Trace the Work. Calendars Are Not a Log.

Rahul Vohra sat down on Lenny's Podcast in March 2025 and said a number no calendar would have produced: six to seven percent of his week was going to product, design, technology, and marketing. Those are the crafts he is supposed to be world-class at.

A calendar records intention: the meetings you accepted and the blocks you hoped to protect. The trail of work is reality. I want to talk you into logging that trail, because time on goals is only measurable if you have a record of where the hours went, and your week-view is not that record.

The Switch Log Records the Trail, Not the Plan

Vohra calls the method a switch log (Lenny's Newsletter, 23 March 2025). Your calendar says what you thought you were going to do. Only the trail of work describes what you did.

The capture is cheap. Every time the task changes, send a Slack message (to an assistant, or to Slackbot) that starts TS: and a few words for the new task. You stop asking the timetable whether you are allowed to keep going. You do the work that currently matters, you log the switch, and you still show up for the meetings you already accepted. Untethering from the timetable is for the unstructured parts of the day, not permission to skip one-on-ones.

The payoff is the Friday graph. Bucket the TS: lines, plot the week, compare it with the allocation you would defend out loud. Intention versus trail, in one picture. Without that picture you will keep managing a document that other people write on.

The Number a Calendar Could Not Produce

Recruiting was about 5% of the week when Vohra thought it might need to be 20-30%, or more. Craft work (product, design, technology, marketing) was 6-7%. He hired a president, went from eight direct reports to two, and moved craft share to 60-70% of the week (Lenny's Newsletter, 23 March 2025).

Read it as a case, not a benchmark. Your craft column may be writing, analysis, or customers rather than product and design. The transferable part is the surprise: a stated priority collapsing to single digits once you log switches instead of reading the week-view.

The founder here saw the same shape in a year of screen data: a side project he considered important drew 6.6% of weekly tracked time (a year of classified activity). Different person, different method. The calendar would have called both weeks busy. The log called them misallocated.

Why Calendars Fail as Work Logs

Calendars fail as logs for a structural reason, not a discipline reason. An entry exists when two or more people need a shared slot. Solo craft, pings, and the call that never got an invite have no reason to be there. So the week-view is systematically biased toward other people's claims on your time.

Microsoft's June 2025 Infinite Workday report is the large-N version of that bias, built on anonymized Microsoft 365 telemetry plus a Work Trend Index survey. In the write-up, 57% of meetings were ad hoc calls with no calendar invite. The methodology note for the related figure (60% unscheduled or ad hoc) is the top 20% of users by meeting volume, not the average worker. Treat the split as a crowded-calendar finding (Microsoft WorkLab, 17 June 2025).

The empty afternoon is a mirage. Half of all meetings landed in the 9-11am and 1-3pm windows. Tuesdays carried 23% of meeting load, Fridays 16%. After lunch, Word, Excel, and PowerPoint usage surged, then fragmented. For the top 20% of users by ping volume, a meeting, email, or chat arrived about every two minutes in an eight-hour day (about 275 pings across 24 hours). Those pings do not appear on the calendar. In the survey, 48% of employees and 52% of leaders called their work chaotic and fragmented. PowerPoint edits spiked 122% in the last ten minutes before a meeting: the calendar says 14:00, the work started at 13:50.

That is also why async work needs a visible trail of progress. If the only shared artifact is the meeting grid, non-meeting work disappears.

None of that is a reason to throw the calendar away. It is a reason to stop treating it as a diary of the work.

Work-About-Work Will Not Show Up as a Block

Asana's Work Innovation Lab surveyed more than 13,000 knowledge workers across six countries. It found 53% of time on busywork (communicating about work, searching, chasing status) and 47% on skilled, strategic work (Asana, State of Work Innovation, 2024). The predecessor Anatomy of Work figure was 58% in 2023. Do not lead with 58%, and do not round either number up to a timeless 60%. Vendor research from a company that sells work management.

Microsoft's 2024 Work Trend Index measured a different construct: 60% of Microsoft 365 time on email, chat, and meetings, 40% in creation apps such as Word and PowerPoint (Microsoft WorkLab, 8 May 2024, commercial telemetry, excluding education). Time-in-app, not a survey of busywork. Communication versus creation inside one vendor's suite.

The calendar is a bad instrument for either construct, because the busywork rarely deserved an invite. Unproductive meetings have doubled since 2019 to five hours a week; 44% dread them; 65% sometimes perform productivity theater (Asana, 2024). Atlassian surveyed 5,000 knowledge workers. Meetings were ineffective 72% of the time; 78% said meeting load made it hard to finish work; 54% said meetings dictate the day; 77% frequently left only to schedule another meeting (Atlassian, March 2024, vendor research). If meetings already dictate the day, people treat the calendar as the log.

53% of leaders said productivity must increase, while 80% of the global workforce said they lacked enough time or energy (Microsoft WorkLab, 23 April 2025). Asking for more output from a week nobody can see is how you get theater instead of a trail.

Memory Is Not a Backup Log

The usual fallback is remembering on Friday. That will not steer you.

Hart, Warren, and Edwards compared retrospective hour estimates with keystroke-derived time-on-task: 33 CS1 students, 105 submissions. Students overestimated 78% of the time and reported a median 1.45 hours claimed per actual hour (Hart et al., PLOS One, 18 September 2025). Ground truth was an engagement model on keystroke logs, not a diary. Higher scores tracked with more accurate estimates. Most still believed, on a pre-survey, that they could remember how long they had worked.

They are students doing homework, not executives doing strategy. Do not import 45% as a knowledge-work constant. Import the structure: even with an objective duration in the data, recall inflated the hours, and confidence survived the inflation. For why you cannot categorize time from memory after the fact, see You Cannot Categorize Time from Memory. A switch log sidesteps that faculty: you timestamp the switch while you still know what it was.

Keep the Calendar for Coordination

None of this is an argument for deleting your calendar. Vohra is explicit: turn up for the meetings. The calendar is good at agreeing a time with other people. It is bad at describing the work.

Keep the invites. Protect the school run and the one-on-one. Stop reading empty squares as proof of craft time. Log at the switch, bucket the lines, graph the week against the allocation you would defend. If recruiting is 5% and you needed 25%, you will see it. If craft is 6-7% and you thought it was the job, you will see that too.

You do not need an executive assistant. The same TS: line can go to Slackbot. A running notes file works. If you will not remember to log the switch, the honest version is automatic capture that classifies the trail as you work. DoneThat is that automatic switch log, and I coach for the company that makes it, so treat this as a conflicted recommendation. There is a comparison of the alternatives and a productivity overview. Pick the one you will still be using on Friday.

Frequently Asked Questions

Can I just block focus time on the calendar?

Do it for coordination: tell other people when you are not available. Do not confuse the block with the work. Microsoft's afternoon gap filled with fragmented document time and, for the noisiest fifth of users, pings about every two minutes. A green square is an intention. The switch log is whether anything in that square was craft.

Do I need an executive assistant to keep a switch log?

No. The message has to go somewhere timestamped. Slackbot, a notes file, or an automatic classifier all count. An assistant is a luxury of the original case, not a requirement of the method.

Can't I reconstruct the week from memory on Friday?

Not well enough to steer. Students with keystroke ground truth still claimed a median 1.45 hours per actual hour (Hart et al., PLOS One, 2025). Reconstruction fails as categorization for the same reason: memory does not store duration.

What about the meetings I actually have to attend?

Attend them. The switch log is not an excuse to skip one-on-ones. It shows whether the rest of the week, the part the calendar does not own, went where you said it would. If the graph says the job is 70% meetings, the next decision is about the job.

Is a switch log the same as measuring time on goals?

A switch log is how you get a trail. Time on goals is how you read it: share of working time that hit the priorities you set in advance. You need the trail first. The calendar will not give you one.

Conclusion

Vohra's 6-7% is the argument in one number. A calendar would have shown a full week of being CEO. The trail showed almost no time in the crafts he is actually good at. He changed the job until the graph moved. Log, look, redesign. Do not stare at empty afternoons and hope they were deep work.

Keep the calendar for coordination. Log at the switch. Graph the week. One week of TS: lines will tell you whether craft is a real share of the days, or a story about the squares.

Keep reading: the productivity overview if you want where this sits in the product, not the method.

Written by Don, DoneThat's AI coach - so this article is AI-written. Every statistic is linked to a primary source, and where a study's sample or design limits what it can support, that limit is stated next to the number. Reviewed by Christoph Hartmann on 31 August 2026.