Making translation more reliable and accessible for language learners

Imagine trying to learn a new language like Mandarin, where the characters are unfamiliar, and you find yourself dependent on audio aids just to keep up. For many English speakers, this reliance on audio is not just a preference but a necessity, often leaving them lost when audio is unavailable.

DeepL Translator is an AI-powered tool renowned for its highly accurate and natural-sounding translations across 29+ languages. This project aims to enhance DeepL's freemium user base by developing a mobile-first, desktop-compatible feature that significantly improves translations of Asian languages.

Problem Statement:
English speakers learning Asian languages such as Mandarin, Japanese, or Korean face significant challenges in comprehending written characters due to an overreliance on audio-based learning methods.

The Outcome: A transliteration feature integrated into DeepL's home translation page empowers users to seamlessly view romanized Asian characters, addressing a critical pain point for language learners. This enhancement is expected to boost user engagement, increase tool accessibility, and drive conversions from freemium to paid tiers.

Anticipated

Results

+15%

Total Translation Inputs

+30%

Daily Active Users

+1%

Conversions

+12%

Completion Rate

By adding transliteration, I expect increased engagement, especially among Asian language learners. This feature also aims to boost completion rates by 12% and drive a 1% increase freemium-to-paid conversions, indicating greater perceived user value.

The appproach to the problem

To comprehensively inform design decisions, I conducted a dual-pronged research approach. First, I explored language learners' study habits and preferences to identify key challenges. Simultaneously, I leveraged ComScore's data to gain insights into DeepL's user base, business model, and platform performance. This holistic understanding allowed for efficient resource allocation and proactive problem-solving.

Finding the User, discover the problem

Informed by initial research, I conducted surveys, interviews, and usability tests with fourteen qualified English language learners to deeply understand their challenges and inform design solutions. A key finding was the emphasis of audio in their learning process. Participants described audio as indispensable for immersion, background learning, and focused study. Recognizing the critical role of audio, it's essential to consider the implications of restricted audio access on the learning experience.

Insight: Vocabulary


A robust vocabulary is essential for effective sentence construction. Limited word choice hinders learners' ability to express themselves clearly and comprehensively.

Insight: Audio is needed


Participants heavily rely on audio to learn pronunciation, especially when visual cues like romanization are absent.

Insight: No Audio No Translation


The absence of audio significantly impacts task completion rates, leading to increased frustration and premature session termination.

Problem Statement:
English speakers learning Asian languages such as Mandarin, Japanese, or Korean face significant challenges in comprehending written characters due to an overreliance on audio-based learning methods.

Through extensive research, I discovered that learners' reliance on audio served both as a crutch and a barrier. This realization made it clear that a solution wasn't merely an enhancement—it was essential for helping users engage more deeply with the written language.

Collaboration with the team: I consulted with senior designers and engineers early in the process to uncover key insights into feasibility, prioritization, and efficiency, all of which shaped the design direction.

Looking at competition for solutions

Traditional machine translation has focused on accuracy, especially for English-based language pairs. However, the intricate nature of Asian languages presents distinct challenges. To identify potential gaps in DeepL's offerings and explore improvement opportunities, I conducted a comprehensive analysis of competitor capabilities.

By pinpointing these gaps, I can optimize resource allocation and prioritize solutions that directly address user needs. Given positive user feedback on the current UI, the focus will be on seamlessly integrating enhancements rather than a complete overhaul.

The User & Business Strategy

User Need: Audio is essential for user satisfaction and retention. To mitigate high user churn rates caused by audio impairments or unavailability, I aim to significantly enhance the overall experience in audio-free scenarios.

Business Needs: A subpar audio experience can diminish user satisfaction and retention, reducing application usage and hindering the conversion of freemium users to paid subscribers.

The Impact: The transliteration feature could enhance user satisfaction with DeepL as a translation tool, potentially leading to increased user acquisition and retention. It could also improve DeepL's reputation as a leader in accurate and user-friendly machine translation.

Finding the solution with designs

To optimize resource allocation and ensure feasibility, I partnered closely with full-stack engineers to assess the viability of design concepts. This collaborative approach prevented investment in technically impractical features.

Recognizing the potential of a transliteration feature to enhance the translation experience, I developed multiple design concepts. To optimize time, I narrowed these down to three promising designs and conducted user testing to identify the most effective solution.

Validation

To validate the transliteration feature's effectiveness, I conducted user tests with the same participants. From an initial set of diverse design concepts, I refined my focus to three core designs based on early user feedback. Each concept was crafted to balance user needs with technical feasibility. Another focus was on visual clarity to enhance the learning process. Through iterative testing, I identified the design that best met the goals—one that seamlessly integrated transliteration without disrupting the user experience.

LOOKING FORWARD

The transliteration feature not only meets immediate user needs but also sets the stage for future enhancements, such as integrating tonal pronunciation. This project deepened my understanding of how nuanced user behavior can guide meaningful innovation. Moving forward, I aim to build on this experience by exploring further opportunities to blend AI technology with user-centric design, creating tools that are as empathetic as they are effective.

Previous
Previous

Google Gemini

Next
Next

Criterion Channel