagues (2020). The effect of exposure to long working hours on ischaemic heart disease: a systematic review and meta-analysis from the WHO/ILO Joint Estimates of the Work-related Burden of Disease and Injury. Environment International, 142, 105739 (abstract only).
- Pega, F., Náfrádi, B., Momen, N. C., and colleagues (2021). Global, regional, and national burdens of ischemic heart disease and stroke attributable to exposure to long working hours for 194 countries, 2000-2016: a systematic analysis from the WHO/ILO Joint Estimates of the Work-related Burden of Disease and Injury. Environment International, 154, 106595. Open-access full text.
- Virtanen, M., Heikkilä, K., Jokela, M., Ferrie, J. E., Batty, G. D., Vahtera, J., and Kivimäki, M. (2012). Long working hours and coronary heart disease: a systematic review and meta-analysis. American Journal of Epidemiology, 176(7), 586-596 (abstract only).
- Virtanen, M., Jokela, M., Madsen, I. E., and colleagues (2018). Long working hours and depressive symptoms: systematic review and meta-analysis of published studies and unpublished individual participant data. Scandinavian Journal of Work, Environment and Health, 44(3), 239-250 (abstract only).
- Demerouti, E., Bakker, A. B., Nachreiner, F., and Schaufeli, W. B. (2001). The job demands-resources model of burnout. Journal of Applied Psychology, 86(3), 499-512 (abstract only).
- Demerouti, E., and Bakker, A. B. (2011). The Job Demands-Resources model: challenges for future research. SA Journal of Industrial Psychology, 37(2), article 974. Open-access full text.
- Bakker, A. B., and de Vries, J. D. (2021). Job Demands-Resources theory and self-regulation: new explanations and remedies for job burnout. Anxiety, Stress and Coping, 34(1), 1-21 (abstract only).
- Westman, M., and Eden, D. (1997). Effects of a respite from work on burnout: vacation relief and fade-out. Journal of Applied Psychology, 82(4), 516-527 (abstract only).
- de Bloom, J., Kompier, M., Geurts, S., de Weerth, C., Taris, T., and Sonnentag, S. (2009). Do we recover from vacation? Meta-analysis of vacation effects on health and well-being. Journal of Occupational Health, 51(1), 13-25. Open-access full text.
- Fritz, C., and Sonnentag, S. (2006). Recovery, well-being, and performance-related outcomes: the role of workload and vacation experiences. Journal of Applied Psychology, 91(4), 936-945 (abstract only).
- Grant, R. S., Buchanan, B. E., and Shockley, K. M. (2025). I need a vacation: a meta-analysis of vacation and employee well-being. Journal of Applied Psychology, 110(7), 887-905 (abstract only).
- International Labour Organization, ILOSTAT. Employment by sex and weekly hours actually worked (thousands), indicator EMP_TEMP_SEX_HOW_NB_A, and Average weekly hours actually worked per employed person by sex and economic activity, indicator HOW_TEMP_SEX_ECO_NB_A. Retrieved 6 November 2026.
- International Game Developers Association (2021). Developer Satisfaction Survey 2021: Summary Report. Johanna Weststar and colleagues, Western University, Ontario. Industry survey, 803 responses.
- Gallup (2026). State of the Global Workplace: 2026 Report. Company-reported survey; 2025 data collected January to December 2025.
Tables 1, 2 and 4 report figures as printed in the sources named beneath them. Table 3 is a CEOtudent analysis of ILOSTAT counts; the shares, the 11.6 ratio and the 44.6-point gap are CEOtudent calculations, as is the roughly 55% combination of the two IGDA crunch groups. Tables 5 to 10 are CEOtudent editorial frameworks, and the week counts in Table 8 are design choices rather than research findings. Deliberately not used: the mesocycle durations often attributed to block periodization, because only the abstract of Issurin’s review was opened; the pooled effect sizes of the 2025 vacation meta-analysis, because only its abstract was accessible; and percentage restatements of the WHO/ILO relative risks from press material, which were not opened for this piece; Table 2 gives the relative risks with confidence intervals as printed in the reviews.
This content was compiled with the support of AI following in-depth research, then written and prepared for publication by the CEOtudent editorial team.
This post is also available in:
