How Duolingo reignited user growth

Key takeaways

1. Before borrowing a feature from another product, ask why it works there, whether it will translate to your context, and what you must change. Mazal's Gardenscapes moves counter and Uber-style referral program both failed because the context differed.
2. Build a model that puts every user in exactly one engagement bucket, then simulate moving each retention rate by 2% per quarter. At Duolingo this showed current user retention (CURR) had 5 times the DAU impact of the next best metric.
3. Choose one North Star metric tied to your goal. Focusing on CURR meant Duolingo stopped working on things that had seemed paramount, such as new-user retention.
4. Protect your notification channel. Groupon's escalating email tests drove users to opt out for good, so Duolingo let teams tune timing, copy and templates but required CEO approval to send more notifications.
5. Look for wins in features you already have. Checking which features correlated with retention led Duolingo to optimize streaks, starting with a streak-saver notification, and the work paid off repeatedly.