Look before you instrument
Go and see the actual work
Industrial engineering has a phrase, often translated as 'go and see for yourself', for the discipline of refusing to design improvements without first observing the work in its actual setting. The intuition transfers cleanly to personal knowledge work. Before any new tool, the honest first move is to look at where time has actually gone — not where it felt like it went, not where it was meant to go, but where the digital exhaust says it did. Memory of one's own week is famously unreliable, and the gap between recollection and record tends to be the source of the most useful corrections.
What to pull and how to combine it
A serviceable audit uses four kinds of input drawn from a single recent week. Sent messages from email and any chat tool, filtered to one's own outgoing items. The calendar, attended events only, including the meetings that ran over. Any meeting transcripts a recorder produced. The outputs themselves — documents finished, code committed, decisions written down. Each by itself is partial. Together they form a record dense enough that an assistant asked to summarise patterns can produce something more accurate than the worker's own recollection. The audit is the first useful joint between human memory and machine pattern recognition.
Three questions that produce the answer
With the week assembled, a small set of questions does most of the work. What activity types recur most often, ranked by frequency? Which of these are clearly driving outcomes, and which are filling time? Of the repeating items, which three are the strongest candidates to be removed, automated, or templated first? The three questions are deliberately ordered. Frequency first, because frequency without value is the easiest waste to find. Value second, because high-frequency low-value is the largest available improvement. Candidates third, because by that point the candidates are obvious.
Why the audit must precede tool selection
It is tempting to skip the audit because the workflows feel familiar. The skip is expensive in two ways. First, without the audit, the tool gets pointed at whichever task the worker happens to dislike most that week — which is often not the highest-impact target. Second, the audit acts as a baseline; without it there is no honest way to tell whether the new tool actually shifted anything. A practice that instruments before observing tends to confuse motion with progress. The audit is the cheapest insurance against that confusion.
A 48-hour pause before acting
An audit done well surfaces uncomfortable truths — meetings one cannot defend, reports no one reads, repeating tasks that quietly belong to no one. The temptation is to act immediately. The practical discipline is the opposite: sit with the audit for forty-eight hours before changing anything. The pause exists because the system-one reaction to seeing one's own week tends toward overcorrection, and overcorrection is the second most reliable source of weeks lost. Decisions taken after the pause are typically more honest, narrower, and more likely to survive contact with the next week.
Without checking your calendar, write down the three activities you spent most of last week on. Then check. How close was the recollection to the record?