Sourced

The Numbers

This is not our research. It's every statistic about email, switching costs and life admin that this site quotes — the publisher, the date, what was actually measured, and, honestly, where the evidence runs out.

1 · The hours

What's actually measured about time lost to email.

Until today, this site's homepage led with a widely-repeated 2012 estimate — email eating 28 percent of the average knowledge worker's week. We went to re-verify it at publish time and couldn't: mckinsey.com timed out on every fetch attempt, on three separate URLs, across two sessions. No independent press coverage that isn't itself a listicle restates the figure either. Per our own sourcing rule — can't verify, don't claim — we dropped it, and we're not quoting it here as a live number. We're leading with current, vendor-telemetry evidence instead.

117

Primary telemetry

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

What was actually measured: Aggregated, anonymized Microsoft 365 productivity signals for the global M365 install base (window ending 15 Feb 2025, excluding EU and education tenants), triangulated with a companion Edelman Data x Intelligence survey of 31,000 knowledge workers across 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, and EU and education tenants are excluded entirely from the signal.
Our read: This is the number we lead the homepage with now, instead of a 2012 estimate we can no longer re-verify. Current, from the vendor's own aggregated signal, with its population and its blind spots stated.

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 was actually 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, not specifically an email read.
Caveat: Counts every interruption type together, so it measures how often attention is pulled, not how long each pull actually costs — that's what the switching-cost research in section 2 is for.
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 — which is the actual 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 says nothing about what each one costs you afterward. Three studies below look at that instead — and one of them complicates the story rather than confirming it, which is exactly why it's here.

attention residue

Peer-reviewed

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

What was actually measured: Two controlled experiments, per the abstract: participants switched between tasks, and the contrasts examined were finishing versus not finishing the prior task, and whether there was time pressure while finishing it. The abstract reports that finishing alone was not enough — time pressure while finishing was what enabled disengagement and higher performance on the next task.
Caveat: Lab task-switching experiments, not real inboxes — the effect itself is well replicated in cognitive-psychology research, but its size at real-world inbox scale isn't what this paper measured. 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're not aware of an inbox-scale replication of it.
Our read: This is the mechanism the raw hour-counts in section 1 can't show: 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

57% of the "working spheres" (task periods) in a field study of 24 information workers were interrupted before they were finished.

What was actually measured: Direct shadowing and activity-timing of 24 information/knowledge workers at one organization, over roughly 700 combined observation hours, logging every task period and whether it ended by interruption or by voluntary switching.
Caveat: A field study from the mid-2000s at a single organization, well before smartphones and Slack-style messaging were part of daily work — treat the exact figures as evidence of a pattern, not a current benchmark.
Age: Published 2005 — 21 years old; cited here for the pattern (fragmentation is the norm, not the exception), not as a current figure.
Our read: The oldest study on this page, and still the clearest evidence that a fragmented workday isn't multiple tasks running well — it's one task, constantly restarted.

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

faster, not slower

Peer-reviewed

In a controlled study, people who were interrupted mid-task completed the work in the same time or less, not more — but reported significantly more stress, frustration, time pressure and effort doing it.

What was actually 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's interruptions aren't timed the way this experiment's were, and the stress cost may not transfer 1:1 to daily work.
Age: Published 2008 — 18 years old.
Our read: The honest complication: an interruption's cost may not show up as "lost minutes" in a time-tracking log at all — it shows up as stress you can't see there. That's part of why we don't 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's whole premise is that one assistant has to span work and personal life, not just a work inbox. Government time-use data measures the personal half. No vendor survey we could find does — and that gap is worth stating plainly rather than papering over.

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

Official statistics

UK women averaged 3h 37m a day of unpaid work — housework, caring for others, volunteering — versus men's 2h 43m a day.

What was actually measured: A UK diary-style time-use survey fielded 23 September–1 October 2023; experimental statistics.
Caveat: UK-only, a single one-week measurement window rather than an annual average, and self-reported diary data rather than observed time.
Our read: The clearest government-grade evidence for the "other half" of the workday that most productivity content ignores entirely — and it isn't 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 was actually 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: Self-reported time, not a time-and-motion measurement; what counts as "routine admin" is the vendor's own definition and isn't independently audited, and no fielding window is published, so the figure can't be placed in time.
Conflict of interest: Fyxer is an AI email-assistant vendor and a competitor we compare against at /vs/fyxer — this figure comes from research it commissioned and published about a problem its own product is sold to solve.
Our read: Labelled for exactly what it is: a competitor's own commissioned number, not independent research — which is why it doesn't anchor this section. Worth being precise about scope too: this is routine admin at work, not personal life admin, so it does not answer the question this section is actually asking.

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

distribution, not hours

Peer-reviewed

Household "cognitive labor" — anticipating needs, deciding among options, monitoring outcomes — is real, consistently gendered, and often invisible to both partners, but the rigorous literature measures its distribution and burden, not how many hours it takes.

What was actually 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 via the Work and Family Researchers Network's listing of it rather than reading the paper. A qualitative interview study, not a diary or time-use survey: it cannot support an hours-per-week figure, and doesn't attempt one.
Our read: This is why we don't publish an "hours of life admin per week" figure anywhere on this site: the best research on this measures who carries the mental load, not how many hours it takes, and we'd rather say that plainly than back into a number nobody has actually measured.

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

Where the evidence runs out

There is no defensible hours-per-week figure for personal "life admin" — 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's distributed, not how many hours it takes, and it doesn't attempt to. Note what this means 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, and we're not going to present it as one just because it sits in this section. (Fyxer is also a company we compare Beemy against directly at /vs/fyxer; we name that conflict because it costs us nothing to say and would cost the page's credibility to hide.) So the gap here is total rather than partial: we won't invent a number to fill it, and the closest honest proxy we have is the government time-use data above — which measures unpaid household work in general, not "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 — modelled as a separate type from everything above precisely so they can never be mistaken for one. We'll publish real numbers from our own beta once we have them, and until then we won't put a number on how much time Beemy itself saves anyone.

60%

Our assumption

We assume a triage-and-drafting assistant can realistically hand back about 60% of routine email-handling time — not all of it.

Triage — read, classify, decide, draft a first pass — is repetitive and context-dependent rather than judgment-heavy, and it's most of the time in most inboxes we've looked at early in beta. The remaining share is judgment calls that still come back to you. This is our own working assumption, not a measured industry figure, and we haven't yet published usage data from our own beta to replace it with a real one.

2.2 hrs/day

Our assumption

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

Our own rough translation of the volume in the "117 emails a day" and "interrupted every 2 minutes" telemetry above into a starting hours-per-day figure for the calculator — not a distinct, separately-measured stat about hours spent on email. Move the slider to your own number; this default is a starting point, not a claim.

Get your time back.

Beemy is in private beta. Join the waitlist and go from connect to a quiet, triaged inbox — in your first session, not a week later.

Join the waitlist Talk to us