Built-in Translations Localization System

A powerful system for automating translations and localizing product support, enabling companies to scale globally with ease.

MY ROLE
Product strategy
User experience
Design thinking
UI Design
TEAM
Product managers
Head of engineering
Head of product
Design team
TIMELINE
Jul 2022 - Nov 2022Released
CONTEXT
Mavenoid is a B2B SaaS company that provides an AI-powered product support tool. Previously, the platform relied on a third-party software tool to manage localized translations within the app. While the tool was functional, users reported that it reduced efficiency and negatively impacted the overall user experience.
IMPACT
+12
New enterprise and expansion deals achieved.
+70%
Customer cost reduction with automation and glossaries.
$50K
Savings in third-party software annually in expenses.
Overview
Mavenoid’s AI-powered Virtual Assistant automates product support by providing accurate answers to customer questions. Translation and localization are crucial for helping businesses expand globally and offer smooth customer experiences in any market.
Spanish
English
German
This is the view presented to the end-user.
Problem
What is the problem?Many hardware brands need to provide support in multiple regions and languages, but they often lack in-house agents who speak those languages. This makes self-service the only feasible option for customers. The challenge is that localising support content for different languages is time-consuming, expensive, and inefficient. Currently, businesses spend thousands of dollars each year on translations, and the process can take weeks with a lot of back-and-forth coordination—especially when content is dynamic and frequently updated.
‍
Why is it a problem?The translation process is slow, costly, and resource-intensive. Delays in providing localized content mean customers may not get timely support, leading to frustration and lower satisfaction. Frequent updates to support content also create ongoing operational inefficiencies and increase the cost of maintaining multilingual support.
Solution
We plan to address this challenge by integrating an AI-powered translation engine into the Mavenoid Admin app. This will allow brands to quickly localise support content at a fraction of the cost of human translation, using DeepL Pro for high-quality translations in minutes. The AI will accurately handle even specialised terms.
‍
Additionally, brands can still involve human translators for review or new translations, offering flexibility and control. This approach will streamline the translation process, cut costs, and enable faster delivery of localized support content.
Research
After conducting a series of interviews, I uncovered users’ pain points, challenges with the third-party tool, and key blockers, which included the following key user stories:
Thomas
Implementation Specialist

"I feel blind when publishing with the third-party software because I don’t know what translations have been completed; we need a way to verify everything before publishing."

Torsti
Project Manager (Customer)

"I find it challenging to confirm if text and visuals work correctly in the assistant since there’s no way to preview it with the third-party tool."

Hugo
Customer Success

"I’m often unsure before publishing about what’s been machine translated and needs approval, what’s manually unapproved, which languages are unapproved, or how changes to the original source will impact other languages."

Key features
The design process consisted of two phases: first, developing a solution that could be quickly and efficiently implemented to meet the core requirements, followed by adding incremental enhancements to refine the feature.
We introduced auto-translate capabilities to simplify translations and added verification functions for translators to review and approve.
Bulk actions were added to speed up tasks like selecting keys and applying functions such as approving and auto-translating, saving hours of time.
To enhance task management, we added a summary feature for an overview of incomplete languages and quick filtering.
Users struggled to identify key sources, so we added a breadcrumb component to improve usability and accessibility.
We complemented the tool with a “find and replace” functionality that allowed users to replace phrases or words in bulk, saving hours and simplifying the editing process.
A glossary feature was added to outline preferred terminology and approved translations, reducing costs by eliminating re-translation.
Finally, we added a key verification panel for users to review and verify keys before publishing, an improvement users had strongly requested.
User testing
User testing identified opportunities to improve efficiency. In response, I focused on refining navigation, addressing usability challenges, and ensuring the prototype was complete, feasible, cost-effective, and aligned with user goals.
Torsti
Project Manager (Customer)

"I find that integrating the localisation view with the main canvas could allow previewing or jumping to nodes for each key, checking translation status directly from the canvas, and quickly accessing and translating any untranslated keys."

By integrating translation management within the canvas, users could access it without leaving the page and work on project-specific keys only.
Ideation
It was essential to create a solution that could be easily translated across different areas of the app, seamlessly integrated, and require minimal effort for engineers to implement quickly, given the project’s complexity, while focusing on the MVP (Minimum Viable Product).
Results
New/expansion deals: easy translation management, and coverage of 60+ languages have contributed to winning dozens of new enterprise deals and expanding with existing customers. Currently, ~30% of all conversations are in a non-primary language, which was much lower (less than 10%) before the TMS. Our TMS is now actively used by 50%+ of live orgs, and most of them have 5 or more language

Lower costs:
1) Our own TMS replaced the previous system that relied on a third-party app, which was costly for Mavenoid ($35-50k/year). We saved these expenses after migrating to our own TMS.

2) Thanks to automated translations and the use of glossaries, many of our customers reduced their translation costs by 70% or more, positively influencing their decision to renew their Mavenoid subscription.

Effort/speed of translation: thanks to the smooth UX of our TMS and features like “find & replace”, “glossary” and “bulk actions”, translation tasks that used to take multiple hours of effort and 1-2 weeks to complete can now be done in under 15 minutes. Again, a massive retention factor for our customers.
Customer story
Husqvarna Group
AI Lab Director

"Mavenoid gives us the ability to scale automated support to all our markets, without losing the local and personal touch Husqvarna is known for each key, checking translation status directly from the canvas, and quickly accessing and translating any untranslated keys."

NEXT PROJECT

Integrating ForceManager with Microsoft Teams

See project
Get in touch!
leonkibuka@gmail.com
Designed by León Kibuka