Adding a new feature to

Duolingo

PROJECT OVERVIEW

My encounter with Duolingo was a true discovery. It transformed the language learning process into a playful experience rather than something tedious. However, my motivation took a steep dive after being a loyal user for a long time. When I learned about the objective of the UX Academy's second capstone project, which involved adding a new feature to an existing product, the desire to enhance this application immediately came to mind."

Role: UX/UI focused on research and testing

Date: 23 / 05 / 2023

Tools: Figma, Photoshop, Canva, Chat GPT, Google Meet.

Topics covered: Setting research Goals, Competitive Analysis, Interviewing users, Affinity mapping, Research Analysis, Persona creation, Information Architecture, Prototyping, and Usability Testing.

“Implementing its sessions was a little bit tricky. […] I learned that Duolingo wasn´t quite transferable to real life.”

Josh E. Talking about his trip to México.

FIRST THINGS FIRST

I studied languages ​​to connect with people different from me and to learn distinct interpretations of everyday life. In my case, Duolingo stopped being attractive when I couldn’t put what I had learned into practice with real people, but I needed to know the experience of other users, so I made a research plan where I established research goals and expressed my hypothesis:

“Adding a social feature that permits Duolingo users to talk to each other would help users practice conversation, learn everyday vocabulary, stay motivated, and engage longer with the application.”

As part of my secondary research, I delved into Duolingo´s competitors. I downloaded their apps and dived into them to learn about differences, resemblances, and the features that position them as rivals to our app.

After using them for about a week, I had enough information to conduct a S.W.O.T. competitive analysis and determine the strengths, weaknesses, opportunities, and threats of each one.

LISTENING TO PEOPLE

I wanted to learn about the experiences of other Duolingo users, so I enlisted the help of eight volunteers to conduct interviews. I aimed to delve into their journeys as second language learners and their relationship with the app, using open-answer questions to gain insights into their experiences and any challenges they encountered.

Once all the data was collected, we processed all the information and made an affinity map.

PRODUCT OF OUR RESEARCH

As had already happened to me, four of the five volunteers interviewed confessed to having lost motivation. Gamified activities had helped them keep on, but their primary goal for using the app was improving their communication skills and getting them into practice. Somewhere along the way, they had felt frustrated.

The trade-offs of Duolingo were:

Positive aspects of Duolingo:

  • Interviewees liked the structured lessons in small segments.

  • The gamification and storytelling elements made learning more enjoyable.

Negative aspects of Duolingo:

  • Some interviewees felt the lessons did not adapt well to real-life situations.

  • Lessons were perceived as not adequately preparing users for real-life language usage.

  • Some users stopped using the app mainly due to the lack of improvement in pronunciation and fluency.

One persona

With so much data collected, we needed a representation of our user type, so we created John, a second-generation Korean-American who would like to speak the language of their grandparents more fluently. John would help us think with empathy, understanding users’ needs, motivations, and frustrations. I would get in John´s shoes when the time to solve the question arrived.

One HMW question

The inquiry to be resolve was clear:

How can we make Duolingo users to not lose motivation so they continue practicing and complete the entire language learning program?

The Answer

Although when I started the project, I thought that the solution could be to transform Duolingo into a social community, I had learned that people did not come to the app looking for friends. Still, instead, they just wanted to practice with someone.

I brainstormed possibilities and decided that adding roleplays was the best idea. The roleplays would allow them to practice everyday situations safely and thus be trained for real life.

These conversation opportunities would be set within Duolingo´s linear path and unlocked as they advance through the units. The difficulty would increase in proportion to their level.

From lower to higher difficulty, users could:

  1. Solve simple problems by chatting with a bot

  2. Solve a simple problem by chatting with another user

  3. Solve a more complicated problem by chatting with another user

  4. Solve problems through audio messages with another user

  5. Solve problems through voice calls with other users

These roleplays would be presented as a prize and won´t be mandatory. Once unlocked, they can be accessed from a chat icon in the main menu. We will design an interest form to pair students by affinity, preferably.

THE ARCHITECTURE

Starting from the current Duolingo sitemap and considering the established hierarchy among its elements, we placed the new ones at the level we thought they should belong.

Once everything looked good, we determined the critical task flows and analyzed in detail every step and action in them to make sure everything was correct.

A INITIAL CONCEPT

Finally, we came up with an initial concept, ready to be tested and hopefully iterated to make it more usable and adapted to users.

THREE TASKS TO TEST

I got a prototype ready to be tested. New volunteers would have to perform the below tasks so the design success could be measured and iterated to improve it.

  • TASK FLOW 1:

    Find the chats icon and start a conversation

  • TASK FLOW 3:

    Complete your interest form and send a friend request.

  • TASK FLOW 2:

  • Unlock role-play and start a conversation

TIME TO TEST!

I created a plan to document the approach and methodology for our usability test. This involved recruiting five new volunteers, formulating key questions to answer, and preparing a structured usability test script to ensure none of the critical questions went unaddressed.

USABILITY TEST RESUTLS

The test was a success. All the volunteers achieved all the tasks.

ITERATIONS

The tests led us to make six essential changes in two rounds: The first was immediately after our first volunteer, which worked as a pilot test. He was overwhelmed with task three, so I realized little changes were ned before continuing with the other volunteers.

Interest form

We added an explanatory note to this form to inform the user of its purpose.

We also removed the possibility of adding new interests so it would be easier to match users.

Filtering information

We reduced filtering options to ease the cognitive load.

The chat text box (I)

We hid the conversation suggestion under a help icon.

The chat text box (II)

Our volunteers loved the new chat functionality but only used it when we directly suggested it. We decided to redesign it, making it more intuitive and visible.

Matched users cards

In the suggested user´s cards, we substituted the request button for the possibility of to message and added more information about the user´s performance in the app.

Filter for chats

We added the possibility of filtering chats to help users quickly find roleplays, so they can practice whenever they like.

OUR FINAL PRODUCT

MEASURING SUCCESS

What WERE the users´ needs?

  • Lessons that adapt to real life

  • Practice in real situations

  • Practice pronunciation better

They said they would use A.I. to practice

They would practice conversation with other students

What is the result?

  • Users can now practice roleplays that simulate real-life situations in a safe learning context. Depending on their level, they would practice with A.I. or other students.

  • The new functionality is not designed to help specifically with pronunciation.

FINAL THOUGHTS

    • We need to investigate which situations language students want to practice so that learning specialists in different languages can determine at what points of the curriculum to put each roleplay.

    • We need to research how to display all the roleplays on one screen.

    • A.I. must be trained to detect cyberbullies/trolls and determine an excellent system to reward good practices.

    • We must find new ways to make the Duolingo community interact more.

    • The needs of the interviewees can be very different, and it is not always possible to design something that satisfies everyone.

    • Sometimes it is better to specialize instead of covering all the possibilities.

    • Time constraints require better planning and more substitutes, especially when we are dependent on the schedule of others.

    • The first to do a usability test should be someone close to correct small details that may have escaped.