Running Meetings That Earn Their Cost
A one-hour meeting with eight people is a full day of engineering capacity, spent. I started doing that arithmetic out loud years ago, while coordinating between an offshore engineering team in Vietnam and a head office in Japan, where every synced hour cost double because someone was always staying late for it. Once you price meetings honestly, you stop treating them as a default and start treating them as a purchase. These are the rules I buy them with.
Decide what the meeting is for, or don't hold it
Every meeting I call answers one sentence before the invite goes out: "at the end, we will have decided ___" or "at the end, everyone will know ___." If I can't finish the sentence, the meeting is actually an email, a document, or a status update that belongs in the tracker. The cross-timezone years made this non-negotiable: anything that could move async, moved async, in written English both offices could re-read. Meetings were reserved for the two things async handles badly, real decisions and real disagreements.
An agenda with time boxes, sent before, not narrated during
The agenda goes out with the invite: items, owner per item, minutes per item. Not because the plan survives contact with the discussion, but because the time boxes make drift visible. When item one has eaten twenty of its ten minutes, everyone can see the trade being made, and it becomes a group decision to continue rather than an accident.
Fewer people, deliberately
The invite list is where meetings bloat. My filter is borrowed from the RACI I already keep: people who will decide or contribute attend, people who only need to know get the notes. Cutting the audience also changes the room's honesty. There are things a team lead will say among four people that they will never say in front of eleven, and those things are usually the reason the meeting was needed.
Materials before, questions during
If a meeting requires context, the context is a document sent ahead, and the meeting starts from questions about it, not a recital of it. Reading a slide deck to people who can read is the most expensive form of narration there is. This habit came directly from working with the Japanese side, whose meeting culture expects preparation as a form of respect, and it turned out to be simple good engineering of everyone's time.
Start on time, end early when you can
Starting five minutes late with eight people burns forty person-minutes to punish the punctual. I start on the minute. And when the decision lands early, the meeting ends early; nothing dilutes future attendance like the memory of an hour that was clearly stretched to fill its booking.
Steer by the agenda, park everything else
Good discussions wander, and some wandering finds real problems. The discipline is not suppressing tangents but pricing them: "this matters, it isn't today's decision, parking it" takes ten seconds and preserves both the topic and the meeting. The parked list gets owners at the end, or it was never really important.
Notes with names and dates, same day
A meeting that produced decisions but no written record produced opinions. The close of every meeting is two minutes of reading back: decided X, A owns Y by Friday, parked Z. Notes go out the same day, short enough to actually be read. In cross-cultural settings this is doubly load-bearing, because a polite nod and an agreement are not the same thing, and the written record is where the difference surfaces cheaply, this week instead of at the milestone.
The test
The test of a good meeting is not whether it felt productive. It is whether, a week later, you can point to something that exists or a decision that held because that hour happened. Run the arithmetic on your own calendar for one week, price every meeting in person-hours, and ask what each one bought. The two or three that fail the test were the easiest budget cut you will find all year.
Nguyễn Hải Nam
Project Management Lead. 16+ years from code to delivery. PMP®. Writing here about project management and engineering.