How to run an international town hall across languages
Decide whether the town hall is broadcast or conversation, because that determines everything else. Translate the Q&A, not just the presentation — and confirm before you choose a platform that the recording will keep the translation, because Teams and Meet drop it.
An all-hands or town hall has a structural problem no other event format has: it is a broadcast that is supposed to feel like a conversation. Those two halves need different handling.
Separate the two halves
The presentation is one-to-many and predictable. You know the speakers, roughly what they will say, and how long it runs. This part is easy to translate well: brief the speakers, load a glossary, rehearse once with translation running.
The Q&A is many-to-many and unpredictable, and it is the part that makes a town hall a town hall. It is also where language exclusion becomes visible, because the people who ask questions live in a second language are the confident minority.
Organisations routinely translate the first half and not the second, and then wonder why engagement outside headquarters is low.
Translate the questions, not just the answers
The most common design error is one-directional translation: the executive is translated into five languages, and questions arrive in English only.
That structure tells everyone whose language is the working language. It also means the questions you receive are filtered by English confidence rather than by importance.
Practical fixes, roughly in order of impact:
- Collect written questions in any language, before and during, and commit to answering a stated number of them
- Translate the questions on screen when they are read out, so the asker sees their words treated as first-class
- Let people speak their own language live if your tooling supports symmetric translation
- Publish the unanswered questions afterwards, with answers, in each language
Timing across zones
A single town hall convenient for headquarters is inconvenient for someone. Two options, both defensible:
- Run it twice, at times that suit two halves of the world. Expensive in executive time, and the second one is always flatter.
- Run it once and invest properly in the recording — which only works if the recording keeps the translation.
Confirm that before choosing the platform. Microsoft’s own documentation states that Teams meeting recordings only capture the original meeting audio and don’t include interpretation audio, and translated transcripts disappear at the end of the meeting. Google Meet does not translate recordings or live streams.
If half your company watches asynchronously and the recording is monolingual, you have translated the live audience and excluded the larger one.
Platform reality for this specific format
Town halls are the format native tooling handles worst:
- Teams Interpreter is not available in town halls, by design, and requires Microsoft 365 Copilot rather than Teams Premium. External guests without a licence are excluded.
- Google Meet supports one language pair per meeting. A company with Italian, Spanish and German offices cannot run this on Meet at all — not a licensing problem, an architectural one.
- Zoom translates to captions only. Workable if your audience is happy reading, which for a 45-minute leadership presentation many are not.
If you have three or more languages, external participants, and an asynchronous audience, none of the native options fits, and it is better to know that during planning.
A short checklist
- Presentation and Q&A both translated
- Questions accepted in writing, in any language
- Speakers briefed on pace, idiom and acronyms
- Recording behaviour confirmed before platform choice
- Unanswered questions published afterwards, translated