Writing Custom Prompts for Better Translations

Generic translations sound generic. "Click here" becomes a technically correct but lifeless translation in every language. The fix? Tell the AI about your app.

XCStrings Translator lets you set a custom prompt for each project. This prompt gets sent with every translation request, giving the AI context it desperately needs.

Try the Configurator#

Play with the options below to see how different settings affect your prompt:

This is a productivity app that helps users manage tasks and stay organized.

Our users are everyday people of all ages.

Use a warm, friendly tone that makes users feel welcome.

Why Context Matters#

Without context, an AI translator makes assumptions. It doesn't know if "Feed" means a social media feed, an RSS feed, or animal feed. It doesn't know if your app is for kids or executives. It guesses—and often guesses wrong.

A good custom prompt answers three questions:

1. What is this app? A fitness tracker? A note-taking app? A game?
2. Who uses it? Developers? Children? Business professionals?
3. How should it sound? Casual? Professional? Playful?

Writing Your Prompt#

Keep it short. You don't need an essay—a few sentences work fine. Here's a real example:

This is a meditation app for busy professionals.
Use a calm, reassuring tone. Keep sentences short.
Avoid technical jargon.

That's it. Three lines that make every translation better.

What to Include#

App description. One sentence explaining what your app does. This helps the AI understand domain-specific terms.

Audience. Who uses your app? Age group matters—translations for kids should use simpler vocabulary.

Tone. Casual? Formal? Friendly? Playful? Pick one and stick with it.

What to Avoid#

Being too long. The AI reads your prompt for every single string. Keep it focused.

Conflicting instructions. "Be casual but also professional" just confuses things.

Obvious stuff. You don't need to say "translate accurately"—that's implied.

Setting It Up#

In XCStrings Translator, open your project settings and find the Custom Prompt field. Paste your prompt, save, and you're done. Every translation from that point forward uses your context.

The prompt is stored in .xcstrings-translator/config.json, so you can commit it to git and share it with your team.

A minute spent writing a good prompt saves hours of fixing awkward translations. Your German users will thank you.

Related#

• AI Translation Tips — More ways to improve translation quality
• Team Workflows — Share your custom prompt with teammates
• Custom Prompts Documentation — Technical reference