What is Duolingo?

Duolingo is a language learning app that has over 40 languages and over 74 million monthly active users. It is the biggest player in the industry, and is fondly known amongst learners for its quirky yet effective design.

Project Overview

Figured out major pain points behind Duolingo’s user abandonment, which has been huge despite its near-flawless design

Ideated, prototyped and tested novel ideas to address the pain points

The Team

Just me! :)

Timeline

8 weeks

How It Started

Final Designs

How It Started

Final Designs

How It Started

Final Designs

Duolingo’s design is near perfect. So why try to fix what isn’t broken?

Well, I didn’t think anything was broken until I noticed some patterns that led to some very intriguing revelations. Almost every time a friend of mine joined Duolingo, they would drop off within weeks, irrespective of their initial enthusiasm.

Research

Research

Research

So... What gives?

I spoke to some friends (and their friends) about their preferences, why they dropped off, or why they’re sticking with Duolingo.

I saw some patterns emerging amongst all learners, which had me intrigued, so I took up the challenge and dove right into some research.

Talking to Learners

Online Survey

Sample Size: 110

Discussions

I also had discussions with a couple of language learners who go to in-person language classes to understand why they chose offline classes and how the learning methods are different.

Here are 3 methods used in offline classes that stood out:

Learning in Pairs

Language teachers often group students in pairs for peer-to-peer learning as well as to create accountability.

Only Using the Target Language

Extensively used in fast-paced crash courses, this method advocates the use of only the language to be learnt for all written and oral communications in the class premises, regardless of the level the student is at.

Interaction With Native Speakers

Offline classes also bring in native speakers of the target language, especially for students at higher levels, so they can improve listening and speaking skills with respect to local dialects.

What I Learned

Synthesizing The Research

Synthesizing The Research

Synthesizing The Research

Emerging Archetypes

Based on the research, learners can be categorized into 3 archetypes based on how and why they use Duolingo:

Learning For Fun

Learning as a hobby to be able to understand songs/shows in target language. These learners love the gamification aspect of Duolingo.

Career Opportunities

Learning to improve their chances of getting opportunities abroad. These learners often maintain written notes and long streaks.

Casual Learning

Using Duolingo as a supplement to language classes or when bored. These learners often use Duolingo only few times a week.

Spotting Points Of Tension

To document the struggles of every type of learner and pinpoint points of abandonment, I made journey maps for the three archetypes.

Here’s a summary:

Problem Areas

aka things that get Duo worked up

Understanding the Competition

Analyzing Duolingo

Strengths

Gamification

Fun branding and UI

Unique content

Freemium model

Big learner base and community

Large number of languages

Weaknesses

Monotonous

XP mining

Competition can become overwhelming

Lack of interaction between users

Discoverability issues with some features

Opportunities

Interactive and gamified engagement

Engaging with native speakers

Pushing features like stories and events

Realistic progress measurement

Peer-to-peer connection

Threats

Platforms providing classes with native speakers

Courses based on real-life scenarios like ordering in restaurants or understanding local dialects

Platforms with peer engagement

Offline classes

Platforms using entertainment for learning

Defining Next Steps

Defining Next Steps

Defining Next Steps

What’s Duolingo worried about?

Duolingo has previously attempted to capitalize on some seemingly straightforward opportunity areas, like letting learners connect on a forum and also chat with each other. However, this did not succeed as it was difficult to monitor interactions on such large scale, and upholding the tenets of safety and inclusivity was not possible.

Hence, all further solutions must be proposed such that there is no need to monitor learners’ interactions constantly.

So, what’s the goal?

To design/redesign features in Duolingo to increase engagement and interaction in learners, while keeping the brand’s values and identity intact.

Inching Closer To The Goal

Mind-mapping, brainstorming, ideating, ranting and the sorts

Priority Matrix

There was still a long list of ideas after filtering for feasibility, so I used a priority matrix to analyze the impact vs effort of each idea. This helped me choose the best ideas to elaborate on and sketch out.

Here’s what I came up with!

Feature 1

Feature 1

Feature 1

Group Stories

The quirky stories we know and love, but with a twist of friends and a dash of interactivity!

Interaction Between Learners

Insight leveraged in this feature

This feature aims at helping learners interact better with their friends and have a more engaging experience with the app.

The fun of stories, with a dash of excitement

All learners take turns to choose dialogues for the characters in the story, which can go down multiple paths depending on the choices everyone makes on their turn.

Reactions To Dialogues

Learners can react to dialogues chosen by others with different emojis.

Interact with friends, old and new!

Up to 5 other learners can join your story. You can choose to invite them from your friends list, or have Duo match you with learners from around the world!

Feature 2

Feature 2

Feature 2

Strategic XP Boosts

A new way to make sure learners are making efficient progress through their learning journey.

Spaced Repetition

Insight leveraged in this feature

Spaced repetition is an evidence-based learning technique that encourages learning a topic in stages over a period of time, instead of grinding through it for a long duration.

STEP 1

Strategic Boosts

Duo notices the learner has been doing the same topic for a while, and gives a 2x XP boost for the next few topics

STEP 2

Push To Learn Something New

The XP boost serves as an immediate incentive to move to newer topics

STEP 3

Breaking The Grind

The XP boost is exhausted after doing 1 lesson of each highlighted topic, or at the end of the day.

This ensures that the learner doesn’t grind on the boosted topics again, hence maintaining a balance between the number of topics learned vs the depth of learning.

STEP 4

Back To Normal

The learner can return to their desired topic at the end of it, or during the boost too!

Feature 3

Feature 3

Feature 3

Joint Quests

A feature that leverages collaboration, healthy competition and accountability to help learners progress through their course!

Collaboration > Competition

Insight leveraged in this feature

This feature rests on this one simple insight that came up repeatedly throughout the research stage, regardless of the language and the user’s demographics.

Collaborate With Friends

Work together with a friend instead of competing against a friend to earn a reward.

Accountability

The Quest is complete only when both of you complete your part!

Make Real Progress

These challenges are quantified by crowns, which are earned only when new lessons are completed; whereas quantifying by XP would lead learners back into the rut of repeating the same lessons over and over just to earn more XP and finish the challenge.

Choose Your Level

You can send and receive invites to quests for varying levels of difficulty to commit to.

Track Your Quests

Find all quest invites and active quests on the challenges page

Testing And Validation

Testing And Validation

Testing And Validation

Putting My Ideas To Test

I showed prototypes of these three ideas to people who regularly use Duolingo, as well as people new to the app.

Here’s some feedback I received.

Originally, this is where the project ended. But Duolingo had different plans for me! Read on to see how a few updates from Duolingo validated my project more than I could’ve ever hoped for!

Update: July 2022

Update: July 2022

Update: July 2022

Some Updates From Duolingo Themselves!

Shortly after I did this project, Duolingo rolled out their very own quests feature, which was shockingly similar to the one I’d made.

But on looking closer, I realized that while the execution was very similar, my feature differed from Duolingo’s in some fundamental ways, outlined below.

Difference In Measures Of Progress

My Version

I’ve used crowns to quantify the quests, which can only be earned by doing new lessons, hence making real progress through the course.

Duolingo’s Version

Duolingo uses XP and number of lessons to quantify their quests, which can be earned by just redoing easy challenges and old lessons, hence making no real progress through the course.

Lack Of Accountability

My Version

To ensure that both learners participating make progress, the quest is considered complete only when both learners earn their share of crowns.

Duolingo’s Version

Even if just one learner earns the full XP required for completing the quest, both participants end up reaping the benefits, creating a lack of accountability.

Choose Your Challenge

My Version

I allowed learners to choose both - the person they want to do a quest with, as well as the level of the challenge they want to commit to.

Duolingo’s Version

Duolingo pairs up random learners and gives them a fixed quest challenge, which can sometimes be either too easy or too difficult.

The Way I See It...

While the two versions appear similar, their purposes and intentions differ. Any resemblance can be attributed to ample research data indicating learners sought collaboration, not competition. While I was initially taken aback, I later realized I had conducted extensive research and tapped into a similar thought process as the Duolingo team, which was super validating!

Update: June 2024

Update: June 2024

Update: June 2024

Duolingo’s New Proprietary Metric: TSLW

Duolingo measures their numbers using metrics developed in-house specifically for their products, and it looks like the latest metric they’ve developed touches on all the points that I’ve addressed in this project!

Time Spent Learning Well (TSLW) is the metric the Duolingo team has developed to accurately measure learners’ progress, recognizing that time spent doing new lessons on the path is better spent than time spent on other lessons like reviews, stories, challenges, etc.

Read more about this on the Duolingo blog: blog.duolingo.com/time-spent-learning-well/

This is the exact basis on which this project rests, especially the Joint Quests and Strategic XP Boosts features. This provides further validation for how strong the foundation of this project is!

That’s all for now, folks!

There's beauty all around us…

And the Blooms app helps you spot it by showing you all the beautiful flowers in your city!