Journal / Working hours
Meetings that earn their slot
A meeting costs the number of people in the room multiplied by its length, plus the fragments of time on either side that nobody counts.
In short
- Cluster meetings; scattered slots destroy the gaps between them.
- No decision-maker present means no meeting.
- Close with decision, owner and date — then send those lines.
The fragments nobody counts
A thirty-minute meeting at eleven does not cost thirty minutes. It costs the ten minutes before, which are too short to start anything, and often ten after, spent re-reading where you were. For work that needs a running start, a single mid-morning meeting can remove the entire morning.
This is why two meetings back to back are usually cheaper than the same two spread across a day, even though the calendar shows an identical total. Clustering is the least controversial improvement available to most teams, and it requires no policy — only a willingness to move a slot.
Three questions before accepting
Is there a decision to make, or is this a broadcast? Broadcasts are better as writing, because writing can be skimmed at a personal pace and searched later.
Who actually decides? If that person is not attending, the meeting produces opinions and a second meeting.
What is the smallest group that can decide it? Every optional attendee is a real cost paid to avoid a small social discomfort.
Write the note first
The strongest version of a meeting starts with a short document circulated beforehand: the problem, the options, the recommendation. Reading it takes five minutes and replaces the twenty minutes normally spent bringing everyone to the same starting point.
It also has a filtering effect. Roughly a third of meetings die while the note is being written, because the act of writing the recommendation reveals that the decision is obvious or that someone can simply make it alone.
Ending on time as a habit
Ending five minutes early is more useful than starting five minutes late is harmful, and it is entirely within the organiser's control. Close by restating the decision, the owner and the date — three sentences, out loud, while everyone is still present.
Send those three sentences afterwards in the same words. Most of the confusion attributed to bad meetings is actually the absence of that short written trace: everyone leaves agreeing, and disagrees a week later about what they agreed to.
The recurring meeting problem
Standing meetings are the ones worth auditing, because they were justified once and have been renewing themselves ever since. A weekly sync of eight people costs roughly four hundred hours a year, and almost nobody has ever seen that number written down.
Audit them twice a year with a blunt experiment: cancel the meeting for three weeks and see what breaks. Usually one thing breaks, and it turns out to be a single handover that a short written update covers. Occasionally nothing breaks at all, which is its own answer.
When a recurring meeting survives the audit, shorten it rather than defending it. Twenty-five minutes instead of thirty, forty-five instead of an hour — the work compresses to fit, and the leftover minutes are the gap that lets the next block of real work start.
The half-hour default is arbitrary
Calendar software offers thirty minutes because a grid needs a unit, not because decisions take half an hour. Left alone, the default expands work to fill it: the twelve-minute conversation becomes a thirty-minute one, complete with a warm-up and a wandering ending.
Setting fifteen or twenty as your own default changes the texture of a week immediately. The conversations that genuinely need longer will announce themselves, and you can extend those deliberately rather than paying the tax on all of them.
The other lever is the number of people. A discussion of four is a discussion; a discussion of nine is a broadcast with occasional interruptions. If nine people must know the outcome, the outcome can be written down — that is what the three closing sentences are for.
Async is not automatically better
The current fashion is to replace meetings with written threads, and threads have their own failure mode: a decision that would have taken eleven minutes in a room takes four days, twenty-eight messages and two misunderstandings about tone.
The distinction that holds up in practice is between converging and diverging work. Diverging work — gathering options, collecting objections, sharing context — is better written, because people can contribute at their own pace and the record survives. Converging work — choosing between two options when both have advocates — is faster live, because the objection can be answered in the same breath it is raised.
So the useful pattern is written first, live at the end: circulate the options, collect the objections in the document, then hold a short call whose only agenda is choosing. That call rarely needs half an hour.
The Thursday letter
One short piece a week — usually a method someone on the desk tried and either kept or dropped. No attachments, no course, no schedule creep.
Also on the desk
