Localisation testing (also written localization testing, or L10n testing) checks whether a translated website, app or software product works correctly and feels appropriate for a specific language and market. It examines more than the words: testers look for broken layouts, untranslated text, incorrect dates or prices, awkward language, right-to-left display problems and features that fail when the locale changes.
For example, a checkout might look correct in English but cut off its German “Continue to payment” button, reject a valid local postcode or display a price with the wrong separators. Translation alone will not uncover all these issues. Localisation testing evaluates the actual product experience before a regional release.
This guide explains the types of localisation QA, a step-by-step workflow, realistic defect examples and a practical checklist that product, marketing and translation teams can use.
What is localisation testing? A simple example
Imagine a UK online shop launching a French-language version for customers in France. The translations might be accurate, but the site could still have a mix of English and French navigation, poorly formatted euro prices, forms that assume a UK postcode or a translated button that overlaps another control on mobile.
The localisation test asks: does the French experience read correctly, display correctly and function correctly with the intended locale? It checks the build, not simply the standalone translation file. A local-language reviewer and a functional tester may both be needed.
If you first need the broader concept, see What Is Localisation?. If you are planning the adaptation of an application, see our separate software localisation guide. This article focuses on testing and release quality assurance.
Localisation testing vs translation, internationalisation and linguistic QA
| Activity | Primary question | Typical output |
|---|---|---|
| Translation | Does the target-language text convey the intended meaning? | Translated strings, documents and approved terminology |
| Localisation | Does the content and product suit a named language and market? | Adapted text, assets, formats and locale-specific features |
| Internationalisation (i18n) testing | Can the underlying product support different languages and conventions? | Engineering defects such as hard-coded strings or non-Unicode assumptions |
| Linguistic quality assurance (LQA) | Is the text correct, natural and consistent in its product context? | Language defects with suggested corrections |
| Localisation testing (L10n) | Does the localised product work, look and read properly in the target locale? | Functional, visual, linguistic and locale-specific defect reports |
These activities are connected, not interchangeable. Internationalisation should be considered early because architectural issues are often expensive to repair after translation. Microsoft distinguishes functional validation, visual validation and linguistic validation when describing testing of localised products. See Microsoft’s localisation testing guidance.
Types of localisation testing
1. Linguistic and in-context testing
A reviewer who understands the source, target language and product checks meaning, spelling, tone, terminology, terminology consistency, plural forms, gender agreement, instructions and contextual fit. A string can be correct on its own yet wrong when shown next to a button or a warning. In-context review catches that difference.
2. Visual and layout testing
Testers review headings, buttons, navigation, tables, mobile layouts, image text, line wrapping and text expansion. German or Finnish copy may be longer than the English source; some East Asian scripts have different font and line-breaking needs. A translation is not ready if customers cannot read or interact with its controls.
3. Functional and locale-data testing
This verifies that forms, filters, search, sorting, account settings, checkout and other key journeys work with the chosen language and regional settings. Currency display, decimal separators, names, telephone numbers, addresses, time zones and date parsing should follow product requirements. Formatting and validation are separate: a field may show the correct date format but still reject the value on submission.
4. Right-to-left (RTL) and bidirectional testing
For Arabic, Hebrew and other RTL-language interfaces, check layout direction, mixed numbers or Latin-language product codes, punctuation and icon behaviour. Some interface elements should mirror; others, such as brand marks and many media controls, may not. RTL testing needs realistic text and interactions rather than simply changing text alignment.
5. Cultural, media and accessibility review
Assess whether imagery, idioms, humour, support details, subtitles and content are suitable for the intended market. Verify that the localised interface retains accessible names, readable labels, keyboard operation and appropriate screen-reader reading order. A market-specific legal or regulatory assessment may also be needed, but localisation QA by itself does not certify legal compliance.
Real localisation testing examples: defect and expected result
For a practical gaming-specific workflow, see our game localisation developer guide, with examples of broken placeholders, plural rules, text expansion and runtime testing in Unity and Unreal Engine.
The following are illustrative test cases, not claims about any named company’s software. Use them as templates for your own testing.
| Test case | Defect a tester might find | Expected result |
|---|---|---|
| German mobile checkout | Long translated checkout button is clipped at 320px screen width | Full meaning remains visible or accessible; button works |
| France: price display | A total appears with ambiguous separators or the wrong currency symbol | Values and currency are accurate and presented for the agreed locale |
| Japanese account form | A form rejects a valid Japanese name or address because it assumes UK fields | Input rules support the intended audience and business rules |
| Arabic settings page | Mixed Arabic and Latin serial numbers display in the wrong reading order | Text, numerals and punctuation read correctly; layout direction is usable |
| Spanish purchase flow | Cart contains untranslated English error messages | All user-facing messages use the intended locale, subject to approved brand terms |
| Polish notification | Wrong plural form appears when an item count changes from 1 to 5 | Plural message matches the language and actual count |
| French password reset | Email link returns to the English site unexpectedly | The user stays in the intended locale unless switching is intentional |
| English–UK date entry | A date such as 04/05 is interpreted contrary to the specified UK format | Displayed and submitted date match the correct calendar date |
Localisation testing process: seven practical steps
Step 1: Define the markets and what success means
List language-region combinations such as fr-FR, de-DE or es-MX, plus devices, supported browsers, release dates and critical user journeys. Specify whether the product uses a language selector, automatic locale detection or both. A country may need more than one language; a language can serve multiple markets. Agree the expected behaviour rather than guessing.
Step 2: Prepare source content and a testable build
Ensure the source strings, terminology, screenshots, product context and translated resources are up to date. The build should identify its version, language and environment so defects can be reproduced. Missing or outdated strings are easier to fix before reviewers spend time testing the wrong version.
Step 3: Run internationalisation checks and pseudolocalisation
Before spending effort on every translated locale, expose technical weaknesses with pseudolocalised text. Google’s Android documentation describes two useful examples: en-XA, which expands and accents text, and ar-XB, which simulates right-to-left behaviour. These can reveal clipped labels, hard-coded English, incorrect string concatenation and directionality problems. Read Android’s pseudolocale guidance.
Pseudolocalisation is a screening tool, not a replacement for human linguistic review in an actual target language. A pseudo-RTL build may reveal an engineering issue but cannot assess whether real Arabic sounds natural.
Step 4: Test the critical user journeys
Run the same core acceptance tests in the default and target locale: registration, login, search, product selection, purchase, password reset, help and contact forms. For apps, also test notifications, offline content, deep links and app-store or device-specific flows where relevant. Include smaller devices and realistic input data.
Step 5: Complete in-context linguistic and cultural review
Have a suitable regional-language reviewer inspect the actual screens, emails, error messages and support text. Ask them to record the original string, where it appears, why it is wrong and a suggested correction. Distinguish a critical meaning error from a stylistic preference so the engineering and content teams can prioritise fixes.
Step 6: Log, prioritise, fix and retest defects
Every actionable report should include the locale, build, device or browser, reproduction steps, expected and actual behaviour, a screenshot when appropriate, severity and assigned owner. Retest each fix and related user journeys. Avoid treating “all translations delivered” as equivalent to “ready for release”.
Step 7: Sign off and monitor after release
Set agreed exit criteria, such as no open critical localisation defects, all core journeys passed for priority markets and acceptance from the language owner. Monitor new releases because features, prices and strings change. Add regression checks for previously fixed issues.
Localisation QA checklist before a release
For an education-specific example of multilingual QA, see the e-learning content localisation guide, including assessments, SCORM completion, caption review and learner acceptance.
Use this as a compact, repeatable checklist. Record Pass / Fail / Not applicable for each relevant combination of locale, device and user journey.
| Area | Checks to complete |
|---|---|
| Language resources | Correct locale loads; no unintended source-language strings; approved terminology used |
| Meaning and tone | Buttons, warnings and instructions are accurate in context; plural and gender forms work |
| Layout and fonts | No text clipping or overlap; required characters render; small screens remain usable |
| Navigation and links | Menus, language switchers, emails, downloads, redirects and deep links lead to intended locales |
| Numbers and money | Currency codes, symbols, decimals, thousand separators and calculations behave correctly |
| Dates and times | Input, output, time zones and date interpretation follow stated product rules |
| Forms and identity | Names, addresses, postcodes, telephone numbers and keyboard input support intended users |
| RTL and mixed scripts | Reading direction, icon treatment, mixed-script punctuation and screen-reader behaviour checked |
| Media and culture | Images, alt text, subtitles, local contact details and market references reviewed |
| Core functions | Registration, search, checkout, login, notifications and errors work in the target locale |
| Quality and release | Defects have owners; critical issues resolved; fixes retested; release sign-off recorded |
For a wider pre-launch website audit that also covers planning, content and post-launch tasks, use our website localisation checklist.
How to record a localisation bug correctly
A useful defect report describes an observable problem rather than simply saying a translation “looks wrong”. Here is a fictional example.
| Field | Illustrative entry |
|---|---|
| Locale / build | de-DE / staging build 1.8 |
| Device / viewport | Mobile browser, 360px wide |
| Location | Basket → Checkout → Payment |
| Steps | Switch to German; add an item; open checkout |
| Actual result | Translated primary button label overlaps the price total |
| Expected result | The button and price total remain readable and usable without horizontal scrolling |
| Severity / owner | High priority if checkout cannot be completed; front-end team |
| Evidence | Screenshot, page URL and translated string identifier |
| Retest | Check fixed layout at narrow sizes and verify checkout still functions |
Where a defect could materially mislead users—for example, a wrong price, medical warning or legal consent message—prioritise it with the product owner and relevant specialists. A minor typography preference should not automatically receive the same severity.
Tools used for localisation testing
The right tool depends on the product, team and languages; no single application replaces a reliable test strategy.
- Translation management systems: Organise language files, versions, reviewer comments and terminology. They support localisation workflows but do not establish that a live website works.
- Pseudolocalisation: Exposes hard-coded strings, text expansion and some RTL defects before final translation; common in software and Android testing.
- Browser and device testing: Checks responsive layouts, fonts, keyboard interactions and key journeys across relevant environments.
- Automated tests: Verify functional parity and key calculations, navigation and forms across locales where the application supports such tests.
- Screenshot and visual comparison: Helps teams find cut-off content or unexpected layout changes; human review is still needed.
- Issue trackers and linguistic QA sheets: Record reproducible defects, translation context, suggested fixes and retest status.
Automated tools are useful for repeatable checks; human reviewers remain important for idiom, tone, cultural suitability and subtle meaning differences. Microsoft recommends in-context review after earlier linguistic validation, particularly for new languages. See the Microsoft testing framework.
Who should perform localisation testing?
A small project may combine roles, but it is helpful to distinguish responsibilities. Developers fix hard-coded strings, formatting logic and RTL defects. Functional testers verify flows against acceptance criteria. Translators or in-market linguistic reviewers judge terminology and contextual meaning. Designers address text wrapping and interface adaptation. Product owners define which issues are release blockers.
The approach should match the project’s risk. A static information page may need a focused content and visual pass. An account-based product handling payments, healthcare information or legal decisions warrants deeper testing, suitable specialists and independent legal review where necessary.
Common localisation testing mistakes
- Testing only the translated text file: Misses UI context and functional failures.
- Using a single “Spanish” or “Arabic” locale for every market: Hides country-specific assumptions and terminology.
- Skipping narrow screens: Longer labels often cause mobile-only defects.
- Assuming formatted dates and valid dates are the same: Display and parsing can disagree.
- Relying on machine output alone: Automation cannot reliably judge local tone or every meaning risk.
- Calling localisation QA legal approval: Potential compliance issues still need a qualified legal assessment.
- Failing to retest: A language fix may introduce another visual or functional defect.
Localisation testing FAQs
What is the difference between localisation testing and linguistic testing?
Linguistic testing concentrates on wording, accuracy, tone, terminology and context. Localisation testing also checks presentation, functionality, regional formats and key customer journeys. Linguistic testing is usually one part of the wider localisation QA process.
What is the difference between localisation testing and internationalisation testing?
Internationalisation testing checks whether the underlying software can support different languages and regional conventions. Localisation testing checks the actual adapted product for a named locale. Catching internationalisation problems early reduces repeated defects across translated releases.
Is localisation testing the same as translating a website?
No. Translating produces target-language text. Localisation testing assesses whether the localised website actually reads, displays and operates as intended. A fully translated website can still fail localisation tests.
Can localisation testing be automated?
Many repeatable checks can be automated, such as string coverage, some screenshot comparisons and critical functional journeys. Human language and cultural review remains important. Automation is a complement, not a substitute for in-context judgement.
How much localisation testing is enough?
It depends on supported locales, product changes and business risk. At minimum, confirm the priority locale and device combinations, critical user journeys, language accuracy and release-blocking issues. More complex or regulated services need broader coverage and appropriate specialists.
What is pseudolocalisation testing?
It uses altered but readable stand-in strings or simulated RTL behaviour to reveal text expansion, untranslatable strings and directionality defects before the real translation is complete. It is particularly useful in software development.
What information should I give a localisation reviewer?
Provide target locale, audience, product build or URL, source and target text, screenshots, style guide, glossary, intended use, access to a realistic test environment and a process for logging corrections.
Planning a multilingual release?
Begin with the target markets, an agreed language style, critical user journeys and a repeatable testing checklist. For background on creating a localised product, read What Is Software Localisation?. For cross-market content planning, see What Is Localisation?.
If you need help scoping professional translation, language adaptation or an in-context linguistic review for your project, contact TranscribeLingo about your requirements. Describe the languages and the assets you have; confirm the exact QA or engineering services needed in the project scope rather than assuming every type of software testing is included.
Further reading
- Microsoft Learn: How to perform localisation testing — functional, visual and linguistic validation.
- Microsoft Learn: Internationalisation testing — preparing software for different languages.
- Android Developers: Test with pseudolocales — expanded strings and simulated RTL behaviour.
- W3C: Internationalisation best practices — language direction and locale-sensitive data formats.
