Meeting audit

Harvard Business Review's widely cited analysis of executive calendars found senior leaders spending an average of 23 hours a week in meetings, roughly double what executives reported in the 1960s. That number keeps getting quoted because it's shocking, but the more interesting part is how it accumulated: almost nobody sat down and decided to spend 23 hours a week this way. It happened one recurring invite at a time, each of which was defensible on the day it was created.

That's the structural problem. A meeting is easy to add and socially expensive to remove. Adding one requires a reason; removing one requires arguing with a colleague about whether their meeting is worth your time, which is a conversation most people would rather skip. So recurring meetings accumulate a kind of tenure, and the calendar treats a standup created for a project that shipped last year exactly the same as the decision forum you need on Thursday.

The result is that recurring meetings get treated as equally non-negotiable, when in reality they're the most negotiable thing on the calendar. A one-off meeting costs an hour once. A weekly meeting costs roughly 40 hours a year, every year, until someone deletes it. That asymmetry is exactly backwards from how most people scrutinize their calendar, and it's what makes a periodic audit worth the effort.

Parkinson's Law explains why the meetings that survive also tend to expand: work expands to fill the time available for its completion. A 60-minute recurring slot will reliably generate 60 minutes of discussion, whether or not there were 60 minutes of substance. The corollary is encouraging — cut the same meeting to 30 minutes and, in most cases, the same decisions get made. The meeting was never sized to the agenda; the agenda was sized to the meeting.

Audit on evidence, not impressions. For one month, keep a single line per recurring meeting occurrence recording what came out of it. Not what was discussed — what changed. The three things worth recording are decisions made, commitments assigned to a named person, and information that genuinely needed to be synchronous.

A month is the right window because it covers four instances of a weekly meeting and one instance of a monthly one, which is enough to distinguish a meeting having a slow month from a meeting that produces nothing. Anything shorter and you'll audit on a single bad week; anything longer and you'll abandon the exercise before it finishes.

Three patterns show up almost immediately once the log exists:

• The read-out. Every occurrence produces information transfer and no decisions. This is a written update wearing a calendar invite. Research on distributed teams consistently estimates that a large share of meetings — commonly put around 40% — could be replaced by asynchronous communication, and status read-outs are the bulk of that share.
• The decision forum with three spectators. The meeting produces real decisions, but only three of the nine attendees are involved in making them. The meeting earns its slot; the invite list doesn't.
• The meeting that only works because of one person. Every useful output traces back to one attendee. That's usually a conversation, not a meeting, and it can often become a 1:1 or an async thread.

Eisenhower's matrix sorts work along two axes, urgent and important, into four quadrants that map to four actions: do, schedule, delegate, delete. It's normally applied to tasks, but it works better on recurring meetings, because a recurring meeting is a standing claim on your time and the matrix is a tool for evaluating claims.

The translation is direct. Important and urgent means the meeting drives decisions that can't wait, so it keeps its slot. Important but not urgent means it matters but doesn't need weekly cadence — strategy reviews, planning, career conversations — so it gets scheduled less often rather than cancelled. Urgent but not important is the quadrant that catches status meetings: they feel time-sensitive because they recur, but the underlying information isn't. Neither urgent nor important is the quadrant almost every calendar has at least one of, and it's the easiest win in the whole audit.

The quadrant that repays the most attention is urgent-but-not-important, because it's the one people misclassify. A weekly sync feels urgent every week — there's always something to report — but urgency generated by the meeting's own recurrence isn't real urgency. Ask what would actually go wrong if this week's instance were skipped. If the honest answer is "people would read the update in Slack instead," the meeting is in quadrant three.

An audit that ends in analysis is worse than no audit, because it costs a month and changes nothing. Every recurring meeting gets exactly one of four verdicts, and each verdict has a concrete next action:

1. Keep. The meeting produced decisions or commitments most weeks and the attendee list matches who made them. The only change worth making is trimming the invite list to the people who actually participate.
2. Shorten. Real output, but not enough to fill the slot. Cut 60 to 30, or 30 to 20. Parkinson's Law works in your favour here: the agenda compresses to fit. This is also the lowest-friction verdict socially — nobody objects to getting time back.
3. Merge. Two meetings with substantially overlapping attendees and adjacent subjects. Merging halves the context-switch cost even when total meeting minutes stay the same, which is often the larger saving.
4. Drop. No decisions, no commitments, no information that needed to be synchronous. Replace it with a written update on the same cadence and propose a check-in in six weeks to see whether anything broke. Framing a cancellation as a trial with a review date is what gets it accepted.

Do this quarterly. A meeting audit isn't a one-time cleanup — calendars refill — and a quarterly review is short enough to actually happen and frequent enough that no meeting accumulates more than three months of unearned tenure.

A calendar audit that only counts meeting minutes understates the real bill, because the most expensive parts of a meeting-heavy day happen outside the meetings themselves.

The first hidden cost is fragmentation. Three 30-minute meetings scattered across a day don't cost 90 minutes; they cost 90 minutes plus the two unusable 40-minute gaps between them. Sophie Leroy's research on attention residue at the University of Minnesota found that switching to a new task before the previous one is finished leaves part of your attention stuck on the unfinished one, degrading performance on what comes next. A fragmented day is a day spent paying that switching cost repeatedly, and research on multitasking has put the productivity loss from constant task-switching as high as 40%. When you audit, look at the shape of the day, not just the total: the same meetings batched into one block cost meaningfully less than the same meetings spread out.

The second hidden cost is post-meeting admin. Writing up notes, chasing what was decided, and confirming who owns which follow-up rarely appears on the calendar, which means it's never audited and never protected — it just eats whatever time was left. A meeting whose slot is 30 minutes but which reliably generates 20 minutes of write-up is a 50-minute meeting, and should be audited as one.

Both of these are covered in more depth in protecting focus between meetings, and the tactics for getting through the days themselves are in surviving a day of back-to-back meetings.

The hardest part of step one is that it depends on memory. Sitting down on Friday to reconstruct what four meetings produced this week is exactly the kind of recall that flatters the meetings you enjoyed and penalizes the ones you found boring. An audit built on that is an audit of your impressions.

This is the part that a durable record fixes rather than more discipline. If the meetings were transcribed as they happened, the audit becomes a search rather than a memory exercise: open the transcript for a recurring meeting, look for decisions and named commitments, and note what you find. Earkeep records mic and system audio continuously and transcribes locally on your device, so the record already exists without anyone assigned to take it and without a bot in the call. Meetings are defined retroactively by selecting a span on a timeline, or automatically from a connected calendar, and each transcript is a plain file on disk you can search or hand to an agent — audio is processed only in memory and never written to disk. See meetings after the fact for how that works.

The habit that makes any of this cheaper is closing each meeting with its outputs written down while everyone is still in the room, which is the subject of never leave a meeting empty-handed. Do that consistently and the monthly audit is mostly already written. For the full system this fits into, start from the pillar guide on time management for meeting-heavy roles.

Download Earkeep free for macOS and let the record of what each meeting produced write itself.