The most interrupt-driven workplace on earth is a hospital ward, and it is there that medicine decided to measure what an interruption actually costs. Westbrook and colleagues [1] directly observed nurses preparing and administering medications across six wards of two teaching hospitals and put numbers on what every professional suspects: each interruption was associated with a 12.1% increase in procedural failures and a 12.7% increase in clinical errors, and error severity rose with interruption frequency. Medicine’s response is the instructive part. Many hospitals did not attempt to eliminate interruptions, which in a ward would be absurd and dangerous; interruptions are how urgent information moves. They zoned them. Nurses on medication rounds wear do-not-interrupt tabards, an intervention real enough to have been trialled in controlled studies. Nobody pages a surgeon mid-anastomosis. The ward stays gloriously interruptible, and inside it sit small, fiercely protected zones where the cognitively critical work happens, because the data said that is where interruptions kill.
Now consider the principal investigator, who runs academia’s version of the ward: a role that is interrupt-driven by design, because being interruptible is part of the job, students need unblocking, decisions queue, urgent things are sometimes urgent. And note what the PI’s week is missing: the zones. No tabard, no theatre, no protected round. The result is a career’s strangest arithmetic, the person hired for their research being the only member of the group with no protected research time, writing grants at eleven p.m. because the day belongs to everyone else. The fix is not discipline, and it is certainly not unavailability. It is the hospital’s answer: keep the ward interruptible, and build the zones.
Table of Contents
- Daniel’s calendar, autopsied
- Zoning the week: the five structures
- “I stopped to check my inbox regularly”: the mechanism, at any altitude
- What Daniel’s retrofit produced
- A word for the researcher whose supervisor just went quiet
Daniel’s calendar, autopsied
Daniel, whose transition and whose supervision craft the first two pieces of this article series already followed, performed this exercise I recommend to every group leader: a one-week interruption autopsy. He logged every unplanned demand on his attention. The count came to 37. Reviewing the log squarely: five were truly urgent. About a dozen were developmental questions his students could have batched or, with a nudge, answered themselves. The rest arrived immediately because arriving immediately was free. His deep-work total for the week, the thinking, writing and analysis he is nominally paid for, came to three fragments. None lasted longer than 40 minutes. Each was ended by a knock. The autopsy’s real finding was not the volume. It was that Daniel’s availability had no shape. Every hour was equally interruptible, so the interruptions spread evenly across all of them. And the cognitively critical work, the very class of work Westbrook’s nurses were doing when the errors spiked, happened in whatever was left. At night. Degraded.
Zoning the week: the five structures
The hospital model translates into five structures, and together they answer the objection every caring PI raises immediately, which is that their people need them.
Maker blocks: the medication round. Two to four half-day blocks per week, calendar-blocked, visibly named, for the work that cannot survive fragmentation: writing, analysis, thinking. The software world calls this the maker’s schedule as opposed to the manager’s, and the underlying cognition is the attention-residue effect [2], the fact that part of your mind stays on the interrupted task long after the interruption ends, a mechanism I take apart in my article on deep work for researchers. During a maker block you are, professionally speaking, mid-anastomosis: door closed, notifications off, and the block defended in your calendar with the same seriousness as a funder meeting.
Office hours: the interruptible ward, made better. Zoning only works if availability is generous somewhere, and this is the piece PIs skip at their peril. Daily or near-daily open slots, published, where interruption is not tolerated but welcomed. This is not less access than the open door; it is better access: a student with a stuck experiment gets your full attention at 14:00 instead of your irritated half-attention at a random 10:40, and the twelve batchable questions from Daniel’s autopsy become one focused conversation.
The urgency protocol: one agreed channel. Genuine emergencies bypass everything, and saying so explicitly is what makes the zones safe: agree with your group what counts as urgent and how it reaches you (a phone call, a specific signal), and suddenly nobody has to knock “just in case”. Hospitals do not protect the theatre by hoping nothing urgent happens; they protect it with a crash protocol. Yours costs one lab-meeting agenda item.
The default-forward rule, issued from above. The deepest source of interruption load is students waiting for permission that was never required. The rule I teach researchers to request from below, if I have raised a decision and heard nothing in an agreed window, I proceed and you correct me later, works even better issued by the PI, because it converts your silence from a blocker into a decision and deletes a whole class of knocks at the source. Notice what it also is: delegation, which is to say development. Every question answered instantly is a small decision-making rep stolen from the person who asked it, and the situational framework in the previous piece of this cluster tells you who is ready to be defaulted forward.
Do not become your students’ Google. I tell every first-time supervisor the same sentence, because it deletes more interruptions than any calendar trick: do not become your students’ Google. Google is what people turn to when they want a quick answer instead of thinking a problem through, and an always-answering PI serves the same function at a higher price, because your researchers need to learn to work a problem. Many of the questions in Daniel’s dozen simply disappear when a ready answer stops being free. In Fast Forward we teach a best-practice format for asking: outline the situation, state the problem you are trying to solve, name at least two solutions you considered, and say which one you are leaning towards and why. It trains the asker to work the problem first. And it pays you back directly: a specific problem in front of you gets actionable feedback, where generic questions get generic answers. The other half of the rule is response speed. An instant answer teaches your group that instant answers exist, and from then on every question arrives expecting one. So do not respond instantly, even when you could. Set specific times when you deal with requests, tell your students what they are, and hold them. The question that can wait until your 14:00 window either gets worked on by its owner in the meantime or dissolves entirely. Both outcomes are the system working.
“I stopped to check my inbox regularly”: the mechanism, at any altitude
The zoning mechanism is not a seniority perk; it is the same physics at every level, which is why one of the clearest accounts of it in our feedback comes from a researcher, not a group leader:
“I started to implement a low energy time box for proceeding emails and stopped to check my inbox regulary. I have the feeling that, thereby, my stress level is reduced.”
Look at the two moves inside that sentence, because they are the whole model in miniature. First, the zoning: email, the self-inflicted interruption, gets a designated low-energy container instead of ambient, all-day access to the person’s attention. Second, the reversal of default: stopped checking regularly means attention is no longer interrupt-driven by habit, and the reported result is the one the whole literature predicts, lower stress, because a mind that knows when the interruptions are allowed stops bracing for them the rest of the time. Another participant named the downstream effect on boundaries in six words that every overloaded PI should tape to a monitor: “More confidence especially in saying no referring to my timetable.” That is what a zoned calendar buys that willpower cannot: the no stops being personal, because the timetable says it for you. A researcher who learns this in their doctorate arrives at the PI role already zoned. The rest of us, like Daniel, retrofit.
What Daniel’s retrofit produced
Daniel installed the five structures over a month, in the order given, and braced for resentment that never came. The autopsy the following term told the story: interruptions down by half, not because his group needed him less but because the needs now had addresses, office hours for the batchable, the protocol for the urgent, the default-forward rule for the merely blocked, and the asking format for the questions that needed working first. His maker blocks held on about four weeks in five, which is what success actually looks like. And the effect he had not predicted echoes the one from his supervision retrofit: his students describe him as more available, not less, because available now means fully present at known times rather than half-present at random ones. The eleven p.m. grant writing did not vanish entirely. It stopped being the system and went back to being the exception, which is the real ceiling for this role. The ward is still a ward. It just has a theatre in it now.
A word for the researcher whose supervisor just went quiet
And if you are not a PI at all, but a doctoral researcher whose supervisor recently announced office hours and stopped answering pings within the hour, this close is for you, because the change is easy to misread. Your supervisor has not become less committed to you; they have moved from random half-attention to scheduled full attention, and almost everything in it is yours to use: the office hours are an invitation to bring the question whole instead of in fragments, the urgency protocol is your guarantee that real emergencies still cut through, and the default-forward rule is the largest grant of trust available in a supervision relationship, the explicit right to proceed. Your side of the machinery, the agendas, the summaries, the specific requests, the well-built question, has an article of its own: my piece on the PhD supervisor relationship, and the skills underneath all of it, for both sides of the desk, are what Fast Forward, our project management and personal productivity training, trains across seven weeks, with over 1,500 individuals trained at more than 30 institutions. The hospital never apologised for the tabard, and the patients were safer for it. Your supervisor’s calendar just grew a theatre. Bring your work to the ward, and let the theatre do what theatres are for.
FAQ
How do PIs find time for research?
The ones who manage it stop treating research time as what remains and start zoning the week, the way hospitals zone interruptions: two to four protected maker blocks for deep work, generous published office hours for everything batchable, one agreed channel for genuine emergencies, and a default-forward rule so students can proceed through silence. Availability with a shape beats availability everywhere, for the PI and the group alike.
Should a PI have an open-door policy?
Not literally, and the interruption data explains why: each interruption measurably degrades cognitively demanding work, and a door open at all hours spreads interruptions evenly across every hour, including the ones your hardest work needed. Better than the open door is the open window: daily office hours where interruption is welcomed at full attention, plus an emergency channel that bypasses everything. Students get more of you, not less.
How many hours of deep work can a PI realistically protect?
Two to four half-day blocks a week is the realistic, sustainable range for a truly interrupt-driven role, held about four weeks in five. Chasing the researcher-era ideal of unbroken days produces only guilt; the PI role is legitimately a manager’s schedule with a maker’s core, and the win is making the core structural instead of nocturnal.
How do I stop my group interrupting me constantly?
Give the interruptions somewhere better to go rather than suppressing them: batchable questions get office hours, emergencies get an explicit protocol, and permission-seeking, the biggest hidden category, gets deleted at the source with a default-forward rule (raise a decision, wait an agreed window, proceed). And stop being your group’s Google: ask for the situation, the problem, at least two solutions they considered and which one they lean towards, and half the questions answer themselves. Answer at set times you have communicated, not instantly; instant answers train instant asking. Interruptions are usually a symptom of shapeless availability, not of a needy group.
References
1. Westbrook, J. I., Woods, A., Rob, M. I., Dunsmuir, W. T. M., & Day, R. O. (2010). Association of interruptions with an increased risk and severity of medication administration errors. Archives of Internal Medicine, 170(8), 683–690.
2. 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), 168–181.
Last Updated on August 5, 2026 by Dr Nadine Sinclair
Dr Nadine Sinclair
Dr. Nadine Sinclair is a molecular biologist, strategy consultant and co-founder of Mind Matters. A scientist by training and at heart, she conducted her doctoral research at the Max Planck Institute for Biophysical Chemistry. She has 30,000+ hours of hands-on project management experience, built across 18 years as a strategy consultant at McKinsey and as an independent consultant, working with research institutions, foundations, pharmaceutical and biotech companies and governments. She is the co-creator of the Personal Resilience Indicator and the author of On Track. Since 2018, over 1,500 researchers from more than 30 institutions have trained with Mind Matters.