Running a Town Hall for a Distributed Company

A town hall meeting for a distributed company has a harder job than the version that used to happen in a canteen. It has to hold the attention of several hundred people who are alone at desks, each one three clicks away from something more urgent, and it has to do that without the social pressure of sitting in a room where standing up and leaving would be noticed.

The shape that works is tight: forty-five minutes, three content segments of roughly ten minutes, fifteen minutes of questions that were submitted and ranked before the meeting started, and at least three moments where every attendee does something with their hands. Everything else - the deck, the guest speaker, the quarterly numbers - is negotiable around that skeleton.

This guide covers what a distributed town hall is genuinely for, the minute-by-minute segment structure, how to collect and rank questions so the uncomfortable ones actually get asked, what to do when the company spans twelve time zones, the numbers worth holding yourself to, how to handle a hostile question without freezing, and what has to happen in the twenty-four hours afterwards.

What a distributed town hall is for - and what it is not

A town hall exists to do three things a document cannot. It puts the whole company on the same page at the same moment, so nobody learns about a strategy change third-hand a week late. It lets people watch leaders answer questions they did not pre-approve, which is the only reliable signal most employees get about how candid the leadership actually is. And it gives the organisation one shared reference point - 'as we said at the town hall' - that makes every follow-up conversation cheaper.

It is not an information channel. If your town hall is mostly the transmission of facts that could be read faster than they can be spoken, you are using forty-five minutes of everyone's week as a very expensive newsletter. Move the facts into the recap and use the live time for the parts that need a human: the reasoning behind a decision, the trade-off that was actually hard, the question somebody was afraid to ask.

It is also not a departmental deep dive. The moment a segment only makes sense to one function, seventy per cent of the audience opens another tab and does not come back. If engineering needs forty minutes on the platform migration, that is an engineering session, not a town hall slot. The test for any segment is simple: would a new joiner in a completely different team be able to say why this matters to them?

The forty-five minute shape

The ordering matters more than the content. The arrival poll goes first because the participation norm is set in the first three minutes and cannot be reset later - a company that spends the opening watching a slide has already decided this is television. The hardest questions go at the start of the Q&A block rather than the end, where they get squeezed out by the clock and everyone notices.

Forty-five minutes is a ceiling, not a target. A month with nothing significant to report should produce a twenty-five minute town hall and a visible relief in the room. Filling the full slot regardless of whether there is anything to say is how a town hall becomes a ritual people learn to skip.

Collect questions 48 hours before, not in the chat

Live chat questions look democratic and are not. They favour whoever types fastest, whoever is awake at that hour, and whoever is confident enough to put their name next to a difficult question in front of the whole company. In a distributed organisation that skews the agenda towards one time zone and one personality type every single month.

Session Flo handles the submission-and-upvote pattern from the same room code as the polls, so people are not being asked to join a second tool halfway through the meeting. That matters less for the question quality and more for the volume - every additional join step costs you a slice of the audience that was already half-committed.

Give people something to do at least three times

Three interaction moments is the minimum for a forty-five minute distributed session and five is comfortable. The purpose is not entertainment, it is attention maintenance - a person who typed something two minutes ago is far more likely to still be listening than one who has been in receive mode since the start.

Space them roughly evenly, and put one immediately after your densest content. If you have just spent ten minutes on a pricing change, a five-second confidence check tells you whether it landed while you can still do something about it. Waiting for the engagement survey in six weeks tells you the same thing far too late.

Resist the urge to make every interaction a quiz. Scored games belong in team sessions, not in a meeting where you are about to discuss headcount; the register clash is jarring and people notice. Save the competitive formats for the sales kick-off and keep the town hall's interactions informational.

Time zones: rotate the pain, do not average it

Companies spanning eight or more hours of time difference usually settle on a compromise slot - typically late afternoon in Europe, early morning on the American west coast, and the middle of the night in Asia-Pacific. This looks fair and is not. It is a permanent tax on whichever region drew the short straw, paid every single month, and the people paying it stop coming within a quarter.

Rotate instead. Run the same town hall at two different times on the same day, or alternate the slot month by month so each region takes an uncomfortable hour a third of the time rather than one region taking it always. Two live runs sounds expensive until you notice the second one only needs the questions block re-run live - the content segments can be the same recording.

Whichever pattern you choose, make the asynchronous path first-class rather than an apology. That means the recording is edited and posted within four hours with timestamps against each segment, the question list stays open for another twenty-four hours, and the poll results are published so people who watched later can see where the room landed. If your recording is a raw ninety-minute file with no chapters, you have quietly told a third of the company that they are the audience for the leftovers.

The numbers worth holding yourself to

These are structural commitments rather than research findings - targets you set and are then visibly held to. Their value is that they are all checkable by anyone in the company, which makes them much harder to quietly abandon than a goal like 'a more engaging town hall'.

Two further numbers are worth tracking privately. The first is the proportion of attendees who interact at least once; if that sits below half, your interactions are badly placed or your audience has already gone passive. The second is the number of questions submitted per hundred employees - a figure that drops month on month is the clearest early warning that people have concluded the question box does not lead anywhere.

Handling the question you would rather not answer

Who runs what on the day

A distributed town hall needs three roles and they should not all be the same person. The host runs the clock, introduces segments, reads the questions and has standing permission to cut a rambling answer short. The question wrangler watches the submissions, merges duplicates live, flags the follow-ups coming in through the chat, and keeps a written list of what went unanswered. The operator drives the polls and the screen share and watches for the technical failure that will otherwise cost you four minutes.

Splitting the host from the most senior person in the company is the highest-value change most organisations can make. A chief executive who is also chairing cannot interrupt themselves, cannot manage the clock while thinking about an answer, and cannot read the room and speak at the same time. Handing the chair to someone else makes every leader in the session sharper.

Run a ten-minute technical check before every session, not just the first one. Screen sharing, audio for remote presenters, the poll display and the recording all need to be confirmed live in the actual meeting room. A distributed audience will forgive one bad question; it will not sit through six minutes of 'can everyone see my screen?' more than twice.

The twenty-four hours after the town hall

Assume that between a third and a half of your company will experience the town hall as a document rather than as a meeting. That includes the people who were asleep, the people who were on customer calls, and the people who joined and then got pulled away. Write the recap for them and it will also serve everyone who attended and forgot half of it by Monday.

The recap has a fixed shape that takes about twenty minutes to produce: what changed, what was shown, what was asked and answered, what is still open with names against it. Session Flo assembles the interactive parts of that automatically - the poll results, the word clouds, the question list with vote counts - which leaves you writing only the narrative connective tissue rather than reconstructing what happened from memory.

The last checklist item is the one that decides whether the whole format survives. Opening each town hall by closing out the previous month's open questions takes ninety seconds and is the single strongest signal that the meeting is a mechanism rather than a performance. Skip it twice and submissions start falling immediately.

What to cut when it overruns

Something will overrun, and the default reflex - cut the questions block - is the worst possible choice. It is the only segment the audience cannot get from a document, and cutting it teaches people that the question box is a suggestion. Protect the fifteen minutes and cut from the content segments instead.

In practice the order to cut is: the department update nobody asked for, the celebratory slide that could be a message, the second half of the guest segment, and the recap of what was already sent in writing. If you are still over, cut the number of questions rather than the time allocated to them - four properly answered questions beat nine rushed ones, and the room can tell the difference immediately.

The exception is a genuine crisis. When something significant has happened - a layoff, an outage, an acquisition - the format changes: shorter content, longer questions, and a second session two days later once people have processed the first. In those weeks the town hall is the most valuable meeting the company has, and the usual forty-five minute discipline can bend without losing the shape.