The old Key Organization site maintained a page called “Time Management Statistics.” Its subject still matters, but figures from different populations and years should not be treated as a single verdict on productivity.
Three useful numbers, with three different denominators
35%: In the US Bureau of Labor Statistics' 2025 American Time Use Survey, 35% of employed people did some or all of their work at home on days they worked. This is not a global remote-work rate, nor the share who worked exclusively at home. The same release says 70% did some work at a workplace on days worked; the groups can overlap.
Two minutes: Microsoft's 2025 Work Trend Index analysis estimates an average two-minute interval between meetings, email and chat pings during core work hours for the Microsoft 365 users in its telemetry. A ping is not necessarily an interruption that a person noticed. The dataset excluded education and EU tenants; it is not a census of workers everywhere.
Thirty countries: The OECD Time Use Database covers 30 OECD countries and groups time into paid work or study, unpaid work, personal care, leisure and other uses. Its country surveys come from different years and methods. Use it to frame a question, then read the country and survey notes before making a comparison.
What these numbers cannot prove
Working at home does not by itself show higher or lower productivity. A count of messages does not measure useful collaboration. Time spent working does not reveal the quality or value of the outcome. The Microsoft result describes a digital environment; the BLS survey describes where time was spent; the OECD database describes categories of time use. Each has a different unit of analysis.
Before repeating a statistic, record its year, geography, sample, exact question or event, denominator and known exclusions. Check whether it describes an average, a median, a subgroup or a change over time. If a chart combines several studies, label the sources separately rather than drawing a false continuous trend.
Measure one week of friction in your own work
- Choose one outcome. For example: fewer delayed handoffs, a shorter turnaround for a recurring report, or more uninterrupted drafting time. Avoid a vague target like “be more productive.”
- Log events, not people. For five days, record the planned work, the interruption type, its duration and whether it was necessary. Do not use the log to rank employees.
- Find the repeated cause. Look for missing information, unclear ownership, duplicate approvals or scheduled meetings without a decision.
- Change one shared rule. Put the current source of truth in one place, clarify who decides, or batch a non-urgent request. Repeat the same measurement the next week.
This local comparison is still observational. A better result may reflect a quieter week or a different project mix. Keep the context visible and repeat before generalizing.
The archived statistics page is evidence of a historical topic, not a source for current numeric claims. We wrote this page anew and linked each published number to its original publisher.
Check the original material
- US Bureau of Labor Statistics — American Time Use Survey, 2025 results ↗
- Microsoft WorkLab — Breaking down the infinite workday, 2025 ↗
- OECD — Time Use Database and comparability notes ↗
- Wayback Machine — former Key Organization statistics page ↗
For a practical next step, see our five-day friction log and measurement guide.