A UX Writer’s Checklist for Localizing UI Copy in Screenshots and Mockups
A UX Writer’s Checklist for Localizing UI Copy in Screenshots and Mockups
A button can sound perfect in a translation file and still fail on screen. Its label wraps onto two lines. Its tone feels too sharp beside a friendly illustration. Its meaning changes because the next screen asks for something unexpected. Reviewing localized UI copy means checking the words where people will actually use them, with the surrounding design, interaction, and task in view.
That review needs more than a spelling pass. It needs a repeatable way to check onboarding screens, error messages, and button labels before a developer turns a mockup into a working experience.
The Review Lens: Read each localized screen as a person trying to finish a task. Check whether the copy explains the next step, sounds appropriate, and fits without hiding meaning. Then inspect the states a polished mockup may leave out: loading, errors, longer names, larger text, and missing information. Give developers approved strings with clear context and layout expectations. A translation is ready when its meaning and behavior hold up inside the interface.
Start With the Task Behind Each Screen
Before reviewing individual phrases, write down what the person needs to accomplish. “Create an account” gives you a better review target than “check the welcome screen.” The task tells you which details matter and which words can be removed.
For each screen, identify what happened before it appeared, what the person already knows, and what happens after the primary action. A translator who sees only “Continue” cannot tell whether the button opens another setup screen, submits personal details, or starts a subscription.
Include the source copy, target locale, screen purpose, and relevant interaction notes in the review package. Name the locale precisely when regional choices affect wording. A language label alone may leave questions about spelling, formality, or terminology unresolved.
Give Reviewers a Usable Screen Set
A single screenshot captures one moment. A useful review set shows the sequence around it. Include the previous screen, the current state, and the next destination. Add error and empty states where they change the wording.
Remove or replace personal information before sharing screenshots. Use realistic sample content, though. An address field filled with “Test” tells you little about how the layout handles a longer address.
Review Onboarding for Clarity and Trust
Onboarding often asks people to make decisions before they understand the product. Localized copy should explain enough to support those decisions without turning each screen into a manual.
Check that the headline, supporting text, and action label tell one coherent story. A headline about personalizing recommendations should not lead into unexplained marketing consent. If a step is optional, make that visible in the localized version.
Read permission requests with particular care. The copy should explain what access supports and what happens if the person declines. Confirm that the explanation matches the actual product behavior. Friendly wording cannot repair a misleading promise.
Keep Tone Consistent With the Moment
Brand voice should guide the translation without forcing the same level of enthusiasm into every interaction. A welcome message can feel warm. A request for sensitive information needs restraint and precision.
Ask a reviewer familiar with the target locale whether the wording feels natural in context. Direct translations of jokes, casual greetings, and idioms may carry a different mood. Formality also needs a consistent approach across headings, instructions, and buttons.
Read the full sequence aloud. An isolated sentence may sound fine, while several screens together reveal sudden changes in politeness or repeated instructions.
- Purpose: Can the person explain what this step helps them do?
- Choice: Are required and optional actions easy to distinguish?
- Expectation: Does the copy describe what happens after the action?
- Tone: Does the language fit both the audience and the request?
- Consistency: Do recurring features and actions use approved terminology?
- Exit: Can the person tell how to skip, return, or leave when those options exist?
Check Text Embedded in Screenshots and Mockups
Some review materials arrive as flattened images. That makes it harder to find every string or compare wording across screens. Text may sit inside an illustration, a sample notification, or a product preview rather than an editable text layer.
Resources that help teams translate images can support work with text embedded in screenshots. Treat the resulting wording as material for review. Compare it with the original image and confirm that small labels, punctuation, and text near icons have been read correctly.
Translated copy still needs an in-layout review because an image alone cannot show how the interface behaves. A phrase may fit in a static screenshot but wrap when a browser renders it with the production font. A label may look acceptable until someone increases the text size or opens the screen on a narrower device.
Place the approved wording into editable components whenever possible. Review it at its intended size, beside the actual controls, with the correct spacing. If the screenshot contains a product example, check that its translated content still makes sense for the feature being explained.
Test Text Expansion Without Cutting Meaning
A translation can need more space than its source. Short labels deserve close attention because a compact source phrase may become several words in another language. Character count can help flag possible pressure points, but it cannot prove that a string will fit.
Inspect the rendered text in the intended font. Look for clipped letters, awkward line breaks, overlapping controls, and supporting text pushed away from the field it explains. Check narrow layouts and larger text settings as separate review conditions.
Agree on What Can Change in the Layout
When copy does not fit, decide whether the component can grow, the text can wrap, or the wording can become shorter. Make that decision with the designer and developer. Removing a useful word should not be the automatic response to a cramped component.
For example, a button container may support a longer label at full width. A row of two buttons may need to stack vertically. A heading may work across three lines if the spacing adapts. These are layout decisions that affect the reading experience.
For right-to-left locales, inspect reading order, alignment, and directional elements in the rendered screen. Mixed content, including email addresses and numbers, also needs a visual check. Record any issue against the specific component and state.
Make Error Messages Useful Under Pressure
Error messages appear when someone is already interrupted. Check whether the localized message explains what went wrong and offers a useful next step. Avoid wording that blames the person or claims certainty the system does not have.
“Payment failed” leaves several questions open. A message that explains the available recovery action can help, but only if that action is valid for the actual failure. Do not tell someone to change card details when the product only knows that a connection timed out.
Review the message beside the field, banner, or dialog that contains it. Confirm that a field error identifies the right input and that any instructions match its localized label. If the message mentions a button, use the button’s exact approved wording.
Check Recovery Copy Against Product Behavior
Ask what happens after the suggested action. Does retrying preserve entered information? Can the person go back without losing progress? Does the support route exist in the target locale?
Those answers shape the copy. A promise that “your changes are saved” needs confirmation from the product team. When behavior is uncertain, flag the missing information instead of polishing an unsupported claim.
Read Button Labels as Commitments
A button label makes a promise about the next interaction. Review that promise against what the control actually does. “Save” and “Submit” may both complete a step, but they set different expectations about where information goes.
Read primary and secondary actions as a pair. Their meanings should remain distinct after translation. Check dialogs especially closely when one option deletes content, abandons changes, or creates an obligation.
Keep labels concise without making them vague. If a longer label is needed to clarify the action, test a layout adjustment before removing meaning. Avoid shortening words into unfamiliar abbreviations just to match the source button’s width.
Give Developers Context They Can Build From
A marked-up screenshot helps locate a problem. An editable string record helps implement the fix. Connect the two with stable string identifiers and a clear record of which wording is approved.
Include enough context to prevent reasonable guesses from becoming product defects. A developer should be able to tell where a string appears, what triggers it, and how its container should respond when the text grows.
- String Identity: Record the string key, target locale, approved copy, and matching screen or component.
- Interaction Context: Describe the trigger, destination, and relevant success or failure state.
- Layout Behavior: State whether wrapping, component growth, stacking, or truncation is permitted.
- Dynamic Content: Explain placeholders, possible values, and any required plural forms.
- Related Copy: Connect field labels, error messages, and actions that must use matching terms.
- Review Status: Name the reviewer, record unresolved questions, and identify the approved version.
Keep instructions separate from display copy. Notes about placeholders or layout should not be easy to mistake for text that belongs in the interface. Avoid splitting a sentence into fragments merely to reuse words. Give localization reviewers enough control to produce a natural sentence.
Finish With Screens People Can Actually Use
Before signing off, walk through the localized flow in an interactive build or prototype. Follow the happy path, trigger an error, go back, and try again. Check the smallest supported layout and a larger text setting.
Record issues with the locale, screen, state, observed problem, and expected result. “The secondary action is clipped at this width” gives the team something concrete to fix. “The translation looks wrong” does not.
Your final check is simple: can someone understand the request, predict the action, and recover when something fails? When those answers hold up on screen, the copy is ready for handoff. The checklist has done its job by connecting language to the experience people will use.