This session draws on real decision-making at a digital health platform serving a geographically dispersed Spanish-speaking user base, tracing how a deliberate, data-led strategy was built to maximize naturalness and closeness to users without producing a separate
version for every market.

When you’re localizing mental health content the stakes are raised considerably: users need to feel met where they are, not addressed from a clinical or cultural distance. That requirement shaped decisions from voice casting to written register to cultural choices. I’ll also discuss how audio and written Spanish were handled through opposite approaches and why Castilian Spanish required a distinct four-stage workflow of its own.

The session closes with a frank look at what AI integration and budget pressure have since changed and the honest gap at the center of all of it: we made these decisions on behalf of users whose feedback we never actually collected.