Today I listened to a fascinating podcast episode from No Priors featuring Duolingo CEO Luis von Ahn, who shared some cool and effective tactics Duolingo has employed to boost user retention:

Surprising Retention Tactics

Short Lessons

He highlighted that a "big game changer" was making lessons only 2 minutes long instead of 30 minutes. Lowering the initial commitment improved user engagement since it sets an easy-to-achieve expectation. Once users start, they often end up spending more time anyway. The key is just to get over the initial hurdle.

Streaks

App usage streaks are super powerful. He mentioned that Duolingo now has "10 million active users that have a streak longer than 365 days". It's pretty techincally too to add to a product.

Passive-Aggressive Notifications

After 5 days of inactivity, Duolingo sends a notification: "These reminders don't seem to be working. We're going to stop sending them for now." Apparently this is really good at bringing users back, since it makes users feel like you've given up on them.

Adaptive Difficulty (83% Rule)

Rather than focusing solely on exercises users struggle with, Duolingo found that providing content where the user has about an 83% chance of success maximizes enjoyment. This keeps users motivated enough to reach the 500–2,000 hours typically required for language proficiency, depending on the language.

The Future of Education and AI

Given he's already at the frontier of education + tech, Luis also shared some interesting predictions about AI's potential in education:

  • AI as the Primary Teaching Method: AI will likely become the primary means of individualized instruction, significantly outperforming the scalability of human teachers alone.

  • Role of Teachers: Human teachers will remain, primarily providing care and supervision rather than direct instruction.

  • Democratized Quality Education: AI can provide access to high-quality teaching globally, helping overcome the scarcity of excellent teachers.

  • Slow Adoption: Despite AI's promise, he expects slow adoption given institutional inertia and regulatory hurdles; expects some developing regions might "leapfrog" ahead, rapidly adopting scalable AI-driven education due to necessity.

  • Dynamics in Private Schools: High-end private schools might face a dilemma in justifying high tuition fees if their primary instructional tool becomes AI-based applications.