24 million characters localized: how automated workflows brought Haufe X360 to new markets

The challenge: millions of words and a complex data format

Haufe X360, a brand of Haufe Group, wanted to introduce a localized version of the American enterprise resource planning (ERP) system Acumatica to the DACH market (which is made up of Germany, Austria and Switzerland).

To do this, the team needed to adapt the entire user interface (UI) and all documentation for this region into more than 4 languages, with plans to later expand into additional international markets.

Localizing such a comprehensive and continuously evolving system posed significant challenges. This was especially true for the documentation, which included all training materials and user manuals for the ERP platform.

Manual translation is not scalable

The project would require the translation of over 60,000 UI strings, as well as 24 million characters (or approximately 4 million words) of documentation .

Initially, all the UI strings were localized manually. But when the team then tried to hand over the extensive documentation to external translation agencies, this approach failed due to communication challenges and limited experience with agency workflows.

As a temporary solution, the team manually checked and adapted texts directly in the system—a time-consuming process that was ultimately not scalable given the enormous volume of content.

Technical hurdles and lack of context

The documentation was stored in DITA-XML, a complex, referenced format specifically designed for technical content. Its tree-like structure makes translation particularly challenging, as transferring it to standard translation tools often causes formatting errors or data loss.

Attempts to integrate DeepL Translator into a traditional CAT system were unsuccessful due to the customized structure of the Haufe X360 ERP system. Additional challenges arose from missing context in the source texts, leading to significant errors such as the translation of “COD”  as “cod” instead of the intended abbreviation for ‘Cash on Delivery’.

The solution: an automated, integrated, accurate localization workflow

Haufe X360 overcame its localization challenges by collaborating closely with an automation partner and leveraging the DeepL Translator API along with custom DeepL glossaries. This combination enabled a fully automated workflow that handles large volumes of content while maintaining accuracy, consistency, and efficiency.

DeepL impresses with its high quality

The decision to use DeepL Translator was driven primarily by its exceptional translation quality. Haufe Group had been relying on DeepL internally for years, and it consistently delivered superior results. The high-quality output allowed the team to optimize machine translations to the point where manual linguistic review became largely unnecessary, significantly speeding up the localization process.

The automated DeepL workflow

An automated translation process was developed in collaboration with a transformation expert. The workflow includes the following steps:

DITA-to-XLIFF conversion and segmentation

Source DITA files are converted to XLIFF format and segmented for translation.

Data transfer via DeepL API

The segmented content is sent to DeepL API for translation

Glossary integration for consistency

A module-specific glossary (terminology database) is generated automatically and sent along with the text to the DeepL API to ensure consistent terminology throughout the translations.

Automated quality assurance

Post-translation, automated checks identify incorrect terms, which are then replaced automatically.

Final output

The resulting German DITA files form the basis for the updated wiki, ready for use by end-users.

The result: high efficiency and a foundation for global scaling

The DeepL-based workflow enables Haufe X360 to manage the localization of millions of words efficiently, cost-effectively, and with continuous quality improvement.

Relief for product management and cost efficiency

Automation has eliminated the need for a large translation team, generating significant cost savings. Employees can now focus on critical terminology work, making optimal use of their time and expertise.

A new foundation for the entire Haufe Group

The high quality of DeepL translations has led to the solution’s adoption across the entire Haufe Group. It’s now used across multiple areas, including:

• An HR automation robot
• Translation of support tickets
• A web-based translation application

Outlook and strategic impact

The next step for Haufe X360 is to further optimize its glossaries in close collaboration with DeepL experts. The aim is to map highly specific, compound technical terms and complex system functions even more consistently.

This technical foundation is what enables flexible responses to market requirements. By deploying a more efficient glossary workflow, Haufe X360 can deliver new language packs with minimal effort in response to customer requests.

Dutch, Slovak, and Czech are currently being implemented, with more to follow as needed, setting the stage for even greater future growth.