Back to Shadow PM
Shadow PM Analysis

Duolingo

What would I do if I were the PM for Duolingo?

Product: Duolingo

FOCUS METRIC

7-Day Streak Retention

WHY THIS MATTERS

Duolingo's business model depends on daily active users. The 7-day streak is critical—users who maintain a 7-day streak have 3x higher long-term retention.

Current State Analysis

What's Happening Today

60% of new users drop off before completing day 7

Day 3 sees the biggest drop-off (25% of remaining users)

Users report feeling overwhelmed by too many features in onboarding

Streak pressure creates anxiety rather than motivation for beginners

HYPOTHESIS

New users fail to build the habit because the first week feels too hard, not because they lose interest in learning.

Research to Validate Hypothesis

Interview 40 users who dropped off before day 7

Analyze session data: when do users struggle and quit within a lesson?

A/B test easier first-week difficulty vs current baseline

Survey power users: what kept them going in week 1?

90-Day Execution Plan

Research & Validation (Weeks 1-4)

Phase 1

Conduct user interviews with early drop-offs

Analyze behavioral data: exactly where in week 1 do users quit?

Test hypothesis: Is week 1 too hard, or are expectations misaligned?

Define success metric: Increase day-7 streak completion from X% to Y%

Experiment & Test (Weeks 5-8)

Phase 2

Build 'Easy Mode' first week: shorter lessons, more encouragement

Test 'Micro-wins' notifications: celebrate every small achievement

Prototype 'Flexible Streaks': Allow one 'freeze' in first week without losing progress

Run controlled experiment with 5,000 new users

Launch & Iterate (Weeks 9-12)

Phase 3

Roll out winning variant to 20% of new users

Monitor day-7 completion rate vs control group

Gather feedback: Does 'Easy Mode' week 1 hurt long-term rigor?

Plan week 2-4 improvements based on learnings

What I Would NOT Build

Just as important as what to build is what NOT to build. These options might seem appealing but would distract from the core goal:

❌ AI tutor feature – Adds complexity, doesn't solve week 1 drop-off

❌ Social learning groups – Too much friction for new users

❌ Gamification overhaul – Risks breaking what already works

❌ New languages in week 1 – Confuses users, doesn't address retention

Expected Impact

If week 1 feels achievable without compromising quality, predict 20-25% improvement in 7-day streak completion rate.