.png)
Duolingo Conversations
CONTEXT
Increase Duolingo users’ accessibility to real-world conversation practice in their target language with a new feature called Conversations. This case study was part of a project as a DesignLab student.
Overview
BACKGROUND
Duolingo is a popular language-learning app with more than 500 million registered users. Despite Duolingo’s success, long-time users can’t help but feel discouraged that they aren't making progress when it comes to speaking their target language.
THE CHALLENGE
Duolingo users are frustrated that they're not able to have a conversation in their target language.
Duolingo users feel that the vocabulary used in the lessons isn't applicable to real-life scenarios. They also feel that having a conversation with someone is the best way to practice speaking a language.
THE SOLUTION
Give language learners real-world conversation practice by matching them with native speakers.
The conversations feature randomly matches users with native speakers.
PROJECT CONSTRAINTS
Inability to implement the solution.
This case study was for a course assignment at DesignLab, so I didn't work directly with Duolingo or have access to their developers.
Keep branding intact.
I used brand elements, such as their fonts, color palettes, and illustration guidelines, found on the Duolingo website, during the design process.
Language-focused feature.
I set a constraint for myself to create a feature that's specifically centered around language learning.
Create an MVP.
Due to time constraints, I created an MVP of a new feature.
The Process
Discover
RESEARCH GOALS
Determine Duolingo's current standing in the language-app market.
Understand Duolingo user's needs and frustrations when learning a new language.
COMPETITOR ANALYSIS
Competitors offer immersive features, such as conversation practice, videos of native speakers having a conversation, and music videos.
Duolingo is the only app (out of the four below) that offers gamified lessons, however, Duolingo does not offer immersive lessons. While Duolingo does offer a "Roleplay" feature for Duolingo Max users, this feature is no substitute for a conversation with a native speaker.
.png)
BABBEL
FLUENTU
MEMRISE
YOUTUBE
USER INTERVIEWS
Participants felt frustrated that they weren't able to have a conversation in the language they were learning.
I interviewed five participants, and all five didn't feel confident having a conversation in their target language using Duolingo alone. The majority of participants mentioned that having a conversation with others would help them learn, but several participants mentioned that they don’t always have someone to practice with.
.png)
Define
AFFINITY MAP INSIGHTS
Theme 1: Learning from scratch vs. learning as a refresher
-
80% of participants were learning a language from scratch on Duolingo for a trip abroad or to speak with friends and family*
-
60% of participants wanted to use Duolingo as a refresher after taking a formal language course in college or high school*
Theme 2: Inability to have a conversation
-
80% of participants did not feel confident about having a conversation in their target language
-
60% of participants didn’t have anyone to talk to
-
-
80% of participants felt that speaking with real people is the best way to practice
-
60% of participants felt that the vocabulary and phrases in lessons weren’t applicable to real-life situations
*Three participants fit into both categories
4/5 participants used Duolingo to learn a new language from scratch
3/5 participants used Duolingo as a refresher after taking a formal language class
4/5 participants didn't feel confident about having a conversation in their target language
3/5 participants didn't have anyone to practice having a conversation with
4/5 participants felt that speaking with another person is the best way to practice
3/5 participants felt that the vocabulary and phrases in the lessons weren't applicable to real-life scenarios
PERSONAS
Based on the insights I gathered from the user interviews and affinity mapping, I discovered that the participants fell into one of two groups: language learners who use the app to learn a new language from scratch and those who use Duolingo as a refresher. Since the majority of participants (80%) were learning a new language from scratch, I decided to focus my efforts on solving the problem for this persona:
.png)
THE PROBLEM
Duolingo users are frustrated that they're not able to have a conversation in their target language.
Duolingo users do not feel that the current lessons and features are sufficient for learning how to have a conversation when learning a new language from scratch on the app. Users feel that having a conversation with someone is the best way to practice learning a language, but they don't always have anyone to practice with.
Ideate
IDEATION PROCESS
Once I determined the problem, I used the “creative constraints” technique to brainstorm solutions for Florence. After turning on a five-minute timer and writing down as many ideas as I could think of, I decided to go with “Duo Discourse” (which was later named Conversations).
Feature alowing users to chat with others learning the same language.
Language Forum
Feature that matches users with a tutor who is fluent in a language.
Language Tutor
Users sign up for group video calls led by a facilitator and practice speaking with eachother.
Conversation Groups
Match users with a native speaker and exchange voice messages
Duo Discourse
STORYBOARDS
To help build empathy for Florence and convey how she'll interact with “Duo Discourse,” I created a storyboard from her perspective. The storyboard helped me further connect my research findings to the design solution that I devised.
Florence Using Duo Discourse
Florence recently connected with distant family members who live in Italy. She planned a trip to visit them and started learning Italian every day on Duolingo leading up to the trip. When visiting with them on a trip, she felt frustrated that she wasn’t able to communicate with them because she didn’t speak Italian and they didn’t speak English.
Although Florence used Duolingo every day for a year, she felt that she was mostly just learning vocabulary related to animals. She was frustrated that it wasn’t helping her learn conversational phrases that would actually help her communicate with her family. Her family wasn’t always available to practice with her, so she wasn’t making much progress.
One day, Florence got a notification on her iPhone from Duolingo. The notification said that there is a new feature that matches users with a native speaker of the language they’re learning so users can practice their conversation skills and learn more applicable phrases.
.png)
Florence was curious about this new feature, so she opened up the app and tapped Duo Discourse under the “Practice” tab. After pressing “Get Matched,” she was matched with Fred, a native Italian speaker who is learning English.
Fred introduced himself, told her about his day, and talked about his family in Italian, which helped Florence learn new phrases to use while speaking with her family. She practiced responding back to him and gave him some pointers on English. After a few months of having conversations with Fred, Florence felt confident about having a conversation in Italian.
During her next trip to Italy, Florence was thrilled that she was able to have a conversation with her family in Italy. She felt that their connection grew stronger since they were finally able to communicate.
Design
USER FLOWS
To determine how Florence would interact with the new Duo Discourse feature, I created the user flow below.
TASK FLOWS
After creating the user flow, I created a task flow to show the most direct way Florence could use the new Duo Discourse feature in the Duolingo app.
Low-Fidelity Wireframes
Next, I used the task and user flows as a guide to sketch out the necessary screens for Duo Discourse. I made these wireframes more detailed so I could conduct user testing with them. After sketching out the flow for the first time, I iterated on the screens to come up with alternate layouts.
USER TESTING WITH LOW-FIDELITY WIREFRAMES
I wanted to validate my solution to Florence's problem before investing time in designing high-fidelity mockups, so I conducted user testing with five participants using low-fidelity wireframes. My goal was to determine if the information architecture and task flow made sense to Duolingo users and confirm that they would find this feature useful.
RESULTS & ITERATION RECOMMENDATIONS
Participants were confused by the name "Duo Discourse," the way to start a conversation, and the purpose of the "Your Matches" section.
Participants validated the idea of the conversation feature and my proposed location within the app’s current sitemap. However, the name (Duo Discourse), the way in which users started a conversation, and the purpose of the “Your Matches” section were confusing to participants. Based on these results, I determined that I needed to iterate on the name, clarify how to start a conversation, and clarify the meaning of the “Your Matches” section.
HIGH-FIDELITY MOCKUPS
The next step in my design process was to incorporate the insights that I gained from user testing and bring the low-fidelity wireframes to life in Figma. To approximate the look and feel of the current Duolingo interface and branding, I referenced their brand guide and screenshots of the app.
Test
I conducted unmoderated usability tests with eight participants on Maze. Each participant completed two tasks: trying out the conversation feature and returning to a conversation.
RESULTS
100%
Completion Rate
9.2/10
Satisfaction Rating
UNEXPECTED RESULTS
"What do I do next?"
1.
38% of participants wondered what to do next after "recording" their message.
2.
"How do I earn XP?"
25% of participants wondered how they earned XP while using the conversations feature.
Clicking on the profile card.
3.
38% of participants clicked on the profile card to return to a conversation instead of the "chat" button.
4.
Issues with Maze.
Maze did not detect the "delay interaction" that included in my Figma prototype.
HIGH-PRIORITY REVISIONS
1. Add step-by-step help text for first-time users
Some usability test participants were confused about what to do after pressing the microphone. I added step-by-step instructions for first-time users to guide them through their first time recording a message. I used the Duo Owl design paired with help text so it matches common design patterns found in the app.
Before
After
2. Clarify how XP is earned
Some usability test participants were confused about how they could earn XP with the new conversation feature. I added text to the “Welcome to Conversations” page explaining that users earn 20 XP for every 10 minutes of conversation and added an “XP progress” bar showing how much XP the user has earned after sending a message.
Before
After
3. Make the entire profile card a "touch target"
Some usability test participants preferred tapping the “profile card” instead of the chat button to navigate back to a conversation. I added a way for the user to navigate to a previous conversation by touching the user profile card.
Before
Default
Pressed
After
Default
Pressed
Next Steps
ACCESSIBILITY CONCERNS
Some of the colors used throughout the Duolingo app do not meet accessibility standards, which could prevent some users from being able to use the app. Since this could negatively impact the business, making the app 100% accessible would be an imperative next step. Due to time constraints, I was not able to address these issues within the app.
FUTURE FEATURES
During usability testing, participants mentioned that they’d like to see a “text” feature included when having a conversation. After fixing the accessibility issues with the app, I'd add this feature to the chat window.
Reflection
CONSTRAINTS REVISITED
Inability to implement the solution.
Since I didn't work with Duolingo for this project, the new Conversations feature isn't found on the app today.
On-brand feature.
Thanks to the simplistic nature of the illustrations and icons on the app, I was able to re-create the elements that weren’t in their brand guide (i.e. icons and buttons).
Language-focused feature.
Setting this constraint helped me narrow down the user problem to solve for this project.
Create an MVP.
Keeping the features simple helped me finish the project within my 80-hour timeframe.
LESSONS LEARNED
1. Using Maze for Usability Tests
My biggest takeaway from this project was that Maze does not accurately record results if “delay interactions” are used in the Figma prototype. The Maze report for the usability tests stated that the screens after the “Matching” screen were “Not reached.” From now on, if I need to test a flow with complex interactions, I’ll conduct moderated usability tests or try out another unmoderated usability testing website instead.
2. Importance of brand voice in UX writing
It’s just as important for the copy in a design to align with a brand as the design elements. The original name that I came up with for the new feature (Duo Discourse) didn't align with the brand and confused users. I ended up using the name “Conversations” because it was a one-word name that mimicked the other exercise names in the "Practice Hub."