Building for people whose cultural context differs from your own can become mystical very quickly.
We start talking about “taste” as if it were a passport some people are born with: you grew up between cultures, you spent years in Silicon Valley, you have the right accent, so you understand. Everyone else is supposed to find the right cultural oracle and trust their instincts.
I do not believe that.
Cross-cultural understanding is not magic. It is a practice.
Culture is not a spreadsheet, and people are not a dataset. But culture does leave observable patterns: vocabulary, rituals, taboos, humor, memory, status signals, and rhythms of attention. Those patterns can be studied, documented, tested, and revised.
That does not make culture simple. It makes humility operational.
Once you accept this, building across cultures stops looking like a mysterious talent and starts looking like work: difficult, imperfect, learnable work. The method travels, even when the context does not. What helps a team learn one market can help it approach the next — as long as it begins again with curiosity instead of copy-pasting yesterday's conclusions.
I keep returning to three principles.
PRINCIPLE 01
Input Determines Output
You cannot build convincingly for a culture you encounter only through a market report.
Culture lives in small signals: the tone of a subway ad, the layout of a supermarket aisle, the punchline on a late-night show, a meme dropped into a group chat, the way a brand replies on X, or the texture of everyone's timeline before a major holiday. These details rarely survive intact inside a slide deck.
Research matters. Interviews matter. Data matters. But they are abstractions of lived experience, not substitutes for it.
If a team wants to build for a different cultural context, that context has to become part of its daily sensory diet. Use the products people use. Follow the creators they follow. Read the arguments, not just the trend reports. Notice what makes people laugh, what makes them cringe, and what they do not bother explaining to outsiders.
Most importantly, build with people who carry that lived context.
Their value is not that they can translate a culture into a neat list of rules. It is that they can disagree with you in high resolution. An instinctive reaction to a line of copy, impatience with an interaction, or discomfort with a visual choice may reveal something no formal report could capture.
Do not treat those people as cultural interpreters stationed at the edge of the team. Treat them as fellow builders. Ask, listen, push back, and stay in the conversation long enough for the real question to appear.
Immersion does not make you native. It makes you less ignorant — and more aware of what you still do not know.
PRINCIPLE 02
Deliberate Annotation Beats Vague Caution
Cross-cultural work requires an antenna. A line of copy, an icon, an interaction, or a social post can carry a meaning its creator never intended.
But awareness without structure turns into anxiety. Everyone becomes vaguely careful, vaguely self-censoring, and vaguely afraid of making a move. The result may avoid obvious mistakes, but it rarely produces anything alive.
The better move is to turn uncertainty into a living map.
Which words shift meaning across communities? Which images carry historical baggage? Why does a joke work on one platform and fail on another? What associations can a color, gesture, or metaphor trigger in a particular context? Which questions require local judgment rather than a universal rule?
Write the answers down. Attach context. Record disagreement. Revisit them when the culture moves — because it will.
AI can help here. It can generate alternative readings, compare connotations, surface possible blind spots, and help a team ask better questions. But it should not be treated as the final cultural authority. Models learn from cultural traces; they can also reproduce bias, flatten subcultures, and express uncertainty with undeserved confidence.
Use AI to widen the field of attention. Use people with lived context to make the call.
Over time, these annotations become shared memory: part of the team's references, prompts, critique, and review habits. The goal is not a frozen rulebook. It is a system that keeps learning.
When the risks become visible, the space outside them becomes easier to explore with confidence.
PRINCIPLE 03
Mistakes Are Part of the Path
No builder is naturally immune to cultural blind spots. Even teams deeply embedded in a culture misread tone, identity, history, and each other.
That is not an excuse for carelessness. A public mistake can embarrass a company, but it can also hurt real people. Responsibility begins before launch, not after backlash.
Still, “never be wrong” is not a workable product strategy. It rewards silence, punishes experimentation, and teaches teams to hide uncertainty instead of learning from it.
A better standard is to make every mistake legible:
- Own it without becoming defensive.
- Understand why the team did not see it.
- Record what changed as a result.
- Turn the lesson into a better default for the next decision.
An apology can be fast. A useful postmortem should be slower. The real proof is durable change.
A team that acts, listens, and corrects itself will get closer to cultural fluency than one that mistakes caution for understanding.
CODA
Just Try
I often think about a reflection Feng Ji shared during the development of Black Myth: Wukong, one of my favorite games in recent years.
He described progress as a series of attempts: features that looked obvious until they failed, ideas that felt brilliant in discussion and awkward in the product, and ambitious prototypes that repeatedly ran into walls. Uncertainty was not evidence that the work was misguided. It was part of making something the team had never made before.
“Just try. It won't kill you.”
That is how I think about cross-cultural work too.
Study seriously. Build with people, not around them. Make the risks visible. Correct mistakes without losing momentum. Then try again — with better inputs, a sharper antenna, and a little less certainty than before.