Why measure instead of estimate?

Because the estimate is not a weaker version of the measurement, it is a different thing. The American Time Use Survey, run by the US Bureau of Labor Statistics, measures how much time people spend on activities such as paid work, childcare, volunteering and socializing, and it does not do it by asking people how their week generally goes. Interviewers collect a 24-hour diary running from 4 in the morning on the day before the interview to 4 in the morning on the interview day, asking respondents to report every activity in that period, including where they were and who they were with. A same-day diary, not a recalled average, is what a national statistics agency considers reliable enough to publish.

How to run one week

Pick an ordinary week, not an unusually quiet or unusually brutal one. Record as you go rather than at the end of the day, because the end of the day is where recall starts substituting for memory. Use a small set of categories, roughly five to ten, that reflect the questions you have, and keep them stable all week; renaming categories on Wednesday destroys the totals. Include the unpaid and unglamorous parts, admin, commuting, interruptions and recovery from them, since those are usually the finding. Then stop. One week is enough to see the shape, and a longer audit mostly produces a longer record and a lower chance you finish it.

Reading the result without flinching

Total each category first and resist interpreting anything until every category has a number, because the interesting result is almost always a ratio rather than a single figure. Useful questions: how much of the week went to the work you would describe as your job, how much to the work around it, and how the two split by client or project. Compare the totals against what you would have guessed before the week started, and treat any large gap as the actual finding. The research on prediction is unambiguous that people underestimate their own completion times, so a category that came in far above your guess is the normal result rather than a personal failing.

What to do with the finding

One change, not five. An audit that produces a reorganization usually produces nothing, because everything moves at once and no effect can be attributed to anything. Pick the single largest gap between what you expected and what you measured, make one adjustment aimed at it, and re-measure the same week's categories a month later. If billing is involved, the same record answers a second question about which clients cost more than they look like they cost. And keep the raw export, because the value of a time audit compounds: the second one is only interesting against the first.

In short

  • Record as the day happens. National time-use statistics use a same-day 24-hour diary, not recalled averages.
  • One ordinary week, five to ten stable categories, including the unpaid work.
  • Total everything before interpreting anything. The finding is usually a ratio.
  • Expect the gap between your guess and the measurement to be large; underestimating your own time is well documented.
  • Change one thing, then re-measure the same categories a month later.

Related guides

Sources