When I started localizing the app into English, Spanish, German, French, Italian, Swedish, and Polish, I expected translation work. I did not expect localization to expose so many product decisions.

Here are the lessons that mattered most.

A short sentence is not automatically clear

The first temptation was to translate every English string and then trim anything that wrapped onto a second line.

That is layout work, not localization.

A sentence can be short and still be wrong for the situation. “Watch the baby” is brief but vague. “Stay close enough to see the baby's face” gives the parent an action. In a high-stress interface, verbs and observable cues matter more than raw character count.

Our rule became: one card, one decision, one next action.

Tone is part of the interface

New parents already receive plenty of warnings. If every screen sounds urgent, users stop noticing the guidance that really is urgent.

We separated the voice into four levels:

• an everyday idea should feel encouraging;
• a safety reminder should be direct;
• “call the doctor if” should be unmistakable;
• an emergency instruction should not compete with anything else.

This hierarchy had to survive translation. The goal was the voice of an experienced friend: gentle, calm, and specific. Not cute when safety matters, and not clinical when a dad simply needs an activity for five quiet minutes.

Units reveal hidden assumptions

Temperature was the obvious example: Celsius alone is not enough for the United States, and Fahrenheit alone is not enough elsewhere.

The deeper lesson was that products quietly assume a region everywhere:

• temperature and weight units;
• date and week formatting;
• punctuation and button width;
• healthcare language;
• which family roles a sentence treats as “normal.”

We chose to show both temperature systems where confusion would be costly and let iOS choose the default language, with an in-app language switch and English fallback.

Navigation labels need to be tested as actions

A label may be a correct noun and still be a poor button.

For example, a parent on the dashboard is not looking for an abstract content category. He is choosing between “Quick overview” and “Full guide.” Those are actions with a clear cost in time.

This mattered even more in translation. We reviewed labels by asking: if you saw only this button, would you know what happens next and how long it will take?

The content model must support localization

Hard-coded paragraphs make translation slow and inconsistent. A stage-based content model made it easier to keep the same structure across languages:

1. what is happening now;
2. what dad can do;
3. safety;
4. call the doctor if;
5. what is coming next.

That structure also improved the English version. Localization forced us to separate reusable intent from the exact wording.

The best feedback is situational

“Does the translation sound good?” produces polite answers.

Better questions are:

• Would you understand this while holding a crying baby?
• Which word sounds like a medical promise?
• Which button makes you hesitate?
• What would a new dad in your country expect to see here?
• Which sentence can disappear without losing safety?

Those questions are now how I review copy.

Best Dad is free, iPhone-only, and has no paywall. I am the maker, and I am looking for honest feedback from parents and product people in every supported language:

https://apps.apple.com/us/app/best-dad-baby-guide-for-man/id6792512850

The app is a practical companion, not a replacement for a pediatrician, midwife, or local medical guidance. If you try it, tell me where the interface still makes an exhausted parent work too hard.