Operating cadence: the org's heartbeat

Most organisational dysfunction I've walked into was one missing meeting or one extra one.

That's a smaller claim than it sounds. It doesn't mean meetings fix things. It means the rhythm of an organisation is a real piece of infrastructure, mostly assembled by accident, and it's cheap to change compared to almost anything else you could do.

The accident is usually this: every meeting on the calendar was created in response to a specific failure, by someone reasonable, and none of them were ever deleted when the failure stopped happening. So the calendar is an archaeological record of past problems, and the org is paying rent on all of them forever.

Weekly is the only real cadence

I run one standing rhythm. Weekly.

Weekly is the frequency at which a person can hold the thread. You remember what you said last time. You can course-correct on something before it's compounded. A week is short enough that a problem raised is still a problem, not a post-mortem.

Monthly is where things go to become presentations. By the time a monthly meeting comes around, whatever was true four weeks ago has either resolved itself or metastasised, and either way what gets discussed is a summary of a summary. Everyone prepares for it. That's the tell. When people prepare for a recurring internal meeting, the meeting has stopped being work and started being a performance.

Quarterly is worse, because quarterly is long enough that the meeting has to justify its own existence with ceremony. Two days of planning, a deck per team, a set of commitments that everyone in the room privately knows will be renegotiated in five weeks.

I'm not saying nothing happens at longer intervals. Plenty does. I'm saying it should be written, not met. A monthly board note, written, no surprises in it. A quarterly look at whether the plan still matches the world, done as a document that people read and comment on. The output is the same and the attention cost is a tenth.

The minimum viable calendar at ten engineers

If you have ten engineers, here's the whole thing.

A weekly leadership hour, whoever the four or five people are who make decisions. Not status. Decisions, and specifically the ones that are stuck.

A weekly demo, standing, never cancelled. It's the single highest-leverage recurring event in a small company, and it's covered properly in its own lesson because it does about four jobs at once.

One-to-ones, weekly, thirty minutes, with everyone who reports to you. These are the last thing to cut and the first thing people cut.

A daily standup at the team level if the team wants one, capped at ten minutes, and genuinely optional as a practice rather than optional in theory. Some teams get real value. Some teams are performing attendance at 9:30 for a manager who isn't listening.

That's it. Four things. Everything else is a working session with a start and an end date, which is a completely different object from a standing meeting and should be named differently so nobody confuses the two.

Written updates instead of status

The single biggest cadence upgrade available to most orgs: replace the status meeting with a written note, and hold the line on people actually reading it.

The note is short. What shipped, what's stuck, what I need a decision on, what I'm worried about. Fifteen minutes to write, three to read, and it's searchable in six months when somebody asks when you first raised the thing.

Two failure modes to watch. The first is that the note becomes long and defensive, which happens when people feel judged by it, which is a trust problem and not a format problem. The second is that nobody reads it, and the meeting quietly comes back to life because that's the only way to guarantee information transfer. The fix for the second one is that the decisions get made in response to the note, in writing, visibly. Once people learn that the note is where things actually get decided, they read it.

Written updates also do something a status meeting structurally cannot: they let a person who wasn't there catch up in three minutes, six weeks later. Every status meeting is information that evaporates unless someone takes notes, and the person taking notes is always the person who most needed to be listening.

Killing a meeting without a fight

Don't argue about whether the meeting is valuable. You can't win that argument, because the person defending it will describe the best version of it and you'll be describing the average one.

Cancel it for three weeks instead. Say it's a trial, put the restart date in the calendar, and mean it. Then at the end of three weeks, ask the room what they missed. Sometimes something real broke, and now you know exactly what job the meeting was doing and you can do that job in a cheaper way. Usually nobody missed anything, and the meeting dies without anyone having to be wrong in public.

That last part is why this works. A recurring meeting is almost always somebody's idea, and killing it in a debate makes them the loser of the debate. Killing it with an experiment makes it nobody's fault, and you'll need that person's goodwill for the next thing you want to change.

The audit worth doing once a year

Export your calendar for the last month. Add up the hours in recurring internal meetings. Multiply by the number of people in them.

Then, for each one, write down the specific failure it was created to prevent, and whether that failure is still plausible. Anything you can't answer for, cancel. Anything answerable but stale, cancel. Anything real, keep, and write the answer in the invite so the next person doesn't have to guess.

Most teams doing this the first time find that between a third and a half of the recurring load is protecting against something that stopped being true a year ago. That's not incompetence. It's just that nothing in the normal operation of a company ever tells you a meeting has finished its job.