17 Statistics on the Cost of Fragmented Attention

What the research says about interruption, recovery, and a working week cut into pieces too small to think in

The complaint is old and the measurements are not. Attention at work is now divided finely enough that the unit of the working day is closer to a minute than an hour, and the recovery cost of each division is larger than the interruption itself. The statistics below cover both halves of the problem: the small interruptions that arrive continuously, and the scheduled ones that take the rest.

Key Takeaways

• Workers are interrupted every two minutes, 275 times a day during core hours, from Microsoft’s telemetry rather than a survey.
• 57% of meetings are ad hoc calls without a calendar invite, so most of the interruption never appears on a calendar anyone could protect.
• It takes about 23 minutes to fully refocus after an interruption, the most quoted figure in the field, traceable to Gloria Mark’s research at UC Irvine.
• Average time on a single screen has fallen from 2.5 minutes in 2004 to 47 seconds in Mark’s most recent data.
• 48% of employees and 52% of leaders say their work feels chaotic and fragmented, per the same Microsoft report.
• Reaching creative flow takes 15 to 20 minutes, which is longer than the average gap between interruptions.

The Cost of a Single Interruption

1. Roughly 23 minutes to refocus

Gloria Mark’s work at UC Irvine, published in No Task Left Behind? (CHI 2005) and The Cost of Interrupted Work (CHI 2008), established the cost of interruption in peer-reviewed research rather than a vendor survey.

The precise figure of 23 minutes and 15 seconds circulates far more widely than the papers do, and it traces to Mark’s later work rather than to either CHI paper. Treat the seconds as false precision and the order of magnitude as sound: recovery takes tens of minutes, not seconds.

Mark’s more recent measurements put the gap higher still, at about 25 and a half minutes before someone returns to the original task, because the interruption is typically followed by a second activity of roughly ten minutes before anyone circles back.

2. People complete two other tasks before returning

The recovery is not a straight line back. Mark’s No Task Left Behind? (CHI 2005) found people typically complete two intervening tasks before returning to the interrupted one, which is why the cost compounds beyond the clock time.

3. Interruptions arrive about every ten minutes

Gloria Mark’s measurements put workers at about 10 and a half minutes on a task before something interrupts them. Set against a 23-minute recovery, the arithmetic does not resolve: interruptions arrive faster than recovery completes.

4. Flow takes 15 to 20 minutes to reach

Reaching the state researchers call flow takes 15 to 20 minutes of uninterrupted work. Longer than the average gap between interruptions, which means for many people it is not reached at all on an ordinary day.

How Finely the Day Is Now Divided

5. 47 seconds on a screen, down from 2.5 minutes

Mark’s longitudinal data shows average time on a single screen falling from 2.5 minutes in 2004 to 47 seconds in her most recent measurements. The trend is the finding: attention spans at work have not merely been short, they have shortened by a factor of three in twenty years.

6. Nearly 1,200 application switches a day

Harvard Business Review’s study of 137 users across three Fortune 500 companies found the average person toggled between apps and websites nearly 1,200 times a day. Each switch costs a little over two seconds, adding up to just under four hours a week, about 9% of working time, spent reorienting.

Across an eight-hour day that is a switch roughly every 24 seconds. Most are not interruptions in the classic sense; they are the work itself, spread across tools that do not talk to each other.

7. 48% of employees say work feels fragmented

Microsoft’s Work Trend Index reports 48% of employees and 52% of leaders describing their work as chaotic and fragmented. Notable that leaders report it more than the people they lead.

8. Interrupted every two minutes

Microsoft’s telemetry shows employees using Microsoft 365 are interrupted every 2 minutes by a meeting, email or notification, roughly 275 interruptions a day. This is measured platform data rather than self-report, which is what makes it worth more than a survey answer. The average worker also receives 153 Teams messages per weekday, up 6% year over year.

The Scheduled Half

9. 21.7 meetings a week

The average knowledge worker now attends 21.7 meetings per week. Spread across five days that is more than four a day, each with its own before and after.

10. 15.4 hours in meetings against 12.1 hours of focus

The same research puts 15.4 hours a week in meetings against 12.1 hours of uninterrupted focus work. Meetings now consume more of the week than the work they exist to coordinate.

11. Executives spend nearly 23 hours a week in meetings

Senior leaders reach close to 23 hours weekly, more than double the figure from the 1960s when the average was under ten. Other measurements put C-suite loads nearer 11 hours, or 28.3% of the week, so the range is wide and the direction is consistent.

12. 392 hours a year in meetings

The typical knowledge worker sits through a 392-hour annual meeting load, which the same research puts at 9.8 full workweeks a year, or about 11.3 hours a week, roughly 28% of a standard workweek.

13. 57% of meetings never appear on a calendar

Microsoft’s telemetry finds 57% of meetings are ad hoc calls without a calendar invite, and that one in ten of the meetings that are scheduled get booked at the last minute.

This is the figure that defeats the standard advice. Blocking focus time defends against the minority of meetings someone planned. Most arrive without an invite, and no calendar setting prevents them.

Two more from the same telemetry: nearly a third of meetings now span multiple time zones, up 35% since 2021, and edits in PowerPoint spike 122% in the final ten minutes before a meeting starts.

14. Estimates of the aggregate cost run to tens of billions

Unnecessary meetings are estimated to cost US businesses around $37 billion a year, with broader estimates of unproductive meeting time reaching far higher. Aggregate cost figures like these are built on assumptions about hourly rates and should be read as order-of-magnitude claims.

Why the Two Halves Are One Problem

15. Meetings generate the interruptions between meetings

A calendar full of meetings does not only consume its own hours. It produces the follow-ups, the chased decisions and the “quick question” messages that fill the gaps, which is why reducing meeting count without changing how decisions get recorded tends to move the load rather than remove it.

16. The gaps are too small to use

Between a 23-minute recovery cost and meetings every couple of hours, the intervals left over are frequently shorter than the time required to start. This is the mechanism behind the common experience of a full day that produced nothing: the day was not idle, it was granular.

17. The switching is mostly between tools, not tasks

At one application switch every minute and a half, most of the fragmentation is not people changing what they are working on. It is people going to fetch something: a file, a thread, a status, a number. That is a tooling problem in the guise of a discipline problem.

The Pattern Underneath

The research points somewhere slightly different from where the advice usually goes. Most guidance treats fragmentation as a behavioral failure to be fixed with discipline: block your calendar, turn off notifications, batch your email. That helps, and it is a personal remedy for a structural condition. We have argued elsewhere that your productivity hacks aren’t helping your team, because one person’s protected focus time is the person waiting on them being blocked.

The structural version is that information lives in more places than a person can hold, and the work of assembling it is what generates the switching. Somebody asks about an account, and answering means a mail thread, a chat, and a record in a third system. The interruption cost is not really the question. It is the six switches required to answer it.

Which is the argument for putting the assembly somewhere other than a person’s attention. If the answer to “what happened with this customer” already exists in one place, gathered as the messages arrived, then the question costs one switch instead of six. That is what the Brain is for, and why workflows act on messages as they land rather than waiting for somebody to notice.

How We Sourced This

Gloria Mark’s interruption and recovery figures come from peer-reviewed research at UC Irvine, published at CHI in 2005 and 2008 with later longitudinal data. They are the most solid numbers here. Microsoft’s Work Trend Index figures are drawn from platform telemetry alongside survey data.

The meeting volume and cost figures come from vendors selling meeting tools and from aggregators, and they disagree with each other by wide margins, particularly on executive hours. We have reported the disagreement rather than picking the most dramatic figure, and flagged the aggregate dollar estimates as models built on assumed hourly rates.