Before you read another word, write down two numbers. How many hours you think you worked last week. And of those, how many went to work that actually mattered. Hold on to them, because by the end you will see what they are worth, and the honest answer is: not much. Not because you are careless, but because almost nobody can do this from memory.
Why can’t you trust your memory of the week?
Your sense of time is wrong in two opposite directions at once. Going forward, you underestimate how long things take. Psychologists call it the planning fallacy. In the classic study, students predicted finishing a thesis in about twenty-seven days at best and forty-nine at worst; the actual average was around fifty-five, worse than their pessimistic guess.1 It scales: the Sydney Opera House was projected at six years and seven million dollars and took sixteen years and a hundred and two million.2 Expertise does not cure it. Only measurement does.
Going backward, you make the opposite error. You overestimate how much you worked and how much mattered. A 2025 study found people overstated their working time most of the time, reporting roughly an hour and twenty-seven minutes for every genuine hour, an inflation of about forty-five percent.3 The “full day” you remember is padded with quick inbox checks and the ten minutes lost getting back into a task after an interruption.
What does a measured week actually contain?
Three findings, gathered from millions of real working days. First, most of your week is spent talking about work rather than doing it: on average around fifty-seven percent of time goes to communication and only forty-three percent to creating anything.4 Second, focus has been shattered. The average worker is interrupted roughly every two minutes, the average focused session has shrunk to about thirteen minutes, and recovering concentration after an interruption costs around twenty-three minutes.5 Third, despite long hours on devices, the average knowledge worker produces under three hours of genuinely focused output a day.4
If your audit shows this, you have not discovered you are bad at your job. You have discovered you are normal, and normal is precisely the problem this method exists to solve.
How do you run the audit?
Track time as you spend it, not from memory. Tag each block two ways: by value (did this move something forward, or just keep the lights on?) and by fate (could this be eliminated, delegated to AI, or does it genuinely need you?). Do it for a representative week. Be honest about AI blocks especially: some saved you real time, and some were time spent supervising a machine that produced something you then had to redo.6 The audit is how you tell the difference instead of guessing.
What is the point of seeing it clearly?
Not shame. Direction. Most attempts to “get more efficient” fail because they optimise a fiction, the remembered week rather than the real one. Replace the fiction with a map, and the next two moves, shedding and shifting, have something true to work on. You stop fixing the wrong problems. You also stop treating a structural problem, a world built to interrupt you every two minutes, as a personal failing. The map is the start of getting the time back.
Frequently asked questions
How do you audit your working week?
Track your time as you spend it, in short blocks, rather than reconstructing it from memory. Tag each block by value and by whether it could be eliminated or delegated. The point is to replace your remembered week, which is distorted, with a measured one.
Why can’t you just remember how you spent your week?
Because memory is wrong in two directions: you underestimate how long tasks take (the planning fallacy) and overestimate how much you worked and how much of it mattered, often by around 45 percent. Only measurement corrects this.
What does a measured working week actually contain?
On average, more than half the week is communication rather than creation, focus arrives in roughly 13-minute fragments interrupted every couple of minutes, and under three hours a day is genuinely focused output.
About the author
Tom Goodwin is the author of Don’t Work Harder, a book about taking the time AI gives back as time rather than more work. He is a co-founder of GAMEPLAN and writes on productivity, technology, and the economics of the working week.
Footnotes
Footnotes
-
Daniel Kahneman and Amos Tversky, “Intuitive Prediction: Biases and Corrective Procedures” (1979); Daniel Kahneman, Thinking, Fast and Slow (2011). ↩
-
The Sydney Opera House overrun is a standard illustration of the planning fallacy. ↩
-
2025 time-perception study finding people overstated working time by about 45 percent on average; time-diary researchers find similar inflation when measuring confidently claimed long weeks. ↩
-
Workplace-software telemetry (Microsoft and Asana indices, 2021 to 2025): around 57 to 60 percent of time on communication and coordination, with under three hours a day of focused output. Vendor data, read as directional. ↩ ↩2
-
Microsoft Work Trend Index, “Breaking Down the Infinite Workday” (2025): interruptions roughly every two minutes; focused sessions near 13 minutes; about 23 minutes to recover concentration. ↩
-
Upwork Research Institute (2025): 77 percent of generative-AI users said it added to their workload through review and prompt overhead. ↩