A More Immersive Duolingo

Introducing new features to enhance the language learning experience.

Eliza Edwards
7 min readApr 30, 2021

A two-week design sprint in a team of four UX Design students: myself, Alice, Midori and Hugo. We agreed as a group that we all wanted to be involved in every step of the process, but Alice and Midori would focus more on research while Hugo and I would spend more time on prototyping and developing the UI.

The Brief

“As of 10 February 2021, the language-learning website and app offered 106 different language courses in 38 languages. The app has over 300 million registered users across the world.”

The goal was to create an ‘Immersive Experience’ within the Duolingo mobile application that would allow users to practice ‘real life situations’.

User Research

Sending out a screener survey helped recruit eight users of Duolingo to interview. We put our insights into an affinity map and, based on this, discerned two types of user — casual and serious.

The casual user — wants to be able to communicate with natives when on holiday, so their priority is to practice speaking key words and phrases.

The more serious user — needs to be able to speak more fluently if, for example, they live abroad. Their priority is to practice speaking full conversations.

Both want to be able to practice speaking the language, but currently feel there aren’t enough opportunities to on Duolingo.

Competitor Analysis:

Interviews revealed that users felt Duolingo is more casual and playful than some of its competitors such as Rosetta Stone and Busuu, something that they enjoyed. I found that users felt the illustrated characters, animations and gamification involving rewards made using the app fun and lighthearted rather than pressurising. In fact, one interviewee said this was the reason she used the app over Busuu.

However, while it is more playful, users also reported that Duolingo was less immersive than some of its competitors. Five out of our eight initial interviewees said their learning on Duolingo felt decontextualised, and they couldn’t practice listening and speaking in a realistic scenario. All our interviewees felt that immersive learning is important, with one user calling it ‘fundamental to all learning’. Creating a feature inventory for Duolingo’s competitors revealed some of the features that make them more immersive including:

  • The ability to choose a realistic scene to practice
  • The chance to practice pronunciation
  • The chance to learn about cultural highlights
  • Being able to find native speakers to practice with

Design Studio:

With these features in mind, we conducted a group design studio to generate some ideas, which involved sketching under time pressure then comparing and discussing the results, before another round of sketching. We each drew some rough drafts for screens that would give Duolingo users a more immersive experience, while maintaining its playfulness. Two key ideas emerged from this:

  1. A game/simulation to practice a realistic conversation.
  2. A feature to find a ‘study buddy’, which would allow users to connect with native speakers of the language they’re learning, who are themselves learning the user’s native language.
Design Studio

As a group we mapped out a flow for each of these ideas. Our initial flows were too complex with too many steps so, with Midori taking a lead, we iterated on each of them.

User Flow for Flow 2 (white = first iteration)

This exercise brought into focus the key stages for each of these features. For the game to practice speaking users would need to:

  1. Choose ‘practice speaking’
  2. Choose a scene
  3. Either watch a tutorial or go straight to the exercise
  4. Do the exercise
  5. Stop practicing, practice in a new scene, or continue with a buddy
User Flow for Flow 1 (white = first iteration)

To find a study buddy users would need to:

  1. Either choose an existing study buddy or -
  2. Browse the suggested profiles and send a buddy request
  3. Either message the buddy or request to play a game with them

Paper prototypes into mid-fidelity:


My teammate Hugo created a paper prototype of the game, which we tested on five users.

Hugo’s sketches of the game/exercise

Our tests suggested that:

  • Users got lost as there were too many screens
  • There was a lack of clarity about how to progress from one screen to the next.

As he moved into lo-fi prototyping, Hugo reduced the number of screens and made the ‘continue’ buttons consistent so progression would be more intuitive for users.

Mid-fi wireframes for Flow 1: The Game


I developed some sketches of Flow 2, adding the new ‘Study Buddy’ feature to the existing ‘profile’ section of Duolingo.

Sketches of Flow 2: Find a Study Buddy

Testing my Marvel paper prototype of these sketches, I found that users:

  • Struggled to understand what the symbols on Duolingo’s current main navigation bar represented — they correctly selected the face to find a study buddy, but did not expect to come to the ‘profile’ section
  • They wanted more information about potential study buddies before they add them
  • They felt being able to call strangers immediately was intimidating, and would prefer to play a game or chat with them first

I iterated on the flow of finding a buddy with a more intuitive main navigation bar. I added a compatibility rating for potential study buddies, as well as their interests, location and age, and I removed the ‘call’ button from the profile page.

Home page (left) and suggested study buddies (right)
Profile of a suggested connection (left and centre) and a current connection (right)
Site Map for Duolingo’s current format

Alice and Midori took the lead on creating a site map for Duolingo’s current format, and then revised it according to the insights from our interviews and usability tests. This was used as the basis for our mid-fidelity prototype and brought into focus how we would offer users a more immersive experience. I suggested that, if we called the current homepage ‘learn’, by introducing a new ‘practice’ section where users could then choose between practicing listening and speaking, we could highlight to users the new features that would prepare them for a real situation.

Mid-fidelity into hi-fidelity

Conducting five usability tests on our mid-fi prototype, we found the following insights, and made the following changes:


  • Of the two home page options that Alice created, users found it easier to navigate the screen without the ‘streak’ feature, which they found a distraction, so we removed this


  • Users were unclear on which part of game they were supposed to be engaging with, so we reduced the opacity of the section they no longer needed to address
  • Users were unsure when they should go to the next page, so we made this happen automatically when they had completed a recording and added instructions for the game at onboarding
  • Users were confused by the explanations of ‘listening’ and ‘speaking’, so we removed them and used one large button for each
Lo-fidlity page of the game (left) to hi-fidelity page of the game (right)
Iterations of the listening and speaking page (left) and the first onboarding page (far right)


  • Since our users weren’t sure what was a connection they already had and a suggested connection, we decided to split these into two clear tabs
  • Users had also said they would like a bit more information about the tutors in order to help build up their trust in them, so we included their rating and fee
Iterations of the ‘Connect’ page

Here is our final prototype:

What next?

  • The next step is some usability testing and accessibility testing on our hi-fi prototype
  • We would also love the chance to develop the placement test, as many of Duolingo’s users said they currently felt frustrated by the level they were given by this test

Key Learnings

  • I learned a lot about paying attention to differences in insights from different types of user, and how to cater to these
  • A lot changed from our initial wireframes to our final prototypes. This really highlighted to me the benefit of starting in lo-fi and not getting too attached to initial ideas. I learned to really listen to the users during tests to try and find out where they were struggling
  • Working in a group was fantastic and a steep learning curve — by the end of the project our ability to incorporate Agile processes and work collaboratively had really improved