UX Research Case Study · 2026

Why Users Stop Learning on Duolingo

A survey-based investigation into the gap between Duolingo's engagement mechanics and users' actual learning effectiveness — and what could close it.

Researcher
Diya Shah
Method
Survey Research
Participants
39 Respondents
Focus
Learning Effectiveness
scroll to explore
01 — Overview

The Problem

Duolingo has 500M+ downloads. Yet most users stop. Are they stopping because life got busy — or because the app isn't actually teaching them anything?

This research set out to understand the relationship between learning effectiveness and user drop-off in Duolingo. Drawing on self-determination theory and behavioral psychology, the study explores whether Duolingo's gamification-first design creates genuine learning motivation or merely the illusion of progress.

The central hypothesis: users disengage not simply due to distraction, but because they stop perceiving real language acquisition — creating a motivational collapse that gamification mechanics alone cannot prevent.

01

Survey Design

Structured questionnaire covering usage history, motivation, emotional experience, and feature needs.

02

39 Participants

Mixed usage profiles — active, lapsed, and churned users across 15+ languages.

03

Psych Framework

Analysis grounded in Self-Determination Theory (SDT) and intrinsic vs. extrinsic motivation.

02 — Participant Snapshot

Who Responded

The overwhelming majority of participants were former users — making this a study of failure, not success. This is precisely the most informative group for understanding why Duolingo loses learners.

0

Used Duolingo regularly, then stopped entirely

0

Still use it, but much less frequently than before

0

Still active, regular users of the platform

0

Started for personal interest — their most intrinsic motivation

Most commonly learned languages
Spanish0
French0
German0
English0
Japanese0
03 — Key Findings

What the Data Reveals

Four distinct patterns emerged from the survey data, each pointing to a systemic gap between Duolingo's design priorities and users' psychological needs as learners.

Finding 01

Drop-off Happens Fast

Over half of users (53.9%) significantly reduced usage within the first month. 25.6% disengaged within the very first week. This suggests onboarding fails to build lasting learning motivation before novelty wears off.

Finding 02

Progress Invisibility is the Core Problem

35.9% said they didn't feel real progress — the second biggest drop-off driver after motivation loss. When learners can't perceive growth, SDT predicts motivation collapse. Streak-based feedback masks this gap.

Finding 03

Fun ≠ Effective

43.6% described Duolingo as "fun but not very effective." This is the defining tension: gamification succeeds at entertainment but fails to deliver the core promise — actual language learning. Users eventually notice the mismatch.

Finding 04

Emotional Arc: Motivated → Disengaged

Only 10.3% stayed motivated over time. 17.9% explicitly experienced the "initially motivated, then disengaged" arc — a textbook case of extrinsic reward decay. Once novelty and streaks stop feeling meaningful, there's no deeper pull to stay.

04 — User Needs

What Would Keep Users Coming Back

When asked what would most increase their likelihood of continuing, users responded with striking clarity — a demand for real-world utility over gamification polish.

Real-world conversation practice0
Clearer sense of progress0
Less repetition0
More personalized lessons0
Better motivation / accountability0
More flexibility in pacing0
participant voices — in their own words

"More lessons based on grammar, more scenarios to help with real world conversations."

Survey Respondent

"Duolingo was lowkey too slow for me — I wanted to learn Spanish to actually talk to people."

Survey Respondent

"I didn't appreciate that the app I used to talk to people and engage in real life conversations was being taught by an artificial intelligence."

Survey Respondent
05 — Psychological Analysis

The Motivation Collapse Model

Viewing the data through Self-Determination Theory (SDT), three core psychological needs must be met for sustained learning motivation: competence, autonomy, and relatedness. Duolingo's design structure systematically underserves all three over time.

Competence

Users report not feeling real progress (35.9%). Without genuine competence feedback, intrinsic motivation erodes. Streak counts measure consistency, not capability — they create the feeling of learning without its substance.

Autonomy

30.8% found lessons repetitive and boring, and 17.9% wanted more pacing flexibility. Duolingo's linear, algorithm-driven path limits learner agency, which SDT identifies as a direct predictor of drop-off.

Relatedness

The top user request — real-world conversation practice at 66.7% — directly maps to this need. Language learning is inherently social, yet Duolingo remains a largely solitary, transactional experience.

The result is a predictable behavioral arc: extrinsic engagement (streaks, XP, leaderboards) drives early retention, but without underlying intrinsic reward, motivation collapses — usually within the first month.

06 — Design Recommendations

What Duolingo Could Do Differently

Based on the survey findings and psychological framework, three high-impact design interventions emerge. Each targets a specific failure point in the current user experience.

01

Real-World Conversation Simulator

Scenario-based conversation modules that simulate real interactions: ordering food, asking directions, making small talk. This directly addresses the #1 user request (66.7%) and bridges the gap between gamified exercise and actual communicative competence — delivered via AI conversation partner or peer-matching.

02

Meaningful Progress Architecture

Replace or supplement the streak metric with a "language ability score" reflecting demonstrated skill — vocabulary range, sentence complexity, comprehension accuracy — rather than daily consistency. Users need to feel they are becoming more capable, not just more consistent. Directly targets the 35.9% who reported no sense of real progress.

03

Goal-Anchored Onboarding

The data shows 24.7% of users started with unclear or no learning goals. Design onboarding that surfaces a specific, motivating goal ("have a basic conversation by my trip to Paris in June") and uses it to personalize content, measure meaningful progress, and resurface motivation during low-engagement periods. Goal clarity is a proven predictor of sustained behavior change.

07 — Limitations & Next Steps

What This Research Doesn't Yet Answer

This study provides strong directional signals but has limitations worth acknowledging. The 39-person sample, while sufficient for exploratory research, limits statistical generalizability. Participants were likely recruited from a convenience sample, which may skew toward younger, more educated users.

The survey captures reported behavior and attitudes but not actual in-app behavioral data (session lengths, completion rates, feature usage). Future research phases should include:

  • Follow-up interviews with the 23.7% who indicated willingness (~9 people) for qualitative depth on the motivation collapse experience.
  • Usability testing of the proposed progress architecture with a prototype to validate whether it actually shifts perceived competence.
  • Longitudinal tracking of a new cohort to map the emotional arc in real time rather than through retrospective recall.
08 — Conclusion

The Takeaway

Duolingo has solved the problem of making language learning feel approachable and fun at the start. It has not solved the harder problem: making users feel like they're actually learning a language.

The data tells a consistent story across every question. Users arrive motivated and curious. They engage with the gamification. And then — usually within a month — they begin to notice that the streaks and XP don't translate into anything they can use. The fun stops being enough, and without real perceived progress, there's nothing left to hold them.

The fix isn't more gamification. It's designing for competence, not just engagement. That means real-world practice scenarios, honest and meaningful progress metrics, and an onboarding experience that anchors users to a goal worth caring about. These aren't small feature tweaks — they represent a fundamental reorientation of what success looks like for both the user and the product.

Case Study by Diya Shah · B.A. Business Administration & Psychology · UC Irvine · dvshah1@uci.edu
Back to Portfolio