How bad software localization can ruin your products

Picture the scene. You and your team have just spent months, or even years, developing a groundbreaking piece of software. You’ve spent countless hours testing it to breaking point, making sure the UI works perfectly, and that it has all the functions your customers need – from inception to alpha and beta testing, you’ve worked diligently to make sure this is the best product you could possibly release. It’s finally ready to launch, and you send your new creation into the world, ready to amaze your end-users.

Then the first complaint comes in. And then another one. And another one. Overnight, a flood of complaints comes in from all four corners of the world.

The Spanish version isn’t correctly localized for South American forms of Spanish, so those users can’t understand it. The German version has text overspill on important graphics and UI, so they miss out on key information. And you’ve really annoyed the British English users by missing the ‘u’ in colour. Your amazing software has suffered from a lacklustre localization process, so it doesn’t matter how good it is – if your users can’t understand how to use it, or if the localization actually damages the software’s usability, then your software documentation has harmed the success of your software.

While this may seem like a nightmare scenario, for some software providers, this has been an all too common occurrence. In this article, we are going to look at some examples of when bad localization has scuppered the success of software from very well-known and respected companies, and what lessons you can learn so that you don’t repeat these mistakes.

Grand Theft Awful Localization

If you ever played a video game on any console, you’ve likely heard of Grand Theft Auto. Rockstar’s premier series is up to its 5th incarnation, which just so happens to be the most profitable entertainment product of all time. The internet pretty much stopped for a week when GTA 6 was announced for release in 2025. So, you can imagine how rabid gamers anticipated the remastering and rerelease of some of the classic titles, GTA3, GTA Vice City, and GTA San Andreas back in 2021. It seemed like an easy win – take these well-loved and critically acclaimed games that had already been created (and localized successfully), update them so that they run better on new hardware, release them to the adoring public, and retreat to your mansion to count the piles of money like Scrooge McDuck.

Unfortunately, this was not the case. Even before its release, there were hints that these games had overly relied on MT tools. Pre-release, the Korean version of the Trilogy was rated as ‘Youth not allowed’, rather than the 12+, 15+, or 19+ ratings usually used in the South Korea rating system. Once released, the Grand Theft Auto Trilogy was plagued by bugs, many caused by moving these games into a completely new game engine, but some caused by poor localization efforts.

In-game storefronts, seen throughout the city environments of all three of the remastered games, were mismatched in size, causing issues with text fitting onto textures. Even worse, most of the names were puns, and poor localization meant that, at best, these puns no longer made sense, or at worst, meant something completely different in the target language. There were plenty of other issues reported with the poor translation of subtitling text for the game’s cutscenes, causing mission objectives and game fail states to be unclear, or muddling the game’s plot. This was reportedly caused by an over-reliance on MT and AI-based tools to help with the game’s text generation and upscaling of old assets.

While many of these issues were fixed with later patches, the damage was unfortunately already done. With a Metacritic score of 1.0, the lowest of any Grand Theft Auto game, the trilogy managed to sell 14 million copies. This may seem like a lot, but it pales in comparison to the near 170 million units sold of Grand Theft Auto 5, and has tarnished the series’ reputation with gamers.

So what lessons can we learn from this example of software localization? While we recommend the use of MT and AI-based tools to help with large localization projects, these tools need to be backed up by a robust QA process – something we recommend as part of a good software localization process. Also, part of the screening during a QA process should look for common issues, such as text size changing graphical formatting between different language versions and how cultural and geographical factors change the meaning of certain words and phrases between different cultures, as well as languages. Using human-backed QA can help easily spot and remedy issues such as this.

Microsoft lessons in localization

Next, we look at a company even larger and more ubiquitous than Rockstar games – Microsoft. Anyone who has ever owned a computer in the last 30 years has no doubt used a Microsoft operating system or software package, and even die-hard Apple users occasionally dip their toes into Microsoft programs. However, despite its massive global presence, Microsoft hasn’t been above making the occasional and costly software localization error.

There are numerous mistakes that Microsoft has made over the years (and they have helpfully been catalogued in this Guardian article) ranging from the somewhat harmless, such as accidentally reversing the South Korean flag in localized versions of its software, to the potentially insulting, such as using European Spanish translations in Central American countries, leading to potentially insulting female users at a gender selection screen, to the politically sensitive, such as references to Taiwan, Kurdistan and Kashmir regions in maps, which caused many issues for Microsoft in China, Turkey, and India respectively.

While Microsoft was able to fix these issues (and notably, has now set up in-house geography training courses to educate staff), they have undoubtedly cost the company in terms of sales, reputation damage, and the costs and manpower needed to fix these respective issues. For example, the inclusion of Kashmir as a separate region, illegal in Indian law, meant that Windows 95 was withdrawn from sale in India. This cost Microsoft hundreds of thousands of sales. When Windows 97 was later released, with an updated map for India, it sold over 100,000 copies, demonstrating the appetite for Microsoft products in the region.

So what can we learn from these examples? Again, as with our previous example, having a QA process that involves native speakers who are aware of the differences in cultural and political situations of your targeted regions can help you quickly and easily spot these issues before release, and recommend appropriate fixes before your software goes live. This may seem a time-consuming and resource-intensive task, but it’s much easier and cheaper to spot these mistakes before releasing your software than suffering the reputation damage and cost of patching your software post-release.

Software Localization done right

So what can we learn from these examples of software localization gone wrong? Well, they demonstrates that even the largest, most successful companies are not immune to issues caused by bad software localization. No matter how many resources you have, if you don’t have a well-planned and executed localization strategy, your software product could fall flat in overseas markets – or worse, end up damaging your brand credibility, insulting your users, or risk litigation by unknowingly breaking local laws.

A good software localization process involves a robust QA process, using native speakers to back up technologies such as MT and AI translation tools that can give you the advantages of quicker, more efficient translations, while simultaneously weeding out the kind of errors these tools can inadvertently make.

If you’d like to find out more about our recommendations on good localization practises, or are looking for advice on your latest localization project , why not get in touch with our team here, or read more about our recommendations for a good localization process in our blog ‘Localization to drive growth in new markets: when and how’, or if a using a pivot language is right for your project ‘When should you turn to a Pivot language?’