The setup
This is placeholder content showing the shape a good teardown takes. Swap in your own analysis — the structure is the useful part. Start with what the product is trying to get the user to do, in one sentence. For Duolingo, it's simple: open the app again tomorrow, not just today.
What's actually working
Name the specific mechanic, not the vague concept. It's not "gamification" — it's the streak counter, the loss-averse framing ("don't lose your streak"), and the push notification timed right before the day resets. Explain why each piece does its job, using a concrete moment: the freeze-streak item, the mascot's slightly guilt-tripping copy, whatever detail you actually noticed.
A good critique names the exact screen, the exact word, the exact moment — not the general idea.
Where it breaks down
Every mechanic has a cost. Maybe it's the users who keep a streak alive by doing the laziest possible lesson, learning nothing. Maybe it's the anxiety a broken streak creates. Say what you'd measure to know if this was actually a problem, and what you'd try instead.
What I'd do differently
End with a point of view, not just observations. One or two concrete changes you'd propose, and the trade-off each one makes. This is the part hiring managers actually remember.