Running fewer and better remote meetings

A Tuesday that held four meetings and no work

At 08:30 one Tuesday my calendar looked reasonable. Standup at 09:45, a design review at 11:00, a stakeholder sync at 13:00 and sprint planning at 15:00. About three and a half hours of calls, which left what looked like four free hours for a bug in a billing export that I had been circling since the previous Friday.

I never touched the bug. From 09:00 to 09:45 there was no point loading it into my head, because the standup would wipe it. From 10:00 to 10:45 I was collecting screenshots for the design review. The gap before 13:00 went on lunch and on answering the review comments while they were fresh. The hour before planning was pure residue: I opened the branch, reread my own notes, and the fifteen-minute reminder popped up just as the problem started to make sense again. By 16:30 I had attended everything, said sensible things in all four calls, and the bug was exactly where I had left it.

The calendar said three and a half hours. The day said all of it. That gap, between what an invite claims to cost and what it actually takes, is what this guide is about. Everything below is a way of making it visible, and then making it smaller.

The calendar said three and a half hours.

Price the meeting before you trust the seat time

An invite shows a block. The day also loses the preparation before it and the stretch after it, when your head is still in the call and the work you left has gone cold. Three calls that look like two hours can quietly erase an afternoon.

Until your own log says otherwise, price a live hour at about one and a half times the invite. Sixty minutes, plus roughly fifteen to prepare and fifteen to write notes and climb back into the work, comes to ninety. A pairing session you already prepared for may cost close to its seat time. An unprepared status round often costs more than the multiplier, because the preparation happens during the call, out loud, with everyone waiting. I kept a rough log for a couple of weeks once and my own number landed around there. That is all I would claim for it: it is a planning habit, not a measurement.

The habit changes what you plan. On a day like that Tuesday, do not assign a feature or a hard bug to the gaps between calls. Give those gaps to review comments, small fixes and replies. The cost you never see on the calendar is part of the real cost of context switching, and it is the reason a meeting-heavy day feels so much more tiring than its hours suggest.

The fairness of the hour matters as much as its length. A 21:00 slot for one region is never cheap, whatever multiplier you use, because the multiplier does not include somebody's evening. How to spread that cost properly is in timezone collaboration without chaos. Which work should never have been a call in the first place is in async communication for distributed teams.

One more distinction helps. If nothing changed on a ticket, a document or a calendar after the hour, the hour was social. That is allowed, and remote teams need some of it. Name it honestly, as a social call, so people can decline it without feeling they skipped work. A sync dressed up as a social call, or the reverse, is where resentment starts.

Set a weekly meeting budget by role

A meeting budget is a planning input, the same as overlap hours or a release date. A week that went over budget because of an incident is a week to repair. A week that went over budget and shipped nothing is a week to cancel or shorten something. Write the number down next to your overlap hours, somewhere the team can see it. A figure that lives in someone's head loses to every invite.

These are starting ranges. Edit them, then publish them:

• Engineer or designer: about 6 to 10 live hours in a heavy week, and two mornings with no meetings at all. Those mornings are a team norm, not a favour, and they are not the block you offer when someone wants a roadmap tour.
• Lead: higher, but still capped on status. Protect one decision block inside the shared overlap, because a lead's week that is all status is buying the wrong thing.
• Freelancer: calls only as the contract states, inside a window you chose. Outside that window, price the call or send a decline note.
• Product manager across regions: plan by pairwise overlap rather than one big slot. No brief, no live decision, and the hour goes back into the budget.

If every deadline eats the meeting-free mornings, they were never real, and it is better to admit that than to keep pretending. The freelancer line deserves one more sentence: "happy to jump on a call" is how two calls a week become five without anyone writing a change order. The client is not being unreasonable. You just never told them what a call costs.

Batch reviews the way you would batch a deploy. Put them next to each other, with one preparation window, and keep a contiguous block free on the other side of the day. Five scattered calls spend the multiplier five times. The same five calls after lunch spend it roughly once. If you are over budget by Wednesday, the next yes should name which series you are skipping to make room. "I will make it work" is how the cost moves to Thursday night.

Run the meet-versus-async test on the invite

Two questions catch most bad invites: what is this hour for, and did writing get a fair chance first?

Information belongs in a document or a short recording. If people cannot open it, fix the access, do not read it aloud. A choice between two options starts as a brief with a comment deadline. Go live only when one written round still ties two defensible options and the comments were actually filed. If nobody commented, the default applies. Booking a meeting because nobody commented rewards the people who skipped the reading.

A bug starts as logs, steps to reproduce and what you already tried. Share a screen when the written reproduction still fails, not before. If the steps only exist in your head, write them first, which often solves the bug on its own. Hard feedback is a 1:1 inside overlap hours, never a trailing item at the end of group status. A brainstorm with no owner is theatre, so name the owner on the invite or do not book it. Status belongs in an end-of-day note; the buckets I use are in end-of-day status update.

The longer version of this test, with the order I ask the questions in, is when a remote team needs a meeting. If the topic is already a written argument that has stalled, use the narrower thresholds in when async disagreement needs a live call. Tension on its own is not a reason to meet. Two named options that writing could not separate are.

Before you accept any invite, read the description and check four things: a brief, an exit line, a clear split between required and informed people, and a fair hour for everyone required. If those fail, decline, or propose the brief yourself, or offer fifteen minutes that produces both and then ends. If the same thin invite collects two required declines in one week, cancel it and send a note instead. Reading the note will be faster than the call ever was.

Replace the daily round-robin with a written note

The team I learned most of this on had eight people across four time zones and a daily thirty-minute video standup. The tickets already held the state of the work. The standup mostly existed so that people could hear each other read their tickets aloud, and it sat at the very start of the day for one region and near the end of it for another.

We replaced it for two weeks with written notes on the tickets plus a short daily post in four buckets: Done, Open, Blocked, Next. The rule for the trial was simple. Watch for one thing only: which problems still could not travel in writing?

The first few days were clumsy. People wrote too much, as if length proved effort, and a couple of people wrote nothing and assumed the tickets spoke for themselves. By the end of the first week the notes had settled into three or four lines each, and I was reading the whole team's day in less time than the call used to take just to start.

One thing did break, and it was specific. Blocked items that needed somebody in a different zone sat for most of a day, because the person who could unblock them read the note after the author had logged off, replied, and then the author read the reply the following morning. Two round trips, two days. So we restored only that slice: a weekly twenty-five minutes on blocked items, with an exit line that every blocker leaves with an owner and a UTC date. The daily round never came back, and nobody asked for it.

The swap fails in a predictable way. If the tickets are empty, people conclude that "writing does not work here" and the call returns. Fill the fields first. A recurring sync with no agenda is the same hole with a calendar icon on it. Cancel it, or write the brief the agenda never had.

When the note has to transfer ownership overnight, write a handoff rather than a status. The fields for that are in leave work someone else can continue. Status buckets leave out the "how to check it works" step, which is the one the next person needs. The opposite mistake also happens: a full handoff written every evening when you still own the work tomorrow is a document nobody asked to pick up.

Name the replacement before you delete any invite:

• A status round becomes the four buckets.
• A "quick alignment" becomes two options and a comment deadline.
• A roadmap preview becomes a short recording plus a question thread with a UTC deadline.
• A brainstorm becomes silent writing first, then twenty live minutes only if the clusters still tie.
• A forty-minute lecture becomes the recording, and live time is for objections after people have watched.

One series at a time, two weeks each. Bring back only the slice that broke, and bring it back smaller than it was.

Put an exit line on every invite past fifteen minutes

Anything longer than fifteen minutes needs five things, or the invite waits until it has them.

1. A purpose that is one of decision, review, pairing or unblock. "Sync" is not a purpose. Two purposes means two slots, or a document for one of them.
2. A pre-read that people open before they join. Across zones that do not share a morning, send it one workday ahead. A link posted twenty minutes before the call is a prop.
3. An exit line, written as a sentence you could paste onto the ticket afterwards. "We leave with A or B chosen." "The pull request is approved, or the list of changes is written down." "Align on direction" will not paste.
4. A timebox per topic in the doc. When the time ends, decide, or move the topic to writing. Do not borrow the next topic's minutes; that is how the real decision dies on "we are at time."
5. One note owner, who puts the decision on the ticket within an hour.

If the decision is missing from the ticket the next day, the meeting did not finish. Collect it in writing. Do not book a second hour to remember what the first hour decided; shrink the series instead.

A usable invite title looks like this: "Choose checkout provider A or B. Pre-read linked. 25 minutes. Exit: the decision is on PAY-200." I have yet to see anyone arrive at a call with a title like that and ask what it was for.

Running it is mostly discipline. Start on time and point late joiners at the doc rather than recapping. Cameras are optional. End when the exit line is met, even if that is at minute twelve. Anything that starts with "while we are here" becomes next week's agenda. Keep required and optional people separate in the invite; optional people get the summary. If you cannot write the summary without them, they were required.

If nobody read the pre-read, cancel the read-aloud rather than holding it:

We are missing the pre-read. Comments go on the doc before Thursday 15:00 UTC. We meet only if A and B are still tied.

In a hybrid room, everyone joins from their own laptop, including people sitting in the same office. Call on remote people first. A remark that never reaches the doc did not happen. If a room camera really is unavoidable, rotate a remote advocate whose job is to interrupt when the room goes quiet on the screen. Record only when someone absent needs it, and always pair the recording with the written decision. A recording on its own makes the next person watch the whole hour to find out what you chose.

Copy the decision note template

Decision:
Options considered:
Chosen because:
Owner / next step / date (UTC):
Out of scope today:

Fill it in before you leave the call, or within the hour, and paste it on the ticket named in the exit line. Chat can announce the note. The ticket keeps it, because chat scrolls and tickets do not.

Here is one filled in. Assumptions: a twenty-five minute call, the European and US East engineers required, the pre-read compared two payment providers, and the Asia-Pacific engineer optional and reading this at the start of their next day.

Decision: Ship checkout with provider A for PAY-200
Options considered: A (lower fees) and B (stronger EU SCA defaults)
Chosen because: legal has not flagged an SCA gap, and A is cheaper at the current volume
Owner / next step / date (UTC): backend engineer updates the ticket owner by Thursday 16:00 UTC
Out of scope today: migrating existing subscribers

The "Chosen because" line is the test. If it will not fit in one or two sentences, you did not decide. You agreed the topic is complicated. Put the open question back on the doc, name the missing fact, and stop. More meeting time will not produce a fact that nobody in the room has.

"Out of scope today" sounds like bureaucracy and is the most useful line of the five. It stops the next meeting from relitigating a neighbouring topic. Without it, the neighbour turns up in the last five minutes, the note owner hesitates over whether it was decided too, and the decision you did make ships wrapped in fog. For a fuller record of decisions that need to last longer than a sprint, I use a five-field note on the ticket.

Send the decline note before you accept the hour

With clients, a call often stands in for a feeling of control. They cannot see the work, so they ask to watch it. What helps is a standing window, for example Tuesday 14:00 to 15:00 in their time zone, with delivery kept outside it. Written status carries at most one decision request. The contract names the emergency channel and the reply time. Without that, every anxious message is an emergency. I send the boundary at kickoff, before there is any tension to read into it:

I batch client calls on Tuesdays. Urgent blockers go by email with BLOCKED in the subject. I reply within four business hours.

A bare no teaches people to route around you. A no with a deadline, or with a draft attached, teaches the next invite to arrive in better shape. Shrink when the topic is real and the format is bloated. Pass when you were only invited to watch. Use the doc deadline when nobody has written down the two options yet. These are the notes I paste most often:

Can we resolve this in the doc by Thursday 17:00 UTC? I will book 20 minutes if the comments still tie.

I am optional for context. Please send the decision note. I will read it async and comment only if the exit line names me.

I can do 15 minutes if we leave with an owner and a UTC date. I cannot do 60 minutes of open sync.

Put the decline on the invite itself, next to the response buttons, not only in a side message that the organiser reads tomorrow. If you are required and the local time is outside 08:00 to 18:00, say so plainly, offer the next overlap hour, and offer the written path if that hour is also too late. Do not apologise for a time zone the series already knew about.

Exceptions are fine. An exception without an end date becomes the calendar. Write the end date, or do not make the exception.

Score every recurring series and cancel one

The first time I exported every recurring invite I was on, there were eleven series. I could explain what seven of them produced. Two I had joined for a project that ended months earlier, and two I genuinely could not remember accepting. None of the four were harmful on their own. Together they were about three hours a week, most of it in the mornings.

Export yours and score the artefact, not the atmosphere. Talk about culture after the export, or the series will survive the conversation.

• Keep a series that produced an exit and an artefact last month, then shorten it to the smallest length that still produces them.
• Rewrite a real need in a weak shape: pre-read required, spectators cut, note owner named. If you cannot cut anyone, required and optional are still mixed.
• Cancel status-only series, and any series nobody can defend by naming what would break without it. Say the replacement in the cancellation: Done, Open, Blocked and Next, wherever the state already lives.
• Move a series that puts one region outside a normal day every single time. Rotate the late hour in public, or it is an assignment.

If you lead the team, a four-week version looks like this. Week one, replace the standup with the written buckets and note any escalation that still needed a voice. Week two, cancel the "quick sync" with no agenda; a real decision gets a brief and a short slot, and a wish to watch people work gets the written note. Week three, cut planning from sixty minutes to twenty-five with an exit line. Week four, keep only the slice that broke.

Planning and review earn their place when they create tickets people actually pick up. A refinement block with no exit becomes a brief plus a short overlap for ties. A three-hour backlog tour means the backlog itself is the document. Sales and engineering introductions can be batched with a brief sent first; engineers are optional unless the exit line needs a technical choice made live.

On a back-to-back day, keep ten minutes for the note. A shorter gap is just residue. How meeting time fits around real work across a whole week is in plan a week of remote work, and after a heavy day the honest goal is one finished artefact and a pickup note for tomorrow.

Each month, ask four questions of every series that remains: was the exit line met, was the decision on the ticket within a day, is the required list honest, and was the pre-read used? Three no answers means cancel or redesign. Do not add a new series to discuss the old one. If you need to argue the case upward, the number to take is contiguous maker time left in the week, or an export showing that a series produced no artefact for a month.

A meeting you delete while the writing stays weak will come back, usually with a new name and a longer slot. The real work is making the written path good enough that nobody misses the call: the brief, the deadline, the decision note on the ticket. When a two-week written replacement breaks nothing you can name, the deletion is done. When it does break something, name it in one sentence and fix that sentence with the smallest live touch that works, which is almost always smaller than the meeting you started with.