What to expect, what they’re really testing, and what a strong answer looks like — scored.
Gamification, habit formation, learning science, and engagement-revenue trade-offs. Duolingo PMs must understand how to balance learning outcomes (do users actually learn language?) with engagement metrics (daily streaks, notifications), which can sometimes conflict.
The question below was asked by Duolingo interviewers. The answer is graded on the five dimensions real PM interviewers use: structure, specificity, reasoning, decision quality, and delivery.
“How would you improve Duolingo's streak feature to better support language learning?”
The streak feature is Duolingo's most powerful engagement mechanism, but it has a known problem: it optimizes for daily check-ins, not for learning quality. Users maintain streaks by doing the minimum daily XP — often easy exercises they've already mastered — rather than advancing to harder content. The streak becomes a habit, but not a learning habit.
I'd reframe the goal: can we make the streak track learning progress rather than activity completion?
Proposed improvement: introduce a 'Learning Streak' alongside the existing activity streak. The Learning Streak increments only when a user engages with content at or above their current proficiency level — assessed via Duolingo's own placement and performance data. A user who does five 'easy' exercises to hit daily XP gets their activity streak, but their Learning Streak only increments if they completed at least one challenging exercise (≥70% difficulty, based on their accuracy history).
This creates a second axis of motivation for users who care about actually learning, without breaking the existing streak behavior for users who value the daily habit.
Success metric: Learning Streak completion rate (the percentage of active users who earn both streaks on a given day) and, more importantly, proficiency advancement rate — do users who earn Learning Streaks show faster vocabulary acquisition at the 30-day mark? This is the outcome metric that validates whether the feature is delivering learning value.
Guardrail: don't reduce overall daily active users — if Learning Streak introduces discouragement for casual users, it's a net negative.
Diagnoses the streak problem clearly, proposes a parallel mechanic that doesn't break existing behavior, defines outcome metrics.
Names the ≥70% difficulty threshold, 30-day proficiency advancement check, and the two-streak concept concretely.
The 'streak tracks activity, not learning' insight is correct and non-obvious; the two-streak design follows directly from it.
Commits to the two-streak design and frames it as additive (doesn't break existing behavior) — a smart constraint.
Well-paced; the 'second axis of motivation' framing is efficient.
This is a near-top-tier answer. The core insight — that streaks track activity not learning — is correct, and the two-streak design is elegant because it doesn't require breaking the existing mechanic. Using proficiency advancement as the outcome metric (rather than engagement) shows genuine learning-product maturity. The one gap: the answer doesn't consider how to explain the two-streak concept to users without confusing them — clarity of UI/UX is a real product concern for a dual-metric system.
Add one sentence on how you'd explain the two-streak distinction to users in the UI — for example, a tooltip that says 'Earn your Learning Streak by completing a challenge, not just a check-in.'
Paste any Duolingo PM interview question and your answer. Get scored on the same five dimensions — instantly, free, no signup.
Grade my answer free →First grade is free. No account needed.