Meetings that earn their cost

Problem. Meetings pile up until the calendar is more meetings than work, and a growing share of the team's time goes into keeping each other aligned instead of building. At some point the team notices that less is shipping and starts debating which meetings to cut, and the debate is one-sided from the start. The cost of a meeting is concrete, minutes per attendee and dollars per hour, and every productivity newsletter has a worksheet for the math, but the case for keeping a meeting has no worksheet at all: people say it is good for alignment, or good for the team, with no number behind either claim. That is why most meeting audits I have watched fade inside two quarters. The same recurring blocks come back, the same people sit in them, and nobody remembers what was supposed to change.

An audit that holds does two things differently: it measures value as carefully as it measures cost, and it runs on a schedule.

The test. A meeting earns its place when the value it delivers is greater than the cost it imposes. Both numbers can be moved.

Cost is the half that gets a formula. The published meeting-cost calculators all break it into the same three pieces:

Cost ≈ attendees × hourly rate × (direct time + prep + 23 min recovery)

The first piece is direct time on the calendar, and the second is the prep work going into the meeting. The third is the least visible: a meeting in the middle of an engineer's morning breaks their focus, and it takes about twenty-three minutes on average to get back into a hard problem, so a thirty-minute meeting has cost a good deal more than thirty minutes by the time everyone has paid their way back into whatever they were doing. Most writing about meeting cost gets this half right, especially the broken-focus piece.

Value is the harder half, and the one that disappears from every calculator, because there is no clean formula for it. The two components I have learned to defend explicitly are alignment and culture. Alignment is the work of pulling everyone back to the same plan, the job a regular meeting does as a checkpoint, and the longer a project runs, the more easily a team drifts without one. Culture is the work of being in the same room together, and on a remote team that room is where leadership shows up. Other components appear in particular meetings, like trust between teammates, or mentoring when the room mixes seniority.

Run the audit on a schedule

Most of the meeting cuts I have made came from a recurring audit that fires on its own schedule, even when nothing is going wrong. Run it every three to six months, whichever rhythm you can hold.

Waiting for something to go wrong produces the wrong cuts. An incident reveals the meeting somebody complained about, but never the meeting that has quietly stopped earning its cost. A one-shot calendar purge (the kind Shopify made a headline of) gets the easy cuts once and then loses momentum, because the meetings with political defenders come back, and the second purge inside the same company is always weaker than the first. The recurring audit avoids both traps because it fires on its own, and nobody has to get angry enough to call for one.

One of the cuts from our own calendar was a weekly meeting where engineering and QA talked through new production bugs. It came up in an audit and failed the test, because the work could move into a ticket flow, so we killed it and ran the same conversation through tickets and short written updates. The rate at which production bugs got fixed did not change, and nobody ever had to argue against the meeting, because it had simply come up on schedule and failed.

Rework before killing

Killing and reworking are two halves of the same move, and rework comes first: before a meeting dies, see whether it passes the test after you have tried to

• Shorten it. Sixty minutes down to thirty, thirty down to fifteen.
• Drop attendees who do not need to be there.
• Sharpen the agenda.

Fifteen-minute slots are underused: a short slot forces you to cut everything that does not matter, while a sixty-minute slot lets you keep all of it. Most of the meetings I have tried to shrink, I should have tried to shrink earlier.

If rework cannot get the value above the cost, ask the next question: can the work this meeting is doing live in an artifact instead, a ticket flow, a written update circulated async, a status doc? If the answer is yes, kill the meeting and stand the artifact up in the same week, so nothing rots in the gap. If the answer is no, the meeting is doing something a process cannot, and you defend it on those grounds.

Defend the value, fix the spectators

When someone pushes to make a kept meeting shorter or cheaper, giving in is easy: trim the duration, drop a few people, cut the prep. Show the value analysis instead, and if the spare minutes are doing the work, refuse the trim.

The example I keep coming back to is a product-engineering team standup that runs twice a week, forty-five minutes, and has run that way since covid. We have questioned the length ourselves, more than once, because forty-five minutes across that many people is a real cost, and the answer has come back the same each time: the fifteen minutes beyond the textbook thirty are buying connection on a remote team, which is the whole reason the meeting exists. Trimming it would hurt the value side directly, and the cost argument alone does not earn the trim. When someone asks why we keep it that long, I show them the analysis, the same math with the value half filled in.

Inside any meeting of that size there is a spectator problem. A junior engineer who treats the meeting as something to watch is a real cost the calculator misses, and one you can fix, once you separate three causes:

1. They do not feel like a participant, because nobody has invited them to be one. Ask them direct questions and leave a slot in the agenda for their input.
2. They are shy because the room is large. Offer encouragement, often in private first.
3. The spectator habit is bad meeting hygiene. Give direct feedback.

To make any of this last, write participation into the engineering ladder. On our SDE2 rubric, the collaboration line explicitly rates active participation in team and cross-team discussions, and once speaking up earns credit on a rubric instead of costing social risk for nothing, the room behaves differently, and nobody has to relitigate it every quarter.

Defend the focus block

Push all meetings into one window of the day, so the rest of the day belongs to the engineers writing code, who need unbroken hours to do hard work. On my team that window is three to six in the afternoon, which leaves engineers three to four hours of focus on either side of it, depending on whether they work better in the morning or the evening.

The block is under constant pressure:

• New joiners do not know it exists.
• The broader org does not share it.
• Urgent startup work shows up at unpredictable times and wants the next open slot.

Without a deliberate defense, the block falls apart inside a quarter.

Defend it with a rule of thumb rather than a law, because a law breaks the first time something urgent lands in the middle of the afternoon, and a rule that has been broken once in silence stops binding anyone. Every meeting has to go on the calendar, including the quick calls, because you cannot defend a block you cannot see. Once a month, export the team's meetings and look at them. A handful outside the focus window is fine, but if the drift is adding up, say so to the team, and adjust the window on the same rhythm.

Tradeoffs

Running the team mostly over written, asynchronous channels is the right default, and it is what we do most of the time, but it breaks down on long projects. I once tried to run a year-long initiative without any recurring meeting at all, on the theory that a senior team would coordinate itself even while everyone was juggling other priorities. The team lost the thread: people could not remember what we had done since the last quiet stretch or what was supposed to come next. We added a meeting back in as a regular checkpoint, what got done since last time and what is planned until next time, with written updates in between unless something needed a real conversation. The longer the horizon of a piece of work, the more a regular meeting earns its place.

The second tradeoff is duration. Some meetings should be longer than the shortest possible version, because their job is connection or context-setting, and the looseness in the room is doing that work.

Not every meeting can be replaced by an artifact, and some clear their cost even when they are deliberately loose. The deeper failure mode in this whole discipline is letting the cost calculator do the leader's thinking for them. The audit tests the meeting, and sometimes the meeting passes. Our forty-five-minute standup has passed it more than once.