Deep work in an open office

It is 10:14 a.m. on a Tuesday. You arrived at the office with a clear intention: finish the design document for the new data pipeline before noon. You sat down, opened your laptop, wrote two paragraphs — genuinely good paragraphs — and then your manager pinged you about a vendor call rescheduled for 11:00. You switched to email to check whether the agenda had changed. You noticed a thread about the quarterly planning deck. You answered a quick question in Slack. You glanced at the clock: 10:47. You returned to your document. You reread the two paragraphs to find your thread again. By noon, you had added one more paragraph and attended the call. The document would not be done until Thursday.

This is not a time-management failure. It is a concentration failure — and it is the default mode of almost every office environment in the world.

What deep work actually means

Cal Newport coined the phrase “deep work” to describe professional activity performed in a state of distraction-free concentration that pushes your cognitive capabilities to their limit. The contrast is “shallow work”: logistical, administrative, low-cognitive-demand tasks that are easily replicated and often interrupted — email, status updates, scheduling, quick reviews.

Neither category is optional. Both are real work. The distinction matters because they live on different cognitive rails. Shallow work can be done in five-minute slivers scattered through the day. Deep work cannot. Writing architecture docs, debugging a non-obvious system failure, designing a product strategy, writing a pitch — these require you to hold a large, complex mental model in working memory simultaneously, and that model collapses the moment an interrupt fires.

The trap of the modern workplace is that it optimizes almost entirely for shallow work: open-plan offices, always-on chat tools, back-to-back meetings, the cultural expectation of sub-five-minute response times. If you do not build an explicit system to protect deep work, it will not happen.

The real cost of a single ping

When a notification appears and you glance at it — even if you do not act on it — your brain has already left the task. Cognitive psychologists call this “attention residue”: even after you return to the original work, part of your attention stays with the interrupting thought. A 2001 study by Gloria Mark at UC Irvine found that after a workplace interruption, it takes an average of 23 minutes and 15 seconds to return to the original task at full depth.

Twenty-three minutes. For a single Slack ping.

In a typical open-office environment, workers are interrupted or self-interrupted every three to five minutes. Do the arithmetic: if every interruption costs twenty-plus minutes of recovery, a day with twelve interruptions has no real depth in it at all — only the illusion of productivity measured in activity rather than output.

This is not about being antisocial or precious. It is about understanding what you are actually trading when you keep notifications on all day.

Maker schedule vs manager schedule

In 2009, the programmer and essayist Paul Graham wrote a short piece distinguishing two modes of scheduling. The manager schedule is a calendar divided into one-hour blocks that can each be handed to a different task or person. This works fine when your job is to coordinate, decide, unblock. A meeting from 10:00 to 11:00 costs one block.

The maker schedule — the one software engineers, writers, analysts, and designers need — operates on half-day units. A meeting from 10:00 to 11:00 does not cost one block. It splits the morning into two 90-minute fragments that are each too short for serious depth, effectively costing the entire morning.

Most organizations run on the manager schedule by default and expect makers to adapt. Makers who do not push back will find their entire week colonized by meetings scheduled by people whose workflows are genuinely undamaged by them.

The framework is not adversarial. It is diagnostic. If your output depends on deep concentration, and your calendar is built on the manager schedule, you have a structural problem that no amount of willpower will fix.

A day divided: fragmented vs blocked

The diagram below shows what a real workday looks like under these two conditions — the same eight hours, the same tasks, radically different outputs.

The fragmented day is not lazy — it is exhausting. But exhaustion is not the same as output.

How to build the blocked day

The mechanics are not complicated. The difficulty is social and cultural. Here is the system that holds up in practice.

Time-block before the week begins. On Friday afternoon or Sunday evening, open your calendar for the coming week and place your deep work blocks before anyone else can. Two blocks per day is the ideal; one is the minimum. Three hours in the morning is worth more than three hours in the afternoon for most people, because willpower and glucose are higher. Guard these blocks the same way you would a client meeting. They are not “free time” — they are your most productive scheduled commitment.

Define what goes in each block. A deep block is not “work on the big project whenever I feel like it.” It is a specific pre-decided output: “write the risk section of the architecture doc” or “debug the latency regression in the ETL pipeline.” Vague intentions collapse under the weight of starting inertia. A specific output does not.

Notifications off is the default, not the exception. Turn off Slack notifications, email badges, phone alerts — everything — for the duration of the block. If your team culture requires constant availability, this is the cultural norm to push back on explicitly (more on that below). Start with your early morning block if you have a later-starting team: the first 90 minutes of the day are often protected by default because nobody expects a response at 7:30 a.m.

Batch all shallow work into one window. Schedule a single daily slot — 45 to 90 minutes, usually mid-day or late afternoon — for all the stuff that is genuinely shallow: answering emails, reviewing pull requests, updating trackers, attending routine standups. Process everything in this window. Outside this window, do not check. When colleagues know you will respond within the batch window, most questions become less urgent the moment you do not reply instantly.

The shutdown ritual. This is the part most productivity advice omits. Cal Newport is emphatic about it: at the end of every workday, close out your tasks, review tomorrow’s calendar, make one note about where to pick up the next deep block, and say a literal phrase out loud — “shutdown complete.” The ritual is not performative. It creates a hard psychological boundary that signals to your brain that work is done. Without it, the background process that keeps re-activating task memories never switches off. You end up half-working during dinner and half-distracted during deep blocks.

The async norm you need your team to agree on

Individual systems break if the surrounding culture rejects them. A person who turns off Slack in an always-on team will be perceived as disengaged, will miss urgent things occasionally, and will face social pressure to revert within weeks.

The conversation to have is direct and framed around outcomes, not preferences. It sounds something like: “I’ve noticed that we lose a lot of productive time to real-time interruptions, including for things that are not actually urgent. Can we agree on a norm that questions with a same-day response time are the default, but anything truly urgent gets a phone call or a specific tag we all monitor?” Most teams agree to this when someone names it explicitly — the always-on default persists not because people want it, but because nobody has questioned it.

Document the norm. “Slack response within four hours during business hours; anything with URGENT in the header gets a 30-minute response.” This gives you a defensible standard rather than a personal preference. It also protects everyone on the team, not just you.

A useful asymmetry to point out in that conversation: the person asking a question pays one second to send a Slack message. The person receiving it pays twenty minutes if they were in deep work. The cost is asymmetric. Making that asymmetry visible often changes behavior without further argument.

When you are on the manager schedule

Not every role permits long deep-work blocks, and not every day will go to plan. If you are in a high-coordination role — engineering manager, product lead, operations — your days may be legitimately structured around the manager schedule. The maker framework still applies to the output that requires depth from you personally: strategy documents, difficult decisions, hiring assessments, one-on-ones you actually think about rather than wing.

Even on a manager schedule, protecting one 90-minute block daily is feasible. Put it at 7:00 a.m. before the first meeting if the afternoon is impossible. Use it for the one thing only you can do deeply.

The honest cost of not doing this

Organizations implicitly measure knowledge workers by presence signals: responsiveness, meeting attendance, visible busyness. Output is harder to measure, so activity fills the gap. A developer who responds in two minutes and attends every optional meeting looks productive. The same developer who turns off notifications and ships twice as much each week is harder to see.

This means the stakes of building a deep-work practice are partly political. You need your manager and peers to trust your output metrics rather than your presence metrics. That requires delivering visibly and communicating clearly about what you are doing during those dark blocks — not defensively, but as a professional who takes output seriously.

The best case you can make is a short track record. Defend the blocks for two weeks. Show the work. Most managers will leave you alone after that.

The fragmented day has one virtue: it feels safe. You are always available. You never miss anything. You are perpetually in the loop.

What you are trading for that safety is the work you came here to do.