Case Study: Dakota Systems

1. Introduction

Overview:

Dakota Systems is a IT Services firm based out of Chicago, IL that develops and provides technical documentation management systems and hosting solutions for Fortune 500 enterprises.

Founded in 1999, they have developed a reputation over two decades of being a trusted and reliable firm that top enterprises trust to handle their technical documentation management, publishing, and hosting.

An early adopter of the XML language, JSON, and, now, LLM-driven semantic search and document vectorization; they provide their clients with state-of-the of art solutions for document management.

Challenge:

As Dakota opened its Shanghai office to serve Chinese customers, boutique website translation of their website content from English to Mandarin was needed. Google Translate's automatic translation feature was not sufficient enough to translate from English-to-Mandarin in a culturally acceptable format. Specifically, using only Google translate, certain phrases were not translated with the same accuracy that a human translator that has a handle on both of the languages would. Because Dakota needed to show proficient translation capabilities on its website, manual translation was needed.

Furthermore, the website needed to be able to determine when an enduser was accessing it from a browser that has it's language preferences set to Mandarin, and automatically perform the translation by pulling the translated content from the database.

Objective:

Develop a feature on the Dakota Systems website that determines when an enduser has their browser language preferences set to the Mandarin language, and translate all content on the website when signaled.

2. The Problem

Background:

As previously mentioned, as Dakota generated more business with Chinese clients, it was determined that a physical office was worth the investment. Along with this decision, a website that could serve Mandarin to potential customers was also determined to be a good investment. Because Google Translate was not accurate enough for this use case, it was determined that manual translation of all website content was needed.

Every word on the main website was translated manually by Dakota's internal staff and stored in an XML database alongside its English counterparts. The site was developed in Node.js and could query English or Mandarin content based on a signal.

This being said, the development of the logic surrounding the identification of an enduser having their preferred language on the browser set to Mandarin and the implementation of a comprehensive translation throughout the website needed to be developed.

Pain Points:

Having a grammatically correct and appealing website to potential Chinese customers was imperative to win contracts in the country. Imagine if you were a US based company procurement officer and vetting software vendors. If you came across a vendor that did not support English or had poorly translated English on their main website, it's possible that you would rule them out as a partner. Because of this, an accurate translation feature was needed on the site.

However, this translation feature needed to be robust as Dakota could also not have US based clients encounter situations where their website was automatically translated to Mandarin because of an error.

3. Our Solution

Discovery Process:

Dakota's team was very straight forward in the requirements that were needed for the development of these feature, and it was easy to lay out project must-haves as we were working with another process heavy IT Services firm. Because of this, discovery was streamlined and only required a transfer of credentials and a quick call to verify the project requirements.

Proposed Solution:

As the website was built in Node.js, we determined that including a snippet of Javascript on the client side to determine what a user's bowser preferences were was the best way to tackle the issue.

If the browser language preference was identified to be Mandarin, the website was served with the Mandarin translation, pulling from the XML database.

Technology Stack:

We conformed to Dakota's tech stack: Node.js with embedded Javascript and HTML + Tailwind CSS, and the eXist XML database for the dynamic content.

Development Process:

As we weren't touching any of Dakota's customer facing code (the code they have to maintain for customers), and the website was under development by the rest of the Dakota team during our engagement, we did not have to follow any strict QA or testing processes. Simply, we made out development changes, tested them across multiple browsers and screen sizes, and then committed and merged our branch.

4. Implementation

Challenges Encountered:

Throughout development, building a robust identifier that could pick up a enduser's language preference across multiple browsers and version proved to be complex.

Additionally, after translation, ensuring that the UI of the website maintained with new content length and sizing due to the translation was also difficult. Specifically, when an English sentence is translated to Mandarin, it can be shorter in length. This can lead to formatting issues that need to be debugged so the website doesn't look malformed upon delivery. Managing a responsive design on mobile devices was also taken into consideration.

Timeline:

The total engagement took about one week from discovery to branch merge.

Collaboration:

We engaged with Dakota through Slack utilizing huddle and Google Meets for video syncs or calls.

5. Results

Key Outcomes:

Our feature development was robust and did not require follow up or maintenance. We look forward to engaging with with Dakota as they enter into another website revamp that requires a robust GPT chatbot integration in 2025.

6. Lessons Learned

The biggest lesson that we took away internally from this project was the process of working with another technical team. When working with another technical team, a rapid adoption of their coding practices, internal processes, and communication standards is paramount. When working with a non-technical customer, you can mold the working relationship standards, but coming into a fleshed out team, you have to adapt.

7. Conclusion

Summary:

All in all, Benmore successfully engaged and delivered a custom language translation feature for Dakota Systems and looks to expand on this relationship going into 2024.

Call to Action:

If you need technical or professional services, please don't hesitate to book a call with us! Click the "Let's Talk" button again.