How to Manage Interruptions at Work

How much does one interruption actually cost?

manage interruptions at work is a key topic in modern project management and teamwork. More than the interruption and less than the internet claims. A realistic model has three parts: the time the interruption takes, the time to return to the task, and the quality lost while your attention is split.

Take a morning with four interruptions — a colleague’s question, a client message, a notification you chose to check, and a short call. The interruption time is maybe 12 minutes. Resumption is the expensive part: if each event costs 10–20 minutes of reduced focus, four interruptions can consume 40–80 minutes of a four-hour block. That is a third of your most productive window, gone, with nothing to show on a timesheet.

Example: the same day, two ways

A developer has a four-hour block to finish an API integration. In scenario A, five pings arrive and each is answered within a few minutes. In scenario B, the pings are batched into two windows and one is escalated because it is genuinely blocking.

• Scenario A: 5 interruptions × 15 minutes of effective resumption = 75 minutes lost. The block yields about 2.75 hours of work, and the integration is 60% done.
• Scenario B: 2 windows × 12 minutes = 24 minutes, plus one 8-minute unblocking call. The block yields about 3.5 hours, and the integration ships.

Same person, same day, same requests. The difference is scheduling, not discipline.

Example: the team-level bill

Eight people, each losing an average of 45 minutes a day to interruption recovery, at a blended cost of $65 per hour:

• 8 × 0.75 hours = 6 hours a day
• 6 × $65 = $390 a day
• × 250 working days = $97,500 a year

That figure is a model, not a research finding, and it moves with every assumption. Its value is the comparison it enables: a $97,500 annual cost justifies a process change, while a $9,000 one might not.

What does the research actually say about resumption time?

The most-cited work here is Gloria Mark’s research programme on interrupted work, including “The Cost of Interrupted Work: More Speed and Stress” (Mark, Gudith, and Klocke, CHI 2008). Across that body of field and lab research, people return to an interrupted task after roughly 22–23 minutes on average, and self-interruptions take longer — close to 29 minutes in her field data.

Two findings from that study matter more than the headline number:

People compensate by working faster, and pay in stress. Interrupted participants completed the task in about the same time as uninterrupted ones, but with higher stress, frustration, time pressure, and effort. The output looked fine; the human cost was hidden.

Self-interruptions are not cheaper. Checking a notification yourself is still a switch, and in the study self-interruptions took longer to recover from than external ones. This is why “I chose to look” is not a defence.

The number that circulates as “23 minutes and 15 seconds per interruption” is a rounded public summary. It is close to the study’s external-interruption figure and it is often used to claim that a worker interrupted 275 times a day loses 4,400 minutes. That arithmetic is nonsense — the 23 minutes is time to resume one interrupted task, not a toll charged per notification. Most notifications are ignored, scanned in a second, or handled in a batch without any task switch at all.

A better frame comes from Sophie Leroy’s research on attention residue. When you switch tasks, part of your attention stays behind on the unfinished one, and performance on the new task suffers. Her experiments also found that a brief plan for returning — deciding what you will do next on the task you are leaving — reduces the residue. That is a practical lever you can use in ten seconds, and it is the basis of the restart ritual later in this article.

Which interruptions deserve an immediate answer?

Very few. Triage by consequence of delay, not by who is asking or how loudly.

The test is simple: if the person cannot progress without you, answer. If they can, defer. Most requests fall in the second category, and treating them as urgent is a habit, not a service.

> Important note: a “quick question” is often a request for a decision that has not been written down. If the same question arrives three times from three people, the fix is a decision log or a short document, not faster replies.

How do you signal focus without going dark?

Declare the block, publish the escape route, and keep the declaration credible. Going dark without a path creates anxiety in the team and eventually gets overridden by the loudest person.

Make the block visible. A calendar block called “Focus — pricing model, until 11:30” tells people what you are doing and when you are back. An empty block with “busy” tells them nothing and invites a test.

Give one escalation channel. “If this cannot wait, call me” is a complete policy. It converts a hundred small decisions into one rule, and in practice the phone rings rarely because most things can wait.

Use a status that changes behaviour. In a shared workspace, a status of “heads down until 11:30, call if blocking” is more effective than headphones alone. Headphones signal “maybe”; a written status with a time signals “not now, and here is when”.

Keep office hours for the questions that need a human. Two 30-minute windows a day absorbs most of the “quick question” traffic without fragmenting the rest of the day. Publish them once, and hold them even when you are busy — the predictability is what makes the rest of the day quiet.

What channel rules actually change behaviour?

Rules change behaviour when they are specific, published, and paired with an alternative. “Please use async communication” is a slogan. A table with response-time expectations is a contract.

Three rules make the table stick.

Every channel needs an owner. An unowned shared channel becomes a place where requests go to die, and people route around it by messaging individuals.

The escalation path must be boring. If the only way to get an answer is to interrupt, people will interrupt.

Leaders have to obey the contract first. A manager who pings at 9 p.m. and expects an answer has deleted the policy, whatever the document says.

Example: a meeting audit that pays for itself

A team of six has 11 recurring meetings a week, averaging 30 minutes, with five attendees each.

• 11 × 0.5 hours × 5 attendees = 27.5 person-hours a week
• At $65/hour, that is $1,788 a week, or roughly $89,000 a year

Cutting three meetings and shortening two others to 20 minutes removes about 8 person-hours a week — around $26,000 a year in recovered time, plus the interruption load those meetings created through preparation and follow-ups. This is usually the highest-leverage change available, and it is a calendar edit rather than a culture programme.

How do you cut interruption load at the team level?

Individual discipline has a ceiling. If the environment generates 270 pings per person per day, personal willpower is not the constraint — the system is.

Audit meetings before optimising messages. Meetings are scheduled interruptions with a calendar invite, and they fragment the day into unusable pieces. A 30-minute meeting at 10:00 and another at 11:00 destroys the morning more than two hours of meetings would.

Create one intake queue for requests. If work arrives through chat, email, tickets, and corridor conversations, nobody can prioritise it. One queue with a visible owner makes the load measurable, which turns “we are at capacity” into a fact.

Write decisions down. A decision logged with a date and an owner kills the repeat question and removes the most common reason for a “quick” interruption: nobody can find what was agreed.

Rotate the interruptible role. Support, on-call, and client escalation duty should be explicit and rotating. When everyone is implicitly on duty, everyone is interrupted.

Protect one shared focus window. A team-wide no-meeting block from 9:00 to 11:00 is more effective than eight individual attempts at the same thing, because it removes the social cost of being unavailable.

How do you recover once you have already been interrupted?

Recovery is a skill, and it takes about a minute. The goal is to leave a trail for your future self, not to reconstruct everything from memory.

1. Write the next action before you switch. One line: “Resume at: rewrite the intro paragraph, then check the churn numbers.” This is the “ready to resume” move that reduces attention residue.
2. Park the open question. If you were mid-decision, note the two options and the missing input. Half-formed decisions decay fastest.
3. Handle the interruption fully. A half-answered question produces a follow-up in 20 minutes. Answer, confirm, close.
4. Return to the same task, not to email. The first three minutes after an interruption decide whether you resume or drift. Opening the inbox is a decision to end the block.
5. Set a two-minute timer for re-entry. Re-read your note, then start the smallest physical action: open the file, scroll to the paragraph, type one sentence.

The team version of this is a culture where “where were we?” is a normal question with a written answer. Teams that keep decision logs and task comments resume faster than teams that rely on memory.

When does protecting focus backfire?

Always, in some situations. The trade-offs are real and worth naming.

Customer support and on-call work cannot be batched. If your job is to answer within 15 minutes, a focus block is a service failure. Use rotation: one person interruptible, the rest protected.

Juniors need fast access early on. A new hire who waits three hours for a blocking answer learns to guess, and guessing produces rework.

Batching can delay the wrong thing. A message that looks routine can be a client about to churn. Build one exception: anyone can mark a message “blocking” and get a same-hour answer.

Focus culture can become an excuse. “I was in deep work” is legitimate once or twice a day and corrosive when it means “I do not answer anyone”.

> Common mistake: Announcing a no-interruption policy without changing the escalation path. People will respect the rule until the first thing that cannot wait, then route around it permanently. Publish the exception before you need it.

Where do tools fit?

Most of this is process, not software, and a team with clear channel rules and a shared calendar can implement all of it with tools it already owns. Software helps in two specific places: making the current state of work visible enough that people stop asking, and keeping requests in one queue instead of five.

Doitify is our project and team management platform, and those two jobs are what it is built for: tasks and sub-tasks with owners and due dates, milestones, checklists, project documents, meeting notes, risks, and work reports in one workspace, plus team chat channels and workload views. Being explicit: Doitify is our product, and it will not fix a team that has not agreed on response times — no tool will. What it does is remove the “where are we?” message that generates a third of most interruption traffic.

Conclusion

Interruptions are not a discipline problem to be solved with willpower. They are a design problem: which requests reach you, when they arrive, and how fast you can return to what you were doing. So publish the channel contract, protect one or two focus blocks with a phone-call escape route, batch the rest, and write a one-line restart note every time you switch. Then fix the system: audit the meetings, create one intake queue, log the decisions people keep asking about, and rotate who is interruptible.

The goal is not silence. It is a day where a two-minute question costs two minutes, and the four hours you protected produce four hours of work.

If this post on manage interruptions at work was helpful, you might also enjoy Time Tracking Project Management Tool and Project Management Tool With Gantt Chart.