Fix the Meeting or Kill It: A Redesign Guide

A meeting you dislike is not evidence of anything. A meeting that fails one of four specific tests is, and each failure has exactly one sensible repair. Run the four questions, take the repair they point at, and if the meeting fails the fourth one, stop repairing it and end it.

Most writing about effective meetings is about how to run one well once it exists — the agenda, the timekeeping, the facilitation. This is the prior question: whether the thing should exist in its current form at all, and what to do in each case.

The reason to have a procedure for it is political rather than analytical. You are rarely the person who can simply cancel something. What you can do is arrive at a conversation with a named defect and a specific edit, which is a different kind of request than “can we talk about whether we need this”. One is a proposal. The other is a complaint with a calendar invite attached.

Four questions that sort effective meetings from the rest

Ask them in this order, about one specific recurring meeting. Stop at the first failure; the repairs are not cumulative and stacking them is how a meeting gets redesigned into something worse.

Four failure modes: orphaned, broadcast, miscast, obsolete. Three of them are edits. One is an ending.

The order matters. An orphaned meeting looks like a broadcast meeting, because nobody is driving it so nothing gets decided in it — and if you fix the symptom by converting it to a written update, you have preserved a ritual nobody wanted in a new format. Ownership first.

Why this beats “is this meeting useful”

Because everyone answers yes. Usefulness is not a falsifiable property of a meeting; somebody always got something out of the hour, and that person is usually the one who called it. The four questions are answerable by looking at artifacts — the invite, the last three sets of notes, the decision log if there is one. You can answer them without a debate, and you can show your work.

If you have not yet done the counting that tells you which meeting to point this at, the calendar audit is the procedure for that, and the six meeting types sorts your week into categories before you start.

Question 1: does it have an owner who would notice?

Not a chair. Not the person whose name is on the invite because they created it eighteen months ago and have since changed teams. An owner is someone who, if the meeting quietly did not happen this week, would notice and care.

How to test it without asking anyone. Look at who sent the last agenda. Look at who follows up when nothing gets decided. Look at whether the invite’s organiser is still in the same role. An orphaned meeting has a very specific signature: the invite belongs to someone who has stopped attending, and it runs anyway because it is in everyone’s calendar.

The repair. Either someone takes it, on the record, or it ends. Those are the only two outcomes, and the useful thing about asking is that the question itself often resolves it. Nobody wants to be the owner of a meeting they have privately thought was pointless for a year.

The message is short and it goes to the current organiser, or to the most senior regular attendee if the organiser has left:

“Quick one on the Thursday sync — the invite is still under Priya’s name and she moved to Platform in March. Is this yours now, or should we let it lapse? Happy either way, just want to know who to send the agenda to.”

Three things that message does. It names the specific evidence rather than the feeling. It offers ending as a neutral option rather than a proposal you are advocating. And it makes the status quo cost something — someone now has to claim it — which is the only way a default ever changes.

The risk is real and worth naming: you are the person who raised it, and if the answer is “yes, it’s mine, it’s important”, you have spent a small amount of capital and gained nothing. That is a genuine cost and it is why you pick one meeting rather than auditing all of them out loud.

Question 2: is there a decision, or only information?

A meeting earns synchronous time when something has to be resolved in the room — a decision between options, a disagreement that needs both parties present, a piece of work that only happens if people do it together.

Information moving from one person to several is not that. It is a broadcast, and the hour is being spent on delivery rather than on anything that required delivery to be simultaneous.

How to test it. Read the last three sets of notes. If you can convert them to a document with no loss — if nothing in them is a record of a live exchange — it is a broadcast. Another version of the same test: could someone who missed it catch up entirely by reading? If yes, then everyone who attended could also have read.

The repair is conversion, not cancellation, and the distinction is what makes it acceptable. You are not removing the information. You are moving it to a form that costs fifteen minutes of one person’s writing instead of an hour of six people’s attention. The status update that does this job — what it contains, how it gets read, why most of them fail — is what a written status update has to do to replace the meeting, and it is a genuinely different skill from running the meeting was.

The conversion has a failure mode with a name, and you should raise it before someone else does: the update that becomes a second meeting. The written version goes out, nobody reads it, and within three weeks there is a fifteen-minute call “to go through the update”. Now there are two artifacts and one of them is a meeting again.

The defence is to set the rule at the start rather than after it breaks: the written update is the record, questions go in the thread, and if a question genuinely needs a conversation, that conversation is booked separately with only the people it concerns. A standing slot “in case” is the meeting coming back.

Keep the decision half. Many broadcast meetings have ten minutes of real decision buried in fifty minutes of reporting. The conversion is not “delete this” — it is “the reporting becomes a document, and the ten minutes becomes a fifteen-minute meeting with four people instead of an hour with nine.” That version gets agreed far more often than a deletion, because it concedes that the meeting was doing something.

Question 3: would these people opt in?

Imagine the invite went out tomorrow with attendance genuinely optional and no social cost to declining. Who comes?

If the honest answer is “about half of them”, the meeting is not broken — the attendance is. A miscast meeting has the right purpose and the wrong room, and the fix is the cheapest of the four because it does not touch the meeting’s existence, its owner, or its slot.

The full procedure for who belongs in a room and how to get yourself out of one you do not belong in is the attendance question. The short version for the redesign: people are in meetings because they were in them last quarter, and nobody audits an invite list, ever.

The repair, in one move. Split the list into people who have spoken in the last three sessions and people who have not, then ask the second group directly rather than removing them:

“Rob — you’re on the Tuesday review but I don’t think you’ve needed to speak in a while. Do you want to stay on it, or would notes be enough? No wrong answer, and you’re welcome either way.”

Ask rather than remove. Removing someone from a recurring invite reads as a signal about their standing, and it will be read that way whatever you meant, especially by anyone junior or anyone new. Asking makes it a preference. Most people take the exit gratefully, and the ones who want to stay tell you a real reason you did not know about.

Question 4: has the thing it was for changed?

Every recurring meeting was created for a reason that existed at the time: a launch, a reorganisation, a system that kept breaking, a relationship that needed rebuilding. The meeting outlives the reason, routinely, because there is no mechanism anywhere in a normal company for noticing that a reason has expired.

How to test it. Find out why it started. Ask the longest-serving attendee, or read back far enough in the notes. Then ask whether that thing is still true. This is the only one of the four questions you cannot answer from the last three weeks of artifacts, and it is the one that most often returns a clear no.

The answers that mean obsolete: the project shipped, the team it coordinated no longer exists in that shape, the system it was watching was replaced, the two people it was reconciling now work together fine or no longer work together at all.

There is no repair. This is the case where editing is the wrong instinct — an obsolete meeting with a better agenda is an obsolete meeting that now takes preparation. End it.

Ending is a separate skill from diagnosing, and there is a version of it that survives: the pause rather than the kill, and the message that offers a way back so that nobody has to defend the meeting to agree to stop it. If you are the one leaving rather than the one ending it, the exit from a standing invite is the narrower move and it is available to you even when the meeting itself is untouchable.

The redesign moves, and what each one costs

Every repair above has a price. The advice that does not name the price is advice from someone who does not have to send the message.

Two observations from that table. The cheapest moves are questions, and questions are the only ones you can make unilaterally. And the cost is not proportional to the benefit — trimming an invite list is nearly free and returns more hours across the company than most cancellations do.

What a meeting costs in money is a separate calculation, and the cost arithmetic is there if you need it. Be warned that it persuades almost nobody. Attendance and ownership arguments land; dollar figures read as an accusation.

Running the redesign when the meeting is not yours

This is the situation most readers are in, and most meeting advice quietly assumes otherwise.

You cannot cancel it. You can do these, in ascending order of exposure:

1. Answer the four questions privately. Costs nothing, and you now have a defect rather than a feeling.
2. Ask the ownership question. One message, neutral framing, genuine either-way.
3. Offer to do the work. “I’ll write the Thursday update if it’s useful to try for a fortnight” is a proposal nobody refuses, and it converts a broadcast without anyone having to concede the meeting was bad.
4. Propose the trial, not the change. Two weeks, then we look at it. A trial is a much smaller ask than a decision, and the thing about a two-week trial of not having a meeting is that it very rarely gets reversed.
5. Take it to your own manager as a time problem, not a meeting problem. “I’m at fourteen hours of recurring meetings and I want to protect Wednesday mornings — here’s the one I’d drop and why” is a conversation about your output. “This meeting is useless” is a conversation about someone else’s judgement.

Step 3 is the one people skip and it is the most effective by a distance. Offering to write the thing costs you twenty minutes a week and buys back six people’s hours, and it reframes you as the person improving the process rather than the person resisting it.

When to leave it alone

Three cases where the four questions return a failure and the correct action is still nothing.

The meeting is the relationship. Some recurring slots exist so that two people or two teams stay in contact, and the agenda is a pretext. It will fail question 2 every time — nothing gets decided, it is all information — and converting it to a document destroys the only thing it was doing. A weekly with your manager is usually this. So is the standing call with the team you depend on and have no other reason to talk to.

It is somebody’s only access. A meeting that looks broadcast from your seat may be the one place a junior person hears a senior person think out loud, or the one forum where a quiet team gets visibility. Killing it is efficient and costs that person something they cannot easily replace. If that is what is happening, say so and leave it.

You have spent your capital this quarter. Pushing back has a budget and it is not large. If you have recently declined something visible or renegotiated a deadline, this is not the week to also propose ending a meeting. the honest inventory of what pushing back costs is the same argument at the individual level, and the timing part of it is the part people get wrong.

None of those are reasons to keep a meeting forever. They are reasons the answer this quarter is keep.

The two messages, in full

Almost everything above reduces to sending one of two messages. Here they are complete, to paste and edit.

The diagnosis message, when you want the meeting examined without proposing an outcome:

“Can I ask something about the Tuesday review — not angling for it to go away, I’d genuinely just like to understand it. It started around the migration last year; is it still serving that, or has it turned into something else? Asking because I’m trying to work out where my Tuesdays are going and I don’t want to opt out of something that matters.”

The trial message, when you have a specific edit and you want it tried rather than decided:

“Proposal on the Thursday sync: for the next two weeks I write the status as a doc on Wednesday afternoon, and we keep a 15-minute slot on Thursday for anything that needs discussing. If it’s worse we go back with nothing lost. I’ll do the writing.”

The second one works because it contains an offer, a fixed end date, and a default of reverting. All three of those lower the cost of saying yes, which is the only variable you control.

If the answer to either is no, that is information and not a defeat. A meeting with an owner who will defend it is a meeting with an owner, which means it passed question 1, which means the four questions did their job.

Every meeting in the week, one at a time

One move per post, each with the message you actually send: who belongs in a room, how a recurring set gets reviewed on a schedule, the cancellation that sticks, agendas written so the wrong people opt out, and the meetings that were always a document.

• Who Actually Needs to Be in the Room — Attending non-mandatory meetings is important — until you check what optional really means. Attendance by role, and the message that gets you off the list.
• The Recurring Meeting Review, Every Quarter — A standing meeting survives because nobody schedules the moment to question it. The 45-minute quarterly review, and the sunset clause that does the real work.
• How to Cancel a Meeting Nobody Wants to Defend — How to cancel a meeting that has outlived its purpose: the two-week pause that beats a kill, the message to send, and what to do when it comes back.
• Agendas That Change Who Shows Up — A team meeting agenda is usually a container for a fixed room. Written with owners, times and an opt-out line, it becomes the thing that shrinks the room.
• Meetings That Should Be a Document — An asynchronous meeting is a document with a deadline. Which meeting types convert to writing, which do not, and what each conversion costs somebody.