Sourced

The Numbers

This is not our research. Every statistic about email, switching costs and life admin that this site quotes is here. Each one shows the publisher, the date and what the study measured. We also say where the evidence runs out.

15 min read

On this page

1 · The hours

What studies measure about time lost to email.

This site's homepage used to lead with a widely repeated 2012 estimate. It said email takes 28 percent of the average knowledge worker's week. We tried to re-verify it when we published this page. We could not. The mckinsey.com site timed out on every fetch. We tried three separate URLs in two sessions. No independent press coverage that is not itself a listicle repeats the figure either. Our sourcing rule is: if we cannot verify it, we do not claim it. So we dropped the figure and we do not quote it here as a live number. We lead with current vendor telemetry instead.

117

Primary telemetry

The average worker receives 117 emails a day, most of them skimmed in under 60 seconds.

What the study measured: Aggregated, anonymised Microsoft 365 productivity signals for the global M365 install base. The window ends 15 Feb 2025 and excludes EU and education tenants. Microsoft checked them against a companion Edelman Data x Intelligence survey of 31,000 knowledge workers in 31 markets, fielded 6 Feb–24 Mar 2025.
Caveat: This is Microsoft's own telemetry about Microsoft 365 users specifically. It says nothing about Gmail-only workplaces. The signal excludes EU and education tenants entirely.
Our read: Our homepage leads with this number now. It replaced a 2012 estimate that we can no longer re-verify. It is current and comes from the vendor's own aggregated signal. We state its population and its blind spots.

Microsoft WorkLab — Breaking down the infinite workday (2025 Work Trend Index special report)(2025-06-17 · checked 2026-08-02)

every 2 minutes

Primary telemetry

People using Microsoft 365 are interrupted by a meeting, email, or notification roughly every 2 minutes during the workday.

What the study measured: Same Microsoft 365 telemetry window as above (through 15 Feb 2025, excluding EU and education tenants). An interruption is any meeting, email or notification event. The data does not isolate email reads.
Caveat: It counts every interruption type together. It measures how often attention is pulled. It does not measure what each pull costs. The switching-cost research in section 2 covers that.
Our read: The volume number (117 emails) tells you the pile is big. This one tells you the pile never lets you focus for more than two minutes at a stretch. That is the complaint most people have.

Microsoft WorkLab — Breaking down the infinite workday (2025 Work Trend Index special report)(2025-06-17 · checked 2026-08-02)

2 · Why the hours undercount it

The switching cost the raw hour-counts can't show.

A count of emails or interruptions per day does not show what each one costs you afterward. Three studies below look at that cost. One of them complicates the story instead of confirming it. We include it for that reason.

attention residue

Peer-reviewed

Switching away from an unfinished task leaves "attention residue". Part of your attention stays on the abandoned task. This measurably impairs performance on the next one, even minutes later.

What the study measured: Two controlled experiments, per the abstract. Participants switched between tasks. The studies compared finishing and not finishing the prior task. They also tested time pressure while finishing it. The abstract reports that finishing alone was not enough. Time pressure while finishing let people disengage and perform better on the next task.
Caveat: These are lab task-switching experiments. They do not use real inboxes. Cognitive-psychology research replicates the effect well. But this paper did not measure its size at real-world inbox scale. We verified the abstract only. The full text is not open access.
Age: Published 2009: 17 years old. The attention-residue effect is still standard, frequently cited cognition research. We know of no inbox-scale replication of it.
Our read: The raw hour-counts in section 1 cannot show this mechanism. Even a 10-second glance at an email can cost you more than the 10 seconds it took to read it.

Organizational Behavior and Human Decision Processes (Elsevier) — Why is it so hard to do my work? The challenge of attention residue when switching between work tasks(2009 · checked 2026-08-02)

57%

Peer-reviewed

In a field study of 24 information workers, 57% of the "working spheres" (task periods) ended in an interruption before the work was finished.

What the study measured: Researchers shadowed and timed 24 information/knowledge workers at one organisation, for roughly 700 combined observation hours. They logged every task period. They noted whether it ended by interruption or by voluntary switching.
Caveat: A field study from the mid-2000s at a single organisation. Smartphones and Slack-style messaging were not yet part of daily work. Treat the figures as evidence of a pattern. Do not treat them as a current benchmark.
Age: Published 2005 (21 years old). We cite it for the pattern: fragmentation is the norm. It is a historical marker. It is not a current figure.
Our read: The oldest study on this page. It is still the clearest evidence of what a fragmented workday is: one task, restarted again and again, never several tasks running well at once.

ACM (CHI 2005 Proceedings) — No Task Left Behind? Examining the Nature of Fragmented Work(2005 · checked 2026-08-02)

same speed, more stress

Peer-reviewed

In a controlled study, people who were interrupted mid-task completed the work in the same time or less. But they reported significantly more stress, frustration, time pressure and effort doing it.

What the study measured: Lab experiment with 48 subjects (81 percent German university students, mean age 26), randomly assigned to interrupted vs. uninterrupted email-drafting conditions with telephone or IM interruptions.
Caveat: A university-student lab sample under a designed interruption protocol. A real inbox does not time its interruptions the way this experiment did. The stress cost may not transfer 1:1 to daily work.
Age: Published 2008, 18 years old.
Our read: This study complicates the story. An interruption's cost shows up mostly as stress, frustration and effort. A time-tracking log does not capture those. That is part of why we do not put a single number on "time lost to switching" anywhere on this site.

ACM (CHI 2008 Proceedings) — The Cost of Interrupted Work: More Speed and Stress(2008 · checked 2026-08-02)

3 · The other half of your life

Work isn't the only place busywork lives.

Beemy is built on one idea: one assistant must cover work and personal life, not only a work inbox. Government time-use data measures the personal half. No vendor survey we could find does. We state that gap plainly. For what that half looks like, and what Beemy already takes off it, read Life admin.

3h 37m (women) vs 2h 43m (men)

Official statistics

UK women averaged 3h 37m a day of unpaid work (housework, caring for others, volunteering). Men averaged 2h 43m a day.

What the study measured: A UK diary-style time-use survey fielded 23 September–1 October 2023; experimental statistics.
Caveat: UK-only. It covers a single one-week window. It is not an annual average. The diary data is self-reported. Nobody observed the time directly.
Our read: The clearest government-grade evidence for the "other half" of the workday. Most productivity content ignores it entirely. And it is not split evenly.

Office for National Statistics (UK) — Time use in the UK: 23 September to 1 October 2023(2023-11-09 · checked 2026-08-02)

5.6 hours

Vendor survey

Routine admin takes 5.6 hours per employee a week, with email reported as the single biggest time-waster.

What the study measured: A commissioned survey of 5,000 UK and US office workers, published January 2026. The report does not state when the survey was fielded.
Caveat: The time is self-reported. Nobody measured it with a time-and-motion study. The vendor defines what counts as "routine admin", and nobody audits that definition independently. The report publishes no fielding window, so we cannot place the figure in time.
Conflict of interest: Fyxer is an AI email-assistant vendor and a competitor we compare against at /vs/fyxer. It commissioned and published this research. The research covers a problem that Fyxer's own product is sold to solve.
Our read: We label it for what it is: a competitor's own commissioned number. It is not independent research. That is why it does not anchor this section. The scope matters too. It measures routine admin at work. This section asks about personal-life admin, which is a different question.

Fyxer — The Admin Burden Report(2026-01 · checked 2026-08-02)

who carries it

Peer-reviewed

Household “cognitive labor” (anticipating needs, deciding among options, monitoring outcomes) is real, consistently gendered, and often invisible to both partners. The rigorous literature measures who carries it and how heavy it is. It does not measure how many hours it takes.

What the study measured: 70 in-depth interviews with members of 35 couples; qualitative coding of who performs which cognitive labor tasks.
Caveat: Abstract confirmed only. The full text is paywalled, and the publisher's own article page blocks our fetcher. So we verified the abstract through the Work and Family Researchers Network's listing. We did not read the paper. It is a qualitative interview study, not a diary or time-use survey. It cannot support an hours-per-week figure, and it does not try to.
Our read: This is why we do not publish an "hours of life admin per week" figure anywhere on this site. The best research measures who carries the mental load. It does not measure how many hours it takes. We say that plainly. We will not back into a number that nobody has measured.

American Sociological Review — The Cognitive Dimension of Household Labor(2019 · checked 2026-08-02)

63 million, six in 10 employed

Vendor survey

63 million Americans, nearly 1 in 4 adults, gave ongoing care in the past year. They cared for an adult or a child with a complex medical condition or a disability. Six in 10 family caregivers are employed on top of that care.

What the study measured: A nationally representative, probability-based online panel from Ipsos: 6,858 US caregivers aged 18 or older who had cared for an adult relative or friend, or a child with a complex medical condition or disability, within the past 12 months.
Caveat: US-only, and self-reported: a person's own account of hours and status, not an observed measurement. We state the employment share in the same words as the publisher's press release. A linked summary page on the same site gives a different figure for a narrower age group.
Conflict of interest: AARP is a membership organisation that sells its own products and services to older adults and caregivers. It is not a neutral research body. But this figure comes from a fielded survey. It is not an internal estimate.
Our read: The clearest evidence that a working carer is not a rare case. Most family carers already hold down a job on top of the caring. That is the reader this persona page is for.

AARP and the National Alliance for Caregiving — New Report Reveals Crisis Point for America's 63 Million Family Caregivers(2025-07-24 · checked 2026-09-26)

Where the evidence runs out

There is no defensible hours-per-week figure for personal "life admin". That means the errands, admin and household coordination that never show up on a work timesheet. The rigorous academic literature on household cognitive labor measures who carries it and how it is shared. It does not try to measure how many hours it takes. This matters for the vendor survey above. Fyxer's 5.6 hours is routine admin at work, measured on office workers. It is not a life-admin figure. It sits in this section, but we do not present it as a life-admin figure. (Fyxer is also a company we compare Beemy against at /vs/fyxer. We name that conflict because hiding it would cost this page its credibility.) So the gap here is total, not partial. We will not invent a number to fill it. The closest proxy we have is the government time-use data above. It measures unpaid household work in general. It does not narrow to "admin" specifically.

4 · Our own assumptions

The numbers on this site that aren't measurements.

Two figures on the inbox-cost calculator are ours, not a citation. We store them as a separate type from everything above, so nobody can mistake them for a citation. We will publish real numbers from our own beta when we have them. Until then, we will not put a number on how much time Beemy saves anyone.

60%

Our assumption

We assume a triage-and-drafting assistant can realistically hand back about 60% of routine email-handling time. The rest still needs you.

Triage (read, classify, decide, draft a first pass) is repetitive and depends on context. It needs little judgement. In most inboxes we have looked at in early beta, it is most of the time. The remaining share is judgement calls that still come back to you. This is our own working assumption, not a measured industry figure. We have not yet published usage data from our own beta to replace it.

2.2 hrs/day

Our assumption

The inbox-cost calculator's default slider assumes 2.2 hrs/day spent on email.

This is our own rough translation of the "117 emails a day" and "interrupted every 2 minutes" telemetry above. We turned that volume into a starting hours-per-day figure for the calculator. It is a derived estimate. It is not a separate, measured stat about hours spent on email. Move the slider to your own number. This default is only a starting point, never a claim.

2.6 hrs/day, 6 days/wk, $45/hr, 120 min/wk

Our assumption

The founders preset on the inbox-cost calculator starts from 2.6 hrs/day on email. It uses a 6-day working week, $45/hr, and 120 min/wk of other recurring admin.

This is a starting point, not a measurement of any real founder's time. The inbox runs on customers, candidates and investors. The role often spills into the weekend. A recurring investor update adds more.

2.8 hrs/day, 5 days/wk, $55/hr, 150 min/wk

Our assumption

The team managers preset on the inbox-cost calculator starts from 2.8 hrs/day on email. It uses a 5-day working week, $55/hr, and 150 min/wk of other recurring admin.

This is a starting point, not a measurement of any real manager's time. It assumes more 1:1s than hours in the day, plus a recurring status update.

1.8 hrs/day, 5 days/wk, $40/hr, 180 min/wk

Our assumption

The working parents preset on the inbox-cost calculator starts from 1.8 hrs/day on email. It uses a 5-day working week, $40/hr, and 180 min/wk of other recurring admin.

This is a starting point, not a measurement of any real parent's time. Two inboxes leave less clear time for email. The class newsletter and other recurring reading add more. A work-only figure would miss them.

3 hrs/day, 5 days/wk, $65/hr, 90 min/wk

Our assumption

The consultants preset on the inbox-cost calculator starts from 3 hrs/day on email. It uses a 5-day working week, $65/hr, and 90 min/wk of other recurring admin.

This is a starting point, not a measurement of any real consultant's time. It assumes three client inboxes instead of one. It also assumes a freelance rate that must cover its own overhead.

1.5 hrs/day, 5 days/wk, $40/hr, 240 min/wk

Our assumption

The working carers preset on the inbox-cost calculator starts from 1.5 hrs/day on email. It uses a 5-day working week, $40/hr, and 240 min/wk of other recurring admin.

This is a starting point, not a measurement of any real carer's time. The job runs on Gmail like anyone else's. A parent's appointments, calls and forms come on top, and many fall outside a normal inbox entirely.

Get your time back.

Beemy is in private beta. We send invites in rolling batches. We cannot promise a date. When your invite arrives, connect Gmail. Beemy sorts your inbox in that first session.

Talk to us

Last updated 17 October 2026

The Numbers

What studies measure about time lost to email.

The switching cost the raw hour-counts can't show.

Work isn't the only place busywork lives.

The numbers on this site that aren't measurements.

How trust is earned, one step at a time

What Beemy is built to catch

Plans and pricing

Get your time back.