It's 4:13 PM on a Tuesday. You've been in five meetings since 9 AM and you haven't written anything anyone outside your laptop will see. The slack threads are stacked four deep. The PRD you promised eng on Monday is still a doc with one heading. You haven't replied to the customer who left a calmly-furious message about a feature you don't have yet.
You're not behind because you can't do the work. You're behind because the work, today, was translation. You took an idea from a customer call into a design crit, then carried it into a standup, then re-shaped it for an exec one-pager, then softened it for a reply to the customer. Same idea. Five tones. One exhausted PM.
This is the translation tax. Every product manager pays it. Almost no one names it.
The cost nobody puts on a status report.
Here is what nobody writes down on the quarterly review: the average PM we surveyed spends 22 hours a weekrewriting the same handful of decisions for different audiences. That's more than half a working week, gone — not to thinking, not to deciding — just to shaping the same content five different ways.
It's the work that doesn't appear on roadmaps. It's the work your manager can't see. It's the work that, when you skip it, immediately produces friction — a confused engineer, a frustrated designer, a CEO asking why she didn't know — and you go and do it again.
The translation tax.
The aggregate cost of taking one idea, decision, or piece of news and re-shaping it into the shape each audience needs to consume it. Paid in hours. Almost never reimbursed.
Five teams, five completely different shapes.
The reason PMs can't just "write it once and forward it" is that each audience is looking for an entirely different shape of communication. It's not that they need a longer version or a shorter version. The actual content is different.
Consider the simplest possible decision: we're going to ship bulk export this sprint.Here's what that one decision has to become.
- For engineering — a Linear ticket with acceptance criteria, repro steps if it's a bug, and an explicit out-of-scope list. The narrative is gone. The empathy is gone. What remains is a contract.
- For design — a question. What's the user need behind it, what's the constraint, what's the open thing. Bossy doesn't work. Curious does.
- For marketing — the benefit, the angle, the proof point, the audience. "We built bulk export" is not a marketing message. "We get your data out faster than your warehouse can ingest it" might be.
- For leadership — the number, the bet, the ask. One paragraph. No narrative.
- For the customer who asked for it — warmth. A timeline. A real person. No corporate hedging.
Each of those is the same decision. None of those are the same writing. And here's the part that makes it expensive: you can't draft them in parallel. You have to put on each hat, find the shape, write the words, then take the hat off before you can put the next one on. You context-switch eight times a day, and you context-switch slowly.
That's why your 5 PM feels like 9 PM. You haven't done one job. You've done five.
→ For engineering
→ For leadership
→ For marketing
→ For the customer
The shape of a fix.
Most "solutions" to this problem are someone telling you to write better templates. The template idea has the right instinct — reduce the shape-shifting cost — but it solves the wrong piece. Templates speed up the structure. They do nothing for the part that's actually slow, which is finding the voice for each audience.
The fix that worked for us was different. Two ideas, both small:
- Separate the thinking from the writing.Capture the decision once, in your own voice, while it's still warm — between meetings, on the walk, in the cab. You don't need a template for this. You need a microphone.
- Translate at the end, in batch.Once the decision is captured, the five outputs are mechanical. Same content, different shape. That's a job you can hand off — to a teammate, to a process, to a tool. The expensive part is the thinking. The shaping is cheap when you do it last.
The expensive part of being a PM is the thinking. The translation is cheap — when you do it last, in batch, and not five times across five hours.— overheard at a team offsite, written on a napkin we kept
What we built around it.
VoiceX started as a dictation tool. The thing it became, the thing PMs actually use it for, is collapsing the translation tax. You talk through the decision once. The five outputs come out the other end — in the tone each audience needs. The eng ticket lands in Linear. The warm reply goes into the Intercom draft. The exec one-pager sits in your shared Notion.
We don't think this is a magical claim. The transcription is good and getting better. The tone-matching works because the tones themselves are stable patterns — engineering writing, customer-support writing, executive writing all have shapes you can describe. The hard problem wasn't the model. The hard problem was making it fast enough that someone would actually do it between two meetings instead of waiting until Sunday night.
Try it on this week's translation tax.
Free for 14 days. No card. Cancel from your menu bar.
A note on the part that doesn't go away.
None of this removes the part of the job that's hard — the judgment about which decisions to make, the awareness of who needs to hear what, the political work of getting alignment. That work is yours, and that work is the reason you're a PM and not a writer.
What we're trying to do — what every honest productivity tool is trying to do — is collapse the parts that aren't judgment, so you get more of the day back for the parts that are. The 22 hours a week of re-shaping the same idea is not the part where you're being a PM. It's the part where you're being a copy editor.
Stop being the copy editor. Your team would rather have you back as the PM by 4 PM.