HSBC spent roughly $10 million fixing one slogan. “Assume Nothing” was meant to signal careful banking. In several non-English markets it landed as “Do Nothing.” Customers noticed. The bank had to pull materials, reprint everything, and rebrand. The error was not a bad dictionary choice. Translators worked without seeing the final placements or how the phrase would sit next to the rest of the campaign. That is the classic LQA gap: accurate words that still fail once they leave the spreadsheet.
The same pattern shows up again and again. KFC’s “Finger-lickin’ good” became an invitation to “eat your fingers off” in early Chinese materials. Pepsi’s “Come Alive with the Pepsi Generation” turned into something about raising ancestors from the grave. Mercedes once chose characters that phonetically approximated the brand but literally suggested rushing toward death. These are not urban legends. They are documented cases where the absence of in-context review produced measurable commercial and reputational damage.
Games supply even sharper examples. The European release of Zero Wing delivered the immortal “All your base are belong to us.” The line became a meme, but the deeper problem was that rigid translations also broke UI elements and left players staring at unplayable text. Final Fantasy VII shipped with lines such as “This guy are sick.” Ghostbusters on NES closed with “Conglaturation!!! You have completed a great game and prooved the justice of our culture.” Each of these failures survived because no one with native fluency played the finished build and flagged the problems before the disks shipped.
Translation Is Not LQA
Translation moves meaning from one language to another. LQA checks whether that meaning still works when it appears inside the actual product. A translator looking at a string list cannot see that a button is only 120 pixels wide, that a variable will insert a long German compound, or that the surrounding screen makes a polite form of address feel abrupt. LQA testers run the localized build. They watch text expand, overflow, or wrap. They notice when a German or Russian string clips on a mobile button while the English original fits perfectly. Industry measurements routinely put German expansion at 20–40 percent relative to English; short UI strings often need even more room. Russian and Polish sit in a similar range. Without deliberate layout buffers or flexible components, the interface breaks.
That is why mobile applications are especially vulnerable. Screen real estate is limited, fonts differ, and users expect polished buttons and labels. A practical mobile LQA checklist therefore includes:
Running every target language on real devices of varying sizes and orientations
Checking every button, tab, and form field for truncation or overflow
Verifying that concatenated strings and variables still produce grammatical results
Confirming that date, number, and currency formats match local conventions
Looking for “tofu” characters or missing glyphs when fonts lack coverage
Testing RTL languages for mirrored layouts and correct reading order
Skipping these steps does not save time. It merely moves the cost downstream. Studies and practitioner reports consistently show that fixing a localization defect after release can cost many times more than catching it during LQA—sometimes cited in the 10x range or higher once engineering time, re-certification, support volume, and lost conversions are included. One survey of SaaS teams found that 73 percent of translation errors are first reported by users rather than internal QA.
The Efficiency Trap
Manual regression testing across languages is slow. That is the third pain point. Teams under release pressure often treat LQA as optional or reduce it to a quick linguistic spot-check. Automated tools can help—pseudo-localization that doubles string lengths, pattern matching for broken placeholders, and visual diff tools that flag overflow—but they do not replace native speakers looking at the real interface. The strongest results come from combining automated checks with human LQA performed on playable or clickable builds. Context screenshots inside the translation management system also reduce the number of out-of-context errors that reach the LQA stage in the first place.
The common thread in every high-profile failure is the same: someone assumed that clean translation files equal a clean product experience. They do not. Context disappears the moment strings leave the source environment. Length expands. Cultural register shifts. Layout constraints appear. LQA is the only stage that puts the translated product back under native eyes before customers see it.
Companies that treat LQA as a non-negotiable step avoid becoming the next cautionary tale. Providers with deep experience across hundreds of languages and decades of project work already run these checks as standard practice. Artlangs Translation, for example, has spent more than twenty years delivering multilingual work across 230-plus languages with a network of over 20,000 professional linguists. Its teams handle full translation services, video localization, short-drama subtitle work, game localization, multilingual dubbing for short dramas and audiobooks, and data annotation and transcription. That breadth means the same process that catches a German button overflow also catches a cultural mismatch in a short-form drama or a variable error in a mobile app. The disasters listed above were preventable. The only question is whether the next release will treat LQA as optional or as the final line of defense that keeps the brand out of the headlines.
