ManageTime

Article

Where your working time actually goes: running a time audit on yourself

A time audit means recording what you actually worked on as the day runs, then reading the log against the account you would have given from memory. Write it down rather than reconstructing it later: in a meta-analysis of 138 monitoring experiments, the effect on goal attainment was larger when progress was physically recorded. This article covers what to write in the log, how granular to make it, how long to keep it up before the numbers mean anything, and what the evidence behind watching your own week actually supports.

How do you do a time audit at work?

Three columns, kept for a working week. A time audit is a written record of what you did while you were doing it: when you started, what the task was, and roughly how long it ran before something else took over. You need nothing more elaborate than that, and the elaborate versions are the ones people abandon by Wednesday. If you also want the log to test your estimates, write down what you expected a task to take before you start it.

What you are after is not the total at the bottom. It is the distance between the log and the week you thought you had.

Granularity is the only real setup decision. Rounding to the nearest fifteen minutes is enough for most work and cheap enough to keep up, while logging every switch tells you far more about an interrupted day and costs more attention than that day may have spare. Pick the coarser one if you are unsure. A fifteen-minute log that survives five days beats a minute-by-minute one that stops on Tuesday afternoon.

Write it as you go, though, rather than reconstructing the day at six o'clock. The nearest hard evidence for that comes from a question next door to this one. Harkin and colleagues pooled 138 experiments covering 19,951 people in Psychological Bulletin in 2016, each of them randomly assigning people to something that prompted them to monitor their progress toward a goal. Prompted monitoring produced a modest effect on whether the goal was reached, about 0.4 of a standard deviation, and that effect was larger when the information was physically recorded and when it was reported or made public.

Those studies were not time audits. They prompted people to monitor progress toward goals of many kinds, and none of them watched somebody log an afternoon at a desk. What carries across is the moderator rather than the number: a record you write down does more for you than a record you carry in your head. That is the same argument the Pomodoro Technique makes for marking every finished sprint.

One more column earns its place, and it is the one most people leave out: what pulled you off the task. A log of the work alone tells you where the hours landed. A log that names the interruption tells you who or what moved them, which is the half you can do something about. Keep the entries honest where they are unflattering, because a tidied log measures your intentions all over again, and getting away from your intentions was the point.

How long should you log before the numbers mean anything?

Longer than a day, and shorter than you are probably imagining. No study fixes the length of a useful time audit, so anyone quoting you a number is quoting a preference. What the research does show is where routines like this break, and it is not accuracy. It is keeping them up.

The one workplace trial aimed squarely at this reports its own retention as a finding. Young and colleagues ran a pilot randomised controlled trial in the International Journal of Applied Positive Psychology in 2025, testing a Reflective Time Journal and Time Awareness Coaching against a control group among hybrid workers. Sixty-seven people were randomised across the three arms. Twenty-two finished, a retention rate of 32.8 percent, and the paper puts that feasibility problem in its own abstract rather than burying it.

Read that as a design constraint on your own log. Two thirds of a volunteer sample, people who had signed up for a study about their own time, did not reach the end of a short intervention. You are not more disciplined than a volunteer.

So make the commitment small enough that finishing it is realistic: one ordinary working week, five days, at a granularity you can still manage at four o'clock on a bad Thursday.

The other recent workplace test points the same way on length. Demerouti's self-training study in the Journal of Business and Psychology in 2025 ran its reflection exercises and daily assignments over six consecutive workdays, with 39 people in the intervention group and 46 in a control group. Six days. That is the scale at which these routines have actually been tried on working people, and neither trial is large enough to carry more weight than that.

One week is a sample rather than a portrait, which matters most if your weeks differ from each other. The repair is not a longer log. Run a second short one a month later, in a week of a different shape, and read it beside the first instead of averaging the two. This site puts two honest weeks on any new method, and an audit is the cheapest version of that test, because you are not changing how you work yet. Change nothing while you log, or what you measured is the logging.

What do you do with a time audit once you have the numbers?

Most audits die in the spreadsheet. The log gets totalled, the totals get a wince, and the following week looks like the one before it, because a total on its own names nothing to move.

Three questions turn the log into a decision, and between them they cost less than the logging did.

Start with the estimates. Take the tasks you guessed at in advance and put your guess beside the logged time, and the multiplier that turns up is usually steady per person and per kind of work. That makes it correctable arithmetic rather than a character flaw. It is also the input Time Blocking needs and rarely gets: blocks sized by what the work took last time, not by what you hoped it would take.

Then ask which share of those hours produced the result you are actually paid for. The Pareto Principle is this site's name for that question, and the 80/20 shorthand attached to it is a heuristic rather than a measured ratio, so treat it as a prompt and not a target. Mark each logged block as work that moved something you own, work somebody else needed from you, or upkeep. The proportions tend to be the surprise, not the totals.

The third question is the awkward one: what does the log say you should stop doing, and who has to agree to it? An audit measures hours. It says nothing about whether an hour was worth spending, and that judgment is a ranking question the prioritization methods answer directly. Taking the log into the conversation is what makes a ranking stick, because a request that reads as small in a message reads differently beside the hours your own week says it takes.

Both halves of the research point at that last move. The monitoring meta-analysis found larger effects when progress was reported or made public, and in Demerouti's hybrid-work trial the gains ran through what people changed, work-home facilitation and relaxation, which in turn explained the improvement in goal attainment. Watching came before the change; it was not the change. An audit nobody acts on is a week of admin with a chart at the end of it.

How does your working style change the answer?

None of these studies sorted people by how they organise a week, so what follows is this site's reading rather than a research finding. The four working styles describe how you already operate when nobody is watching, and an audit is one of the places that difference shows on day one. Take the profile that sounds like your ordinary week.

If you build the week in advance, the Architect's pattern, your risk is building the instrument instead of using it. A category scheme with eleven labels is fine on Monday and gone by Thursday, and the audit that reaches Friday is the one that gets read. Three categories, five days, then set the log beside the plan you had already written. That comparison is the entire audit for you. Nobody else can run it.

Burst workers, closest to the Sprinter, already own half the apparatus. A finished Pomodoro sprint is a logged unit, so counting sprints per task is a time audit at 25-minute resolution with no extra bookkeeping at all. What that resolution hides is the gap between sprints, which on a bad day is most of the afternoon. Log the starts as well as the finishes for one week.

If you think in outcomes rather than tasks, nearer the Visionary, do not log by activity. Log what each block was in service of and accept coarse timing, because the number you want is the share of the week that reached the two or three things you actually own. A day that reads as fully booked and returns nothing to those is the finding, and a log sorted by task type will never show it to you.

A week that never repeats, which is the Improviser's situation, breaks the averaging most audits rest on. Do not go looking for a representative week. You do not have one. Count switches and name what caused each, run the log across two unlike weeks, and compare their shape rather than their hours. The request that turns up in both, whatever else changed around it, is the one worth raising with somebody.

One week, three columns, starting tomorrow morning.

Write the interruption beside the task, leave the entries unedited, and read the whole thing on Friday against the week you thought you had. If you are not sure which of these styles is yours, the working style assessment takes a few minutes, and the planning pillar covers what to do with the hours once you know where they went.

Frequently asked questions

How do you do a time audit at work?

Log what you are working on as the day runs, in three columns: the start time, the task, and what pulled you off it. Round to the nearest fifteen minutes rather than chasing every minute, keep it for one ordinary working week, and do not tidy the entries afterwards. On Friday, read the log against the account of the week you would have given from memory. The gap between the two is what you came for.

How long should you log your time before the results mean anything?

One ordinary working week is enough to start with, and the risk in running longer is that you stop before you finish. No study fixes a length. What the research shows is where these routines fail: in a 2025 pilot trial of a reflective time journal for hybrid workers, 67 people were randomised and 22 completed, a retention rate of 32.8 percent. If your weeks differ a lot, run a second short log a month later rather than extending the first.

What should you actually write down in a time log?

The start time, the task in your own words, and the thing that ended it, which is the column most people leave out. Totals tell you where the hours landed, and the interruption column tells you who moved them. Write it as you go rather than at six o'clock: across 138 monitoring experiments, the effect on goal attainment was larger when the information was physically recorded.

Is a time audit worth it when no two of your days look alike?

Yes, but stop hunting for an average day: on an unpredictable week the useful output is the pattern of interruptions rather than the totals. Count switches and write down what caused each one. Then run a second week of a different shape and compare the two instead of merging them. The request that turns up in both, whatever else changed, is the one worth a conversation.

Sources