Your app just launched in Germany. Within hours, users report broken buttons, confusing messages, and a support ticket backlog growing by the minute.
Turns out, your translations were technically accurate, but no one tested the design inconsistencies and functional errors.
This is a translation quality assurance problem.
Poor TQA can cost you more than embarrassment. Broken localized experiences can create an overwhelmingly negative brand perception, increase the pressure on support staff, and, ultimately, tank your market entry.
Translation quality assurance (TQA) prevents these failures by testing translations against critical parameters. In this article, we’re going to unpack what translation quality assurance is and how to build a TQA process.
Webinar replay: Translation quality, quantified
Learn how automatic translation scoring, powered by AI, removes the need for 80% of review tasks.
Translation quality assurance, or TQA, is a set of processes designed to evaluate and verify the accuracy, consistency, and quality of translations.
Languages differ significantly in structure, idioms, cultural nuances, and contextual meanings. These discrepancies make translation a complex task that requires careful QA consideration of these variations to ensure the content is both accurate and resonates appropriately with the target audience.
TQA analyzes your translations against four key dimensions:
| What it tests: Grammar, spelling, punctuation, syntax, terminology consistency, brand voice alignment, and whether the translation serves the same purpose as the source text. |
Linguistic Quality Assurance (LQA) focuses on the textual integrity and semantic precision of translations. It involves scrutinizing grammar, spelling, punctuation, and syntax to ensure technical correctness.
LQA also assesses how well the translation adheres to the original text's style, tone, and intended message. And it considers the brand’s glossary or style guide in this evaluation.
Translators and project managers need to ask themselves: Beyond checking spelling and grammar, does this translation serve the intended purpose of the original word, phrase, or document?
🔖 Build your LQA program in six steps
Check out our best practices for creating your own linguistic quality assurance program from the ground up.
Marvel translated the title Guardians of the Galaxy to "星際異攻隊" — literally “Interplanetary Unusual Attacking Team.” The awkward phrasing confused audiences.
What’s worse, poor subtitle translations stripped out jokes, puns, and wordplay entirely. Local media called it unprofessional, questioning whether translators understood the source material at all.
That’s to say that while some linguistic translation discrepancies are inconsequential, others — like those of Marvel’s — have financial, reputational, and legal repercussions.
| What it tests: Layout integrity, text truncation, button overflow, UI element alignment, readability, and whether translated text works within design constraints. |
Visual quality assurance analyzes visual elements like graphics, layouts, and overall design components. The goal is to make sure all visuals are culturally appropriate and fit well within the translated content.
This approach verifies that the translated text fits aesthetically and functionally within the design parameters. You want to deliver a frictionless user experience in translated interfaces.
Consider the English phrase "Submit feedback." In German, a typical translation may be "Feedback einreichen."
Depending on the font, this German translation is approximately 50% longer than the original English phrase. If this phrase were used in a button on a website or an app, the increased length could cause the text to exceed the button's boundaries or make the layout appear cluttered, unbalanced, and even unusable.
When you consider how this applies to translating extensive amounts of text, the impact on the layout can be substantial and potentially require fundamental adjustments in page structure, menu designs, and overall navigation elements.
| What it tests: Interactive elements (buttons, links, forms), date/time formats, currency display, text direction (LTR vs RTL), input validation, and whether translated content breaks core functionality. |
Functional quality assurance (FQA) ensures that all functional aspects of a product, from its buttons and links to its overall navigational structure, operate correctly in the target language.
In the context of software, apps, or websites, FQA tests the localized version to confirm that it behaves as expected without any element breaking due to translations.
If you design an appointment booking app for a U.S. audience, the date format will be MM/DD/YYYY. However, when localized for a French audience, the date format needs to be changed to DD/MM/YYYY to match local conventions.
Without proper functional quality assurance, the app may incorrectly interpret the dates entered by French users, potentially leading to booking errors or even a loss in bookings due to user frustration.
| What it tests: Cultural norms, local regulations, legal terminology accuracy, imagery appropriateness, color symbolism, and whether content could offend or violate local laws. |
Cultural translation assurance assesses all translations against local norms, regulations, and sensitivities beyond literal language accuracy. This includes verifying that colors, imagery, idioms, and phrasing are culturally appropriate for the target market.
For example, the Turkish language has two distinct versions of the letter "I": one with a dot ("İ", lowercase "i") and one without a dot ("I", lowercase "ı"). These letters are different and are used in different contexts.
If you don’t account for these differences in your translation guidelines, it can lead to legal repercussions.
Also consider the 2012 Shanghai Maritime court case between two corporations: one Chinese and the other British.
This case centered around translation errors with significant legal and financial implications, where "drydocking" was mistakenly translated as "tank washing" and "except fuel used for domestic service" was inaccurately rendered as "except fuel used for domestic flights." These mistranslations led to protracted legal disputes between the involved parties.
Linguistic, visual, functional, and cultural quality assurance don't work in isolation. Companies that address all four dimensions simultaneously in their localization process can see dramatic improvements.
Take Withings, for example.
Withings, a connected health pioneer, improved its localization setup by integrating Lokalise with Figma. This made it easier to assign, coordinate, and track tasks among designers, developers, and translators.
By embedding translation keys directly into the design phase with Figma, the team ensured that every aspect of the user interface was considered from a linguistic, visual, cultural, and functional standpoint early in the development process.
The payoff: Withings reduced the operational burden of manually handling localization tasks and rolled out features at a 90% faster timeline. Withings also experienced a ”big impact on the quality of the user experience.”
We have improved the quality of the localized UI, ensured higher quality translations, and consistency of the overall product experience.
A translation quality assurance process operates in three phases:
This three-phase approach can reduce review time while maintaining quality standards. Automation handles repetitive validation tasks while translators focus on cases that genuinely require human expertise.
Let’s break it down.
Start by auditing your source content for issues that can cause translation problems. Think complex or confusing phrasing, cultural references that won't translate well, and technical information.
Then map out your project scope:
Clear requirements prevent scope creep and help your team focus on what actually needs translation.
Design guidelines in line with your translation expectations. This can include:
These resources create uniformity across translations in various languages and assets. Focus on clearly documenting your preferred terminology, designated voice, and symbols like ™ and ®, while also providing extensive reference materials.
⚡Here's a pro tip
Build a QA checklist that covers locale-specific requirements. Think date and currency format standards for each region, regulatory language, and technical compatibility issues (character limits, special characters).
Most translation quality assurance today happens through software. Professional translators use computer-assisted translation (CAT) tools and translation management systems (TMS) for routine QA checks.
That said, you still need human expertise for:
It all comes down to the scope of your translation projects. You might only need a translator and a TMS. For enterprise launches across 20+ languages, you’re likely looking at a complete team to prevent expensive post-launch fixes.
Translation management systems are central to TQA. They automate the repetitive validation work that would otherwise consume hours of manual review time.
Here's everything a modern TMS like Lokalise can do to level up your quality assurance setup:
Lokalise brings all of this (and more) into one platform. When a translator submits work, automated checks run instantly against your glossary, TM suggests consistent terminology, and designers see layout issues in real-time — all in the same interface where development happens.
💡Find the right TMS for your needs
We evaluated dozens of translation management systems to shortlist and review the five best options in the market. Read this guide to make an informed decision.
Translation memory (TM) plays a preventative role in translation quality assurance. It stores every approved translation in a searchable database, then automatically suggests the right translations when you use similar terms in new projects.
This creates a self-reinforcing quality assurance workflow. Translators work with pre-approved terminology instead of making independent decisions that create inconsistencies.
For example, when “Submit feedback” appears in a new feature, your TM database shows the previously approved German translation “Feedback einreichen.” This prevents terminology drift where the same English phrase gets translated five different ways.
Linguistic QA operates in two stages: automated detection catches obvious errors, then human review handles judgment calls.
Your TMS should automatically flag grammar, punctuation, and syntax errors or spelling mistakes and typos. These automated checks also reveal inconsistent terminology or misplaced variables.
For machine translation, automated quality scoring decides which MT output needs human review and which can proceed with minimal editing. And there are three ways to evaluate MT quality automatically:
Route highly error-prone translations to professional linguists who review the output against multiple parameters, including:
Quality estimation (QE) predicts translation quality by analyzing linguistic features like fluency, terminology consistency, grammar accuracy, and sentence structure. It assigns confidence scores (0-100 scale) in real-time.
Human review provides cultural judgment to identify nuanced errors QE systems miss, such as:
An efficient localization strategy combines both these approaches. That means, QE automatically scores all translations as they're submitted, then routes them for human review based on confidence thresholds.
Translations scoring 85+/100 (typically 70-80% of volume) auto-approve for low-stakes content like UI strings and help documentation. Scores of 70-84 (15-20% of volume) trigger light human review focused only on terminology verification. Scores below 70 (5-10% of volume) route to full linguistic assessment including cultural appropriateness and message preservation checks.
This hybrid approach reduces your team’s workload while maintaining quality standards. Human experts focus exclusively on translations where mistakes have real business consequences.
Visual testing verifies that translated text actually works in your UI. It looks at the translated interface to check if visual elements make sense (both aesthetically and functionally) to end users.
Here are some elements to test for visual coherence:
Lokalise integrates with design tools like Figma, Sketch, and Adobe XD to create a fast-paced testing experience. So designers can see the translated text length in real-time and adjust layouts immediately.
A linguistically perfect and engaging translation is worthless if it crashes your checkout flow and people can’t place an order. Functional testing checks whether translated content breaks any functionality.
You want to start by testing all interactive elements.
Test that buttons, forms, navigation menus, and links function properly in each language. Verify form submissions process correctly with localized input and error messages displayed when expected.
Also pay attention to date and currency formats. They should match regional standards to prevent any mental friction for your audience.
And remember to test complete workflows (signup → purchase → support) instead of checking each feature in isolation. This will allow you to catch issues where localized content creates unexpected failures across multiple steps.
Content that works perfectly in one market can cause serious problems in another. This could range from cultural offenses damaging your brand image to legal violations leading to lawsuits and penalties. Work with local legal experts and cultural consultants to test the accuracy of your translations.
For instance, a cultural consultant would review translations for cultural sensitivity. They’ll flag everything from colors and imagery to design choices and language that can come off as offensive in the market.
For regulated industries (finance, healthcare, legal services), content must meet local requirements. Work with a legal consultant to catch issues in legal documents like privacy policies, product claims, compliance, and more.
Create a structured feedback mechanism to capture detailed user interactions and responses to your localized product.
Use tools to collect:
Then, analyze this data to identify patterns and specific issues with the localized content.
You can draw on these insights to iteratively refine your translations based on actual user experiences and specific feedback. Use version control systems to best manage changes so that each iteration improves upon the last.
Before the public launch, run a final review covering all quality dimensions — linguistic, visual, functional, and cultural.
Here’s a handy TQA checklist for a final quality audit:
It’s best to run this audit cross-functionally and include translators, developers, designers, and QA testers in the process. Each perspective catches different issues, like translators can spot linguistic problems and developers flag technical discrepancies.
At Lokalise, we’re a leading localization and translation management platform that streamlines the adaptation of web and mobile applications, games, and digital content for international markets.
Our TMS platform automates translation processes, facilitates collaborative project management, and integrates with numerous tools to enhance workflow efficiency, among other key QA benefits.
We make quality assurance easy — providing QA tests in hours, not days.
Our quality assurance checks promptly test spelling and grammatical errors, inconsistencies in placeholders and HTML tags, discrepancies in URLs and email addresses, formatting concerns like whitespace and bracket balance, and much more.
That way, you can translate better, and faster. See for yourself: Try out all of Lokalise’s features for free for 14 days.
Author
Meet Rachel, our Content Manager and Lead Copywriter, who pivoted from advertising to SaaS and has never looked back.
Born and raised in the UK, Rachel has lived in London, Paris, Buenos Aires, and now Brussels. Through city-hopping, traveling, and her studies in French and Journalism, she’s picked up French and Spanish, and is now inventing her own language with help from her three-year-old daughter: Franglospanish!
Outside work, Rachel enjoys making (and eating) fresh pasta, drawing, and spending as much time as possible outside, cycling, hiking, or running.
Want to turn a piece of software written in Java into Python without rewriting it from scratch? This is now possible thanks to LLM for translation. What a time to be alive. Large language models (LLMs) like GPT or CodeT5 can now understand and translate programming languages. This helps developers migrate projects, modernize old codebases, and adopt ne
You decided to open your product for the world to discover. Congratulations! Translation is one of the next steps to open the doors of the global marketplace. One problem: you do not have an in-house translator team, and now is neither the time nor the priority to hire them. One solution: connect with vendors who will assist you in your localization projects. One better solution: discover vendor management in loc
The traditional translation management system (TMS) is dead. It doesn’t support the many moving parts of the multilingual content lifecycle, such as managing linguistic assets and syncing with design systems to integrating with codebases, automating QA, and coordinating stakeholders across functions. Modern TMS solutions go far beyond translation. They bring together translators, developers, marketers, and designers under one roof. But with dozens of transla
Behind the scenes of localization with one of Europe’s leading digital health providers
Read more Case studies
Localization workflow for your web and mobile apps, games and digital content.
©2017-2026
All Rights Reserved.
