When a meeting runs in everyone's second language, two losses stack on top of each other. The speaker can only put part of what they mean into English, and the listener can only catch part of what was said. Picture it roughly as six-tenths times six-tenths — and less than half of the original idea survives the trip. That math is a way to see the problem, not a measurement, but the effect is real and most teams never name it. The fix isn't getting everyone's English better; it's letting each person speak and read their own language.
The two losses don't average — they multiply
Here's the part teams miss. When you run a meeting in a shared second language, there are two separate places meaning leaks out:
- The speaker's loss. Expressing a thought in a language you didn't grow up in, you simplify. You reach for the word you have instead of the word you wanted. The careful caveat, the precise verb, the bit of nuance that made the point land — some of it just doesn't come out. Say you get sixty or seventy percent of your actual meaning into the sentence.
- The listener's loss. On the other end, someone who also isn't a native speaker is parsing your accent, your speed, an idiom they half-know. They catch most of it — call it sixty or seventy percent of what you managed to say.
The mistake is to think these average out to "pretty good." They don't. They multiply. If the speaker conveys 0.6 of the idea and the listener catches 0.6 of that, what actually lands is 0.6 × 0.6 ≈ 0.36 — barely a third of the original thought. (To be clear: those numbers are an illustration to make the shape of the problem visible, not a benchmark. The point isn't the exact figure; it's that the two losses compound instead of cancelling.)
And it gets worse with the room size. Add a third non-native participant relaying the point onward and you multiply again. This is the corporate version of the telephone game, except everyone is being perfectly sincere and trying hard.
The quieter cost: people stop talking
There's a second cost that doesn't show up in any transcript: people who aren't confident in their English say less. They wait until they're sure. They let the fluent speakers carry the meeting. They drop the half-formed idea that might have been the best one in the room, because workshopping it out loud in a second language feels too exposed.
So the price of "let's just use English" isn't only the meaning that leaks on each hop — it's the contributions that never get made at all. The most fluent people end up dominating, which has very little to do with who has the best judgment. This part is well documented; non-native speakers in second-language meetings comprehend less and participate less, and we walk through the actual research in the hidden cost of running meetings in everyone's second language and why "everyone just speaks English" quietly fails APAC teams.
"Just improve everyone's English" doesn't scale
The instinct is to fix the people: more training, more practice, hire for fluency. But language training is slow, expensive, and never actually finishes — fluency is a spectrum that shifts with stress, fatigue, and topic. Even strong second-language speakers lose ground in a fast technical argument. You can spend years closing the gap and still have native speakers setting the pace. It's the wrong thing to optimize.
The reframe: one conversion instead of two
So here's the better question: what if nobody had to translate themselves in real time?
Let the speaker talk in their own language. Now they're at full expression — the nuance, the caveat, the precise word all make it into the sentence, because it's the language they think in. Let the listener read in their own language, at full comprehension. The translation happens once, in the middle, by a machine — and machine translation isn't perfect either. But that's one conversion step, and a small loss, replacing two big human ones. Keep the same rough picture from before and the bar barely dips: most of the meaning makes it across, instead of a third.
That's the whole idea. You don't make the losses disappear — you stop them from compounding, and you take the social tax off speaking up, because nobody is performing in a second language anymore.
How Sageio does it
Add bot@sageio.net to your Google Meet calendar invite and it joins on its own — nothing for anyone to install. Each participant picks the language they want captions in, so the same meeting serves everyone at once: you speak Japanese, a colleague reads English, another reads Traditional Chinese, all from the same words, with captions appearing in about two seconds. Sageio translates into 20+ languages, with Asian languages treated as first-class. Within about five minutes of the call ending, a searchable transcript and an AI summary arrive, shared at the host's discretion.
(Today this runs on Google Meet; Zoom and Microsoft Teams support is coming soon.)
For the broader picture of how this works and what to look for in a tool, see real-time translation for remote teams.
Frequently asked questions
Does running meetings in English actually lose information? Yes, in two compounding ways. A non-native speaker conveys only part of what they mean, and a non-native listener catches only part of that — and those losses multiply rather than average, so well under half of the original idea can survive a single exchange. There's also a participation cost: people who aren't confident in their English contribute less, so ideas are lost before they're ever spoken.
Isn't machine translation also lossy? It is — no translation is perfect. The difference is that it's a single conversion step instead of two. When two people each strain through a second language, you stack a speaking loss on top of a listening loss. Translating once, by machine, while each person uses their own language, replaces two large losses with one smaller one.
How does per-person translation help people speak up? Because no one is auditioning in a second language. When you can speak your own language and trust that everyone reads it in theirs, the fear of sounding less competent than you are goes away — and the people who used to stay quiet start contributing.
How fast are the translated captions, and what does it cost to try? Captions appear in about two seconds, with a searchable transcript and summary within about five minutes after the call. Every plan starts with a free 60-minute trial, no credit card required; after that Professional is $49/month and Teams is $99 per seat/month (annual billing includes 2 months free), with custom-priced Enterprise.
"Let's just use English" feels free, but it quietly taxes every cross-language conversation twice — once when the idea is spoken, once when it's heard — and a third time in the ideas no one risks saying. Letting people speak and read their own language is how you stop paying it. Add the bot to your next meeting and watch who starts talking.