To overcome these issues and build an effective localization strategy, it was essential to understand the back-ground, analyze the challenges, solve the problems, and review the final performance of the localized apps.

These four main areas needed to be improved as quickly as possible to ensure that new features and languages could be launched quickly, and with sufficient quality. And, with very limited resources for the localization team, prioritization was essential.

The first step was to create a standardized process. The consensus at the time was that localization means just doing translations, and that these translations should be delivered the next day. There was no agreement on quality standards and no way to determine whether the translations were good or bad.

DiDi created a process for translation, review, localization quality assurance, and content editing. They also created a language style guide and quality standards. This helped the internal stakeholders better understand localization and have confidence in a process that didn’t involve engineering resources.

The second step was to improve launch speed. At the time, DiDi was in a period of accelerated expansion, and supporting this growth was of the utmost importance. Since supporting the launch of a new language took up to several months, DiDi needed to build an internal proxy to connect Git with Memsource, a third-party translation management system, to automate translation submissions and the delivery process. Eliminating the need for engineers to manually copy and paste would save a great deal of time.

The third step was to improve the efficiency of the localization workflow for the engineers, language specialists, and localization project managers. DiDi needed to create a website that would free engineers from hassling with manual translation edits, prevent quality errors due to manual editing, and reduce the communication costs of product managers throughout the entire process.

One of the main challenges was that engineers needed to do code refactoring to support translation search, edits, and auto release function. Code refactoring involves editing and cleaning up previously written software code without changing the function of the code in any way. Of course, this takes time. Balancing the short-term and long-term costs and benefits and convincing engineers to do code refactoring now to support long term global expansion was not easy. The pros of not doing it meant that engineers had more time in the short term to spend on feature development. However, the cons meant that every time DiDi needed to update a translation, the engineers needed to be involved.

In the end, DiDi decided to do code refactoring, sacrificing short-term efficiency but supporting long-term scalability. The team performed a pilot test to prove the code factoring worked.

The internal localization platform provides semi-automation capability — a translation editing function — and has more modules (from previous code refactoring) that connect with the platform. The second stage of the localization platform provides full automation capability (translation editing plus release automation) and integrates with the rest of the modules. A full 90% of the modules finished with the code refactor were integrated to the platform, which helped free the engineers from translation work.

The final step was to ensure not only that the language quality was accurate, but that it also sounded natural to the target audience and was user-friendly. A regular review and localization quality assurance process was established to ensure UI translation quality. The DiDi internal localization content management system provides UI descriptions, screenshots, character limits, and generates reference files that enable linguists to have more context.

After implementing these processes, DiDi built a solid localization system to ensure that new languages and features are launched quickly and efficiently.

Internal stakeholders understand more about the localization process and turnaround time, and urgent requests have decreased dramatically. The style guide was approved and is used by local legal, PR, and marketing teams.

Overall, localization launch time was reduced by up to 40%. The localization content management system saved engineers 185.5 days of manual work time, and saves localization project managers over 60 hours per month. The translation fix time has been reduced from several weeks to one business day. The localization quality assurance projects fixed 2700+ linguistic issues across three main locales within six months.

Lessons from DiDi’s localization journey can help other Chinese companies that plan to enter the global marketplace.

The first hurdle for Chinese companies is adopting an inter-national mindset. Localization teams need to continuously educate internal stakeholders to think about internationalization and localization rules, cultural differences, differing design standards, and so on when deciding to launch a new product.

As business grows, a number of legacy localization problems will need to be fixed. Since many Chinese companies only decide to go international after establishing themselves in the Chinese market, the codes, product features, and designs are not very user-friendly for different locales. Tolerating reduced efficiency in the beginning to fix legacy issues was important and necessary. It allowed the company to set up their infrastructure, processes, and quality standards, providing DiDi localization the ability to scale and automate without any snags.

DiDi has now achieved a higher level of localization maturity, and is well-equipped to deliver high-quality global products, content, and user experience.