The most useful thing I do for a meeting usually happens before it exists: I work out whether it needs to. More often than I’d like to admit, the honest answer is no, and the single most effective meeting habit I have is the one that quietly takes the meeting off the calendar.
The usual version of this advice is a list of habits for running great meetings. I’ll get to the habits, because a few of them are real. But they’re the wrong place to start, because most meetings don’t have a habits problem. They have an existence problem, and no amount of crisp facilitation fixes a meeting that shouldn’t be happening.
What the Meeting Is Actually Standing In For
When I look at a recurring meeting that everyone quietly resents, it’s almost always covering for one of two things.
The first is status that should have been written down. The fifteen-person sync where two people talk and the rest wait for their turn or zone out. Nothing in it needed to be live. It’s a status report being read aloud, badly, to people who’d have absorbed it faster in thirty seconds of text on their own schedule. The meeting exists because writing the update feels like work and scheduling a recurring call feels like progress.
The second is more interesting, and it’s the one worth fixing: a decision nobody is empowered to make. The technical debate that loops for the third week running isn’t looping because the topic is hard. It’s looping because no one in the room has the authority to end it, so the meeting becomes a way to diffuse the decision across a group instead of placing it on a person. A committee is what you convene when you want a decision to have no author. That feels safe and it is enormously slow.
So before I think about agendas or timeboxes, I ask what the meeting is a symptom of. If it’s information, that’s an async problem. If it’s a decision with no owner, that’s an ownership problem. Neither is solved by meeting better.
Default to Async, and Mean It
The fix for the first category is to make written the default and live the exception.
A status sync becomes a short written update people read when it suits them. A design debate becomes a document with the proposal, the alternatives, and the trade-offs, where people comment in line and the disagreement is visible and durable instead of evaporating the moment the call ends. Most of what fills calendars is information transfer, and information transfer is the thing writing does better than talking: it’s searchable, it’s skimmable, it doesn’t require eight people to be free at the same instant, and it forces the author to actually think the point through rather than wave at it on a slide.
The bar I hold for a synchronous meeting is narrow. It’s worth meeting when there’s genuine high-bandwidth disagreement that needs fast back-and-forth, or a decision that has an owner and just needs the inputs gathered quickly in one place. Nearly everything else is a document that hasn’t been written yet.
When You Do Meet, the Slot Is the Enemy
The few meetings that survive that filter still have a problem: the calendar block itself. Give a discussion an hour and it takes an hour, give it thirty minutes and it somehow still resolves. That’s just Parkinson’s Law showing up on your calendar, and meetings are where I see it most reliably. A ten-minute decision will expand to fill whatever slot you booked for it.
So the guardrails that are worth keeping are the ones that fight that expansion, not the ones that decorate the meeting:
- Name the decision-maker first. This is the one that actually matters. Before the meeting, it should be clear who has final say. Without that, a technical debate has no natural end and you’re back to looping. With it, the meeting’s job is to inform a decision, not to manufacture consensus.
- Keep it small. Six to eight people is about the ceiling for a conversation rather than a broadcast. Anyone who’s there “just in case” is usually a sign you’re not sure what the meeting is for. They can read the recap.
- State a goal, not a topic. “Discuss the API” is a topic and it will sprawl. “Decide the API’s pagination approach” is a goal, and you know when you’ve hit it. If I can’t write the goal down, the meeting isn’t ready and I cancel it.
- Stop at the clock. If a new thread opens as time runs out, the right move is almost always to spin it into its own decision with its own owner, not to let a thirty-minute meeting bleed to ninety. Ending on time also tells people their calendar is real.
- Write down what was decided. Decided, owner, due date, in text, after. A decision that lives only in eight people’s memories isn’t a decision, it’s a future argument about what everyone thought they heard.
Notice that most of these are about ending things, whether that’s the debate, the meeting, or the ambiguity. That’s the actual skill.
Where Async-First Fails
Async-first is a position I hold, not a religion, and it has real failure modes.
Writing has a tax. It only works if the team actually writes and actually reads, and a culture that defaults to async without that discipline just trades the silent meeting for the ignored document. Async is also genuinely slower for some things. A contested decision with a lot of nuance can die in a comment thread that a fifteen-minute call would have settled cleanly, and the dogmatic “this could have been an email” reflex sometimes turns a quick resolution into a three-day asynchronous slog. Some work is better live, full stop: early, ambiguous exploration where you’re thinking out loud together, anything with interpersonal friction, and the relationship-building that a distributed team doesn’t get for free.
The goal isn’t zero meetings. The meeting should be the deliberate exception you reach for when async genuinely can’t do the job, not the default container you pour all coordination into because scheduling one is easy.
The Question to Ask Before Scheduling
The habits aren’t wrong. They’re just downstream of the question that actually matters: what decision does this meeting exist to make, and who owns it? If you can answer that, the meeting is usually short and occasionally unnecessary. If you can’t, you don’t have a meeting to improve. You have an ownership problem wearing a calendar invite.
So before you schedule the next one, try answering that question in a single sentence. If you can’t, the most useful thing you can do for everyone’s week might be to not send the invite.