Why Cross-Functional Teams Can't Understand Each Other

Have you ever found yourself in a meeting¹ watching someone two levels above you try to explain your team's roadmap to a visiting executive, and suddenly realized: they're not actually doing anything? They're pointing at a slide deck (which someone on your team made from data you helped compile, which itself came from customer interviews conducted by the research team), and they're... what? Converting? Reformatting? Taking information coded in one language and recoding it into another?

They're translating.

Not translating English to French, obviously. But taking "here's what customers told us they need" and converting it into "here's our strategic direction," which is itself a stepping stone toward the eventual "here's what the board wants to hear." Each person in the chain is performing the same basic operation: input information, process it through the particular filter of their role and audience, output a modified version. You're all human APIs.

And here's what makes this uncomfortable: this isn't some bureaucratic dysfunction. This is the job. The actual job. Most of what any of you do, all day, is translation work.

The pretense of real work

The thing about translation is we pretend it isn't real work. Real work means making something or doing something: writing code, closing deals, shipping product. Translation feels like... overhead. Administrative. The stuff you have to do between the real work. Except when you actually track your hours (not in some official timesheet way, but honestly) how much of your day involves creating something new versus reformatting something already created?

Engineers translate specifications into code, sure, but before they write a single function they're translating vague requirements into clarifying questions, translating answers into technical approaches, translating those approaches into ticket estimates, translating their progress into stand-up updates. An analyst doesn't just "do analysis." They translate raw data into clean datasets, translate findings into charts, translate charts into slide decks, translate slide decks into three-sentence summaries for people who won't read the deck.

Walter Benjamin wrote about translation as a mode existing between languages, a space where meaning lives temporarily before moving into new form. He was talking about literature, but swap "languages" for "contexts" and you've basically described every modern organization. The org chart isn't a hierarchy of command so much as a vertical translation pipeline, with each layer converting information into the dialect the next layer speaks.

So here's an uncomfortable question: if you removed all the translation layers, would anything actually break? Or would you just have... faster decisions? More direct communication? The unnerving possibility your entire professional identity rests on being a necessary intermediary in a chain of necessary intermediaries, where "necessary" might mean "we've built a system requiring this" rather than "this is inherently required."²

Think about the product manager on your team, who spends maybe 10% of their time thinking about what to build and 90% explaining what you're building to different audiences in different formats. Engineers need user stories. Designers need context. Leadership needs business justification. Marketing needs launch messaging. The same information, just reskinned for each constituency. They're not being redundant or inefficient. They're doing their job. The job is translation.

Or take performance reviews, which are basically exercises in translating your actual work (messy, ongoing, context-dependent) into a narrative (clean, retrospective, achievement-focused) translating into a rating (numerical, comparative, future-predicting). You did work, then you translate the work into a document about the work, which your manager translates into their assessment, which gets translated into a calibration session discussion, which translates into a rating, which translates into compensation. At no point in this chain does anyone dispute whether the original work was even any good. They're negotiating the translation of the work.³

What you can't say

The philosopher Michael Polanyi talked about "tacit knowledge," the things we know how to do but can't fully articulate. You may know how to ride a bike, but try explaining it in words detailed enough for someone to actually learn. Most professional expertise is tacit. You've developed instincts, pattern recognition, contextual judgment. None of which exports cleanly. So when you try to share your expertise (mentor someone, write documentation, present your findings) you're attempting an impossible translation from tacit to explicit knowledge. No wonder it feels exhausting. No wonder mentorship is "so hard." You're not just explaining; you're trying to encode something existing in wordless neural networks into something representable in words.⁴

What percentage of your value comes from knowing things versus knowing how to explain things to the right people in the right way? Would you be better at your job if you knew more? Or if you got better at translation?

Training for the wrong job

Here's a bit of a perspective shift: if work is primarily translation, then the skills we should be developing aren't the ones we think we should be developing.

We treat translation as peripheral competency. Communication workshops are half-day seminars. Writing courses are optional. We hire for domain expertise and then wonder why brilliant engineers can't explain their work to stakeholders, why talented analysts produce reports nobody reads, why smart managers write emails creating more confusion than clarity.

But if your job is 70% translation and 30% domain work, shouldn't the training ratio roughly match? Shouldn't we spend more time learning to write clearly, speak precisely, listen actively, read accurately, synthesize disparately? Not as soft skills peripheral to the real work, but as the skills constituting the real work?

You probably learned your domain competence somewhere. School, training, apprenticeship. But where did you learn to translate technical concepts for non-technical audiences? To convert qualitative feedback into quantitative priorities? To take a VP's strategic vision and turn it into a team's quarterly OKRs? Probably nowhere formal. You just... picked it up. Figured it out. Developed workarounds.⁵

Which means most of us are doing our primary job function with self-taught, unexamined, unrefined skills. We're professionals who've never professionalized our main professional activity.

Your role's reality isn't the reality

There's this thing happening in meditation practice where you become aware of the constant mental translation: you're not experiencing the world directly; you're experiencing your mental narrative about the world, your mind translating raw sense-data into stories, judgments, categories. Most people never notice this layer. They think they're experiencing reality, but they're experiencing translation.

Same thing happens at work. You think you're experiencing "the actual situation," but you're experiencing your role's translation of the situation. The engineer sees a technical problem. The PM sees a feature request. The customer success person sees a user pain point. The executive sees a strategic risk. Same situation, different translations, each one shaped by professional vocabulary, incentive structures, and available tools.⁶

And here's the worst part: each translation feels complete. The engineer isn't thinking "I'm seeing only a partial, technically-filtered view." They're thinking "here's what's actually happening." We mistake our translation for the territory.

What would change if you recognized your work as translation work? If you approached meetings not as information-exchange but as translation-negotiation? If you read your colleague's email understanding they've translated their actual thoughts into professional-email-dialect, requiring reverse-translation to access original meaning?

Would you get defensive when someone "misunderstands" you, or would you recognize your translation failed to survive their back-translation? When someone seems unreasonable, would you get frustrated, or would you get curious about what translation schema makes their position coherent within their context?

Working with lions

I keep thinking about something Wittgenstein wrote: "If a lion could speak, we could not understand him." Not because lions lack intelligence or humans lack hearing, but because a lion's form of life is so foreign to ours their words would emerge from an incomprehensible context. Their language would translate experiences we've never had into concerns we've never held.⁷

You work with people who might as well be lions. Not because they're dumb (maybe, sometimes), but because they operate in different professional contexts with different pressures, metrics, and success conditions. The CFO and the designer aren't speaking different languages metaphorically. They're speaking different languages literally, where "investment" and "quality" and "success" have non-overlapping definitions.⁸

And you, the person in the middle, translating between these lions? Your value isn't your native tongue. It's your polyglottism. Your ability to code-switch. To take what the designer means by "user-centered" and express it in CFO-speak as "reduces onboarding costs." Neither translation is lying. Both are necessary. This is the work.

The question isn't whether you're doing translation work. You are. You always were. The question is whether you're doing it consciously or accidentally, skillfully or sloppily, with awareness of what gets lost in translation or with the confident obliviousness of someone who thinks they're just "communicating clearly."⁹

What if your next promotion has nothing to do with getting better at your domain and everything to do with getting better at translation? What if the people above you aren't necessarily smarter or more skilled? What if they're just better translators between more domains?¹⁰

What got translated into your current understanding of your job, and what got left out?

Everything gets smaller

Because here's what nobody tells you about translation: it's always lossy compression. Every time information moves from one context to another, something drops out. The engineer's detailed technical trade-offs become the PM's "performance considerations." The customer's frustrated story about a workflow becomes the analyst's "pain point in the checkout process." The CEO's genuine uncertainty about market direction becomes the all-hands presentation's "strategic pivot."¹¹

We don't lose random information. We lose specific kinds: uncertainty, alternatives considered and rejected, emotional texture, the full context making a decision reasonable at the time. We lose the "why" behind the "what." We lose the conversations happening in hallways and Slack threads. We lose the history making this situation different from the apparently identical situation six months ago.

This isn't anyone's fault. It's structural. Translation requires simplification. You can't forward all context to all people. But the cumulative effect is organizations operating on increasingly degraded information the further you get from the source.¹² Senior leadership makes decisions based on information passing through five translation layers, each one stripping out nuance, adding interpretation, reshaping content for the next audience. Then they wonder why their directives get "misinterpreted" on the way back down. The directives are experiencing the same translation loss in reverse.

You know what's wild? We've built an entire professional infrastructure around managing translation loss. That's what "alignment" means. That's what "communication plans" are for. That's why we have all-hands meetings and strategy docs and quarterly planning and post-mortems. We're not sharing information; we're attempting to minimize information degradation across translation boundaries. And honestly, we're not very good at it.

Your title is your translation pair

Your professional identity is your position in the translation chain. Your job title doesn't describe what you do; it describes which translation you perform. "Senior Engineer" means you translate between code and architecture. "Manager" means you translate between individual work and team objectives. "Director" means you translate between team objectives and organizational strategy.¹³

Moving up means changing which translations you perform, which contexts you bridge. You stop translating customer needs into features and start translating features into roadmaps. You stop translating roadmaps into projects and start translating projects into portfolio strategy. Each promotion is learning a new language pair.

Which explains why promotion often feels like starting over. You were fluent in your old translation pair. You'd internalized the vocabulary, the concerns, the mental models on both sides. Now you're learning fresh contexts, new stakeholder languages, unfamiliar pressures. You're a beginner translator again, and everyone can tell.

It also explains why some people plateau. They've mastered their current translation pair. They're incredibly good at moving information between their two contexts. But they can't (or won't) learn the next language pair up. Not because they lack intelligence or capability, but because learning new professional languages is genuinely hard, and at some point you have to decide whether you want to spend your energy becoming fluent in VP-speak or getting better at the translation you already perform.

Sitting with it

There's a version of this essay where I end with actionable advice. "Five ways to become a better translator!" or "How to recognize translation work in your organization!" But that would be its own kind of translation loss, wouldn't it? Taking something uncomfortable and converting it into something useful, stripping out the discomfort to make it palatable.

So instead: what if you just sat with this? What if you spent a week noticing the translation work you do? Not to optimize it or improve it, but to see it clearly. To recognize when you're reformatting information versus creating new information. To catch yourself mistaking your role's translation for objective reality. To observe what gets lost when you convert your actual thoughts into professional communication.

You might find you're more translator than you realized. You might find your value comes less from what you know and more from your ability to move knowledge between contexts. You might find the org chart less about authority and more about which professional languages each person speaks.

You might find you work in a translation machine, and you're one of the gears.

What would you do differently if you knew?