Case: the 90-minute stand-up

A long meeting feels like one block of time, so it's hard to argue about. Split it into segments and log each one: minutes, who attended, who was actually needed, whether it could have been written, decisions made, action items. Then the meeting can be discussed with numbers instead of impressions.

Klasna's stand-up was logged for ten working days in five segments: status round, blockers, problem-solving, announcements, and waiting and connection. It ran 900 minutes, 90 a day on average. The segments that could have been a written update were the status round (298 minutes) and announcements (89): 387 / 900 = 43.00% of all stand-up minutes.

That number answers the worry "if I cut it, won't people lose track?" Almost half the meeting was people hearing information they could have read in a couple of minutes, at a time that suited their shift and time zone.

The calendar shows a meeting's length; the team pays its length times the number of people in it. Count person-time: minutes × attendees for every segment, added up.

At Klasna every stand-up segment has seven attendees: the lead and six coordinators. Two weeks of stand-ups: 900 minutes × 7 = 6,300 person-minutes = 105 person-hours, so 52.5 person-hours a week. That's 7.5 hours a week for each person, 18.75% of a 40-hour week, spent in one meeting during the exam peak, while the queue piles up.

Person-time also exposes waste that minutes hide. Problem-solving took 335 minutes (37.22% of the stand-up) and usually needed two people, three at most; the other four or five listened. Across the whole stand-up, 2,526 of 6,300 person-minutes (40.10%) went to people the segment didn't need. Iryna's line is the right test for any peak-season meeting: that number had better be small.

Time spent tells you what a meeting costs; outcomes tell you what it's worth. Two outcome measures are easy to log: decisions per hour, and the share of action items with an owner and a date.

Klasna's stand-up ran 900 minutes in two weeks and made 17 decisions: 17 / 15 ≈ 1.13 an hour. The weekly team call makes 3.33 an hour, the refund sync 4. Inside the stand-up the picture is sharper still: 14 of the 17 decisions come from the blockers segment, which takes only 107 minutes (11.89%). The rest of the meeting produces very little.

Action items tell the same story. The stand-up ended with 33 agreed steps, only 10 of them (30.30%) with an owner and a date; all 12 from the other two meetings had both. A meeting that produces few decisions and vague steps isn't a forum for decisions, however long it runs. Find the segment that does produce them, and protect that.

Put the segments side by side and the stand-up's real shape appears. Blockers, the part that genuinely needs a meeting, take 107 of 900 minutes (11.89%) and produce 14 of the 17 decisions. Status and announcements (43.00%) could be written. Problem-solving (37.22%) needs two or three people, not seven. The rest is waiting and connection problems.

So the honest reading: the useful core is the blockers segment; status and announcements belong in writing; problem-solving belongs to the few who need it. The stand-up makes 1.13 decisions an hour against 3.33 for the team call, so it isn't the team's main decision forum and shouldn't get more time. The team call shouldn't be cancelled either: it does its job.

And the data points at the meeting's design, not at people. "Some coordinators talk too much" is a judgement about individuals that the audit doesn't support; the long segments are long because of what the meeting is asked to carry.

Moving the status round to writing removes the biggest block of could-be-written minutes: 298 of the stand-up's 900 in two weeks. But a written status only works if it's easy to write and easy to scan.

Four design choices make it work. Short: three lines per person (for example: yesterday, today, anything blocking). One place: a single team thread, not direct messages to the lead, so everyone sees everyone. A fixed time that suits both time zones: Olha suggests 8:45 Kyiv, so the two coordinators two hours behind read it with their first coffee and arrive at the stand-up already informed. An outage rule: whoever has no power posts one line from their phone when they can, and their buddy covers the queue. The thread can wait; families can't.

Avoid the two failure modes: reading everyone's status aloud (the old status round with one speaker) and making status optional (gaps appear, and the stand-up drifts back to asking everyone).

Once status is written, the stand-up can shrink to what genuinely needs everyone at once. A 15-minute stand-up for the same seven people, five days a week, costs 15 × 7 × 5 = 525 person-minutes, 8.75 person-hours a week, instead of 52.5: 43.75 hours back every week of the peak.

What stays in: blockers, read from the written status, and who needs whom today. What moves out: problem-solving, which goes to a short follow-up call for just the two or three people involved, booked before the stand-up ends. Every stand-up ends with action items with one owner and a date, posted in the thread.

And the time itself. 9:00 Kyiv is 7:00 for two coordinators. Moving the stand-up to 10:00 Kyiv (8:00 for them, with the morning queue already covered) or rotating the slot spreads the early hour fairly. Ask the team before deciding: it's their morning too, and Olha wants the draft in the thread for a day first.

A good redesign answers each finding of the audit with one change. Status and announcements (43.00% of minutes): replace the status round with a three-line written status in one thread before the stand-up. People not needed (40.10% of person-minutes): move problem-solving to short follow-up calls for the two or three people involved. Vague endings (only 30.30% of action items with an owner and a date): end every stand-up with action items that have one owner and a date, posted in the thread.

Changes that don't fit the data: banning cameras saves no minutes; a ranking of who spoke least is surveillance that punishes quiet people and fixes nothing. Write the redesign for the team with the numbers, post the draft for a day as Olha asks, and set a review date after two weeks to check the new numbers.

How it fits together: segment the meeting, price it in person-time, judge it by decisions and owned action items, then keep the core and move the rest to writing and follow-ups. Next, module 6: in June the peak ends, and the pressure shows up between people.