Running Workshops Across Time Zones

Running workshops across time zones starts with one honest calculation: how much of the work truly needs every person awake at the same moment. For most sessions the answer is sixty to ninety minutes, not a full day, and everything else - generating options, ranking them, reading the background, writing the decisions up - can happen asynchronously at each person's own reasonable hour.

That single decision fixes most of what goes wrong. The usual failure is not a bad video connection or a dull agenda; it is a facilitator who booked a three-hour workshop in their own working day, discovered that this put Singapore at 22:00 and Sydney at midnight, and then wondered why one third of the room never spoke. People at the wrong end of the clock do not argue, do not volunteer, and do not push back - and their silence gets misread as agreement.

This guide covers how to map the real overlap window, four workshop shapes that fit different amounts of overlap, how to rotate the unsociable hour fairly, what asynchronous rounds should actually contain, how to hand over between regions, and the failure modes that appear once daylight saving shifts underneath a recurring series.

What actually breaks when a workshop spans time zones

Three separate things break, and they need different fixes. The first is arithmetic: there is often no hour that is civilised for everyone, so any single live session imposes a cost on someone. The second is cognitive state: a person joining at 07:00 before the school run and a person joining at 22:00 after dinner are not bringing the same attention, however willing they are. The third is participation asymmetry, which is the expensive one - the people at the awkward hour contribute less, so the workshop quietly over-weights whichever region owns the comfortable slot.

That third effect compounds across a series. If the same region always gets the late call, they stop preparing, then they stop turning their cameras on, then they send one representative instead of four, and within a quarter you have a workshop that is really a headquarters meeting with observers. Nobody decides this; it accumulates one unsociable invite at a time.

The remedy is structural rather than motivational. You do not fix a 22:00 call with better energiser activities. You fix it by needing less synchronous time, by moving the work that does not need a live call out of the live call, and by sharing the discomfort on a schedule people can see.

The numbers to design against

These are design defaults, not research findings - start here and adjust once you know your specific spread. The one worth defending hardest is the ninety-minute ceiling. A live block that runs to three hours is almost always carrying work that could have been done asynchronously, and the cost of that padding lands entirely on whoever is furthest from your own clock.

Map the real overlap before you design anything

This takes about twenty minutes and it is the highest-leverage twenty minutes in the whole design. Every downstream choice - the shape, the length, what goes async - follows from the size of the intersection you find.

Choose a workshop shape that fits the overlap you have

The relay is the shape most people underuse. Each region gets a ninety-minute block inside its own working day, and the sessions are chained: Singapore runs first and hands over to London, London hands over to San Francisco, and the last block closes the loop. The overlap hour between consecutive regions - London 09:00 with Singapore 16:00, then London 17:00 with San Francisco 09:00 - is where the handover conversation actually happens live, and it is worth protecting.

Mirrored sessions are simpler but cost you more. You run the same ninety minutes twice, which doubles your own load and produces two conversations that diverge by the second run because you already know the answers. If you mirror, keep both sessions genuinely open, and merge the two outputs in public afterwards rather than silently privileging whichever ran second.

Put the right work in the live block

The test for whether something belongs in shared live time is simple: does it need people to react to each other in the moment? Disagreement does. Trade-offs where one team's priority sinks another's do. Anything where tone and hesitation carry information does. A decision that needs to be made in front of everyone, so that nobody can later claim they were not there for it, does.

Almost nothing else does. Presenting context does not - that is a document or a five-minute recording. Generating ideas does not, and generating them live is actively worse, because the people at the awkward hour produce less and the fluent English speakers dominate. Ranking a list does not. Reading a proposal certainly does not, though facilitators still burn twenty minutes of the only shared hour on silent reading.

So the ninety minutes should look thin on paper and feel dense in the room: a two-minute framing, ten minutes on the results of the asynchronous round that everyone completed yesterday, then an hour on the three points where the round showed genuine disagreement, then fifteen minutes to name decisions and owners. That is a workshop. Everything else was already done, at each person's own convenient hour.

Make the asynchronous half do real work

Asynchronous does not mean 'a document nobody opens'. It means structured rounds with a question, a deadline and a visible result - which is roughly what the Delphi method has always been, and what the nominal group technique does in a room. Collect silently first, show the aggregate second, discuss third. Splitting collection from evaluation matters more when people are distributed, because the alternative is that whoever is awake and confident sets the frame for everyone still asleep.

This is where a live session tool earns its place asynchronously. Session Flo activities can be left open across a full day and night, so a word cloud, a ranking or a survey collects from Singapore in the afternoon, London in the evening and San Francisco the following morning, and you walk into the live block with the aggregate already on screen. The room spends its scarce shared time on the split in the results rather than on producing them.

Give each round a name and a hard close, and enforce it. Rounds without deadlines drift for a week and lose the people who answered on day one. A round that closes at 08:00 UTC on Thursday, stated as 16:00 Singapore, 09:00 London and 01:00 Pacific - which is why Pacific answers on Wednesday - is unambiguous in a way that 'by Thursday' never is.

Run the live block so nobody is a spectator

Handing over between regions

The handover note is the load-bearing part of a relay, and it is the part most often skipped because the block overran. Protect the last ten minutes of every block for it, in the same way you would protect a fire exit. A note that says 'we agreed on the two-tier model, we could not agree who owns migration, and the question for London is whether the tier boundary survives EU data rules' is worth more than a full recording nobody will watch at 07:00.

Keep the note to one screen. Its job is to transfer state, not to archive the discussion. If the second region has to read four pages before starting, they will start without reading, and you get the divergence a relay was supposed to prevent.

Rotate the unsociable hour and publish the rota

Someone has to take the bad slot. The only question is whether it is always the same people. Write a rotation for the next quarter, publish it where everyone can see it, and stick to it even when it is inconvenient for you - especially when it is inconvenient for you, because the first time the anchor region takes the 22:00 call is the moment the rotation becomes credible.

A workable pattern for three regions is a three-session cycle: session one favours Asia-Pacific, session two favours Europe, session three favours the Americas, and the region being favoured also does the pre-work summary and writes the recap. Attaching the chore to the comfortable slot balances the ledger a second time and stops the rotation feeling like a tax.

There are limits worth stating out loud. Nothing before 07:00 or after 21:00 local for anyone, no exceptions for 'this one is important' - because they are all important. If your session genuinely cannot fit within those bounds for all three regions, it is telling you that this workshop should not be fully synchronous, and you should move it to a relay or to asynchronous rounds instead of asking someone to take a midnight call.

Recordings, recaps and the people who could not attend

Recording the call is the minimum, and on its own it is close to useless. Nobody in Sydney is going to watch ninety minutes of video before their own working day starts. Treat the recording as an archive for the two people who need to check exactly what was said, and put your effort into the written recap instead.

A recap that works has four parts and fits on one screen: the decisions with owners, the things explicitly not decided and when they will be, the raw output of any activity so people can see their own contribution in it, and one open question for anyone who was not there. That last part is what converts absence into asynchronous participation - a person who missed the call can still change the outcome by answering it before the next round closes.

Timing matters as much as content. Publish the recap while the sleeping region is still asleep so it is waiting at the top of their morning, and keep the follow-up window open for a full working day on their side. A recap that arrives at 16:00 their time, with comments closing that evening, is a notification rather than an invitation.

The failure modes to watch for

The most common failure is the phantom overlap: a slot that looks fine in the calendar because the invite shows your own time zone and everyone else's is collapsed behind a tooltip. Always type the other local times into the invite body. A line reading '09:00 London / 16:00 Singapore / 01:00 Pacific' makes an unreasonable request visible before anyone accepts it.

The second is agenda creep in the shared block. Ninety minutes becomes two hours because a demo was added, and the extra thirty minutes lands entirely on the tired region. Anything added to the live block should displace something, not extend it.

The third is the daylight saving drift that breaks a recurring series twice a year, usually in the fortnight when Europe has changed and North America has not. Anchor the series to one zone, and send a note in the weeks where local times move.

The fourth is quiet substitution: the region at the awkward hour starts sending one delegate instead of four. That is not a scheduling problem any more, it is a signal that the workshop has stopped being worth the sleep, and the fix is to move more of it asynchronous rather than to insist on attendance.

A default plan for a three-region workshop

Run this shape three or four times with the same group and something useful happens: the asynchronous rounds start filling themselves, because people learn that answering the round is how they get influence, and that the live hour will not repeat what the round already settled. That is the point of the whole design. The clock difference stops being the thing everyone works around and becomes just another constraint the format already handles.