In May 2022 I handed in a thesis asking whether pressure makes people more productive at work. I wrote it one month before the final jury of my degree, about the team I was in, while that team was under pressure. The premise was personal: I had always believed I worked better with a deadline breathing down my neck.
Neither the literature nor the project supported that belief. But the finding worth writing down is not that stress is bad. It is that the workload, the thing every team argues about, is the least interesting of the three ingredients that produce stress.
The cost is documented well enough to stop arguing about it
The World Health Organization and the International Labour Organization put long working hours, assumed to act through increased occupational stress, at the top of the list of work-related risk factors by attributable disease burden: around 745,000 deaths from ischaemic heart disease and stroke in 2016. The European Agency for Safety and Health at Work puts stress and psychosocial risk behind half to sixty per cent of all lost working days. The French figure everyone quotes, two to three billion euros a year, is narrower and older than it sounds: it prices job strain alone, on data from the 2000s, and reads as a floor. In early 2014 the consultancy Technologia counted over three million workers at high risk of burnout.
Absence is where it compounds. Two people go off, four cover for them, and if the four carry the extra load long enough they get there too. The losses are not linear.
The productivity link itself is less dramatic than the death toll and more useful to a team lead. Cross-sectional work from 2010 and 2021 finds the expected negative correlation: more stress, less output. The stronger association runs the other way. Job satisfaction tracks quality and productivity closely, and people with energy affect the output of everyone around them.
Three sources, and only one is the deadline
Robert Karasek’s model is the one I kept. Demands and control are his, from 1979. The third factor, support, was added by Johnson and Hall about a decade later. Between them they say occupational stress comes from three factors, not one.
Psychological demands are the load: quantity of work, complexity of tasks, time constraints, contradictory requests, interruptions. This is the factor everyone names.
Decision latitude is how much control someone feels they have over what they do. Plot it against demands and the grid explains more than the demands column alone. High demand with high autonomy is active work, and it is where people perform. High demand with low autonomy is strain, and it is where they break. The same deadline produces both, depending on the second axis.
Social support is being heard, by peers and by the people above you. Low support at work predicts health problems independently of the other two.
Jacques Broda got to a similar place in 1986 with the three Ps: pressure of hours, hierarchical pressure, and no professional prospects. Two of his three are not about the amount of work either.
In a design team, the pressure lands on whoever ships last
Our project had checkpoints at three, two and one month before the jury. Those were the peaks, and the trigger was not the volume of work. It was having the work judged.
Two other things generated more stress than the load itself. Outside feedback that contradicted itself, which meant sorting divergent demands ourselves before we chased a project that no longer made sense. And transferred stress, which in a design team has a direction.
Designers hand over to developers. Developers are at the end of the production chain, so they inherit the pressure mechanically. If the mockups arrive late, the developers are in the rush and there is nothing they can do about it. That is not a personality problem or a planning problem on their side. It is the shape of the chain, and the only people who can fix it are upstream.
What it actually cost
Stress gives you an adrenaline push and then spends the reserve. What I logged in myself and in the group: energy gone by the low hours of the day, concentration lost to wondering what caused the stress in the first place, and rumination.
The clearest case was a feature some of us could not picture yet, weeks before we had settled the parameters we needed to build it. Worrying about that gap nearly had us produce features in the wrong order. We talked it through calmly and found a solution, but I am fairly sure the days spent chewing on it cost more output than the gap did.
Then the parts nobody puts in a retrospective. Creativity narrows, which at a design school is disabling. Personality changes: I snapped at a teammate over something that did not matter, and passed my own stress on to someone else. Lateness, because stress and sleep do not coexist. Withdrawal, because the people who most need to talk are the ones with no energy left to.
What worked, and it was mostly structure
Take a task off whoever is drowning. We did this on purpose. Stress overruns a person quickly if no hand goes out.
Match the load to actual capability, and say so out loud. I told the group I was not competent at illustration. That is a two-second conversation that prevents a week of bad work.
Decide by majority. On a project with a lot of early decisions it was the fastest method we found and the least frustrating.
Give autonomy. The parts of the project where people had real decision latitude are where the game mechanics and the interfaces got interesting. That is Karasek’s active work, observed from the inside.
Ask questions early. Ambiguity does not resolve itself, and screen production is exactly when you discover what nobody wrote down.
Plan, and take breaks. A precise weekly list, work split across the group. On the days we skipped it, things got disordered and the volume felt heavier than it was. I still run Pomodoro.
None of this argues for zero pressure. Hans Selye coined eustress for the good kind, the appetite for performance you need daily, and the dose makes the poison in both directions: too much enthusiasm for a goal causes its own collapse. The psychiatrist Patrick Légeron put the boundary plainly: too much stress kills performance, but nobody performs without any.
Four years on
I ended the thesis on a question rather than an answer: if autonomy and being consulted matter this much, what does that say about control tools and about flat hierarchies, and what organisation of work should we be aiming at.
I have not answered it. What I did keep is the diagnostic. When a team is struggling, the reflex is to talk about the workload, because it is the visible factor and the one you can put in a spreadsheet. Two thirds of the model is somewhere else: how much of their own work people get to decide, and whether anyone is listening. Both of those are design decisions about how a team is arranged, and both of them are free.