Counterintuitively, Duolingo discovered that competitive leaderboards are more engaging when users are pitted against strangers at a similar commitment level. Competing with friends often fails because their dedication rarely matches, making the competition feel unbalanced and demotivating.
Related Insights
The proliferation of AI leaderboards incentivizes companies to optimize models for specific benchmarks. This creates a risk of "acing the SATs" where models excel on tests but don't necessarily make progress on solving real-world problems. This focus on gaming metrics could diverge from creating genuine user value.
Duolingo's most powerful re-engagement notification is one sent after five days of inactivity stating, "these reminders don't seem to be working. We're going to stop sending them." This passive-aggressive message makes users feel the app is "giving up on them," which is surprisingly effective at getting them to return.
To overcome internal resistance to making money from its mission-driven, communist-leaning early team, Duolingo framed its freemium model as wealth redistribution. Wealthier users who pay for premium features effectively subsidize free education for users in poorer countries, aligning financial needs with the company's core social mission.
To amplify word-of-mouth, Duolingo identified existing sharing behavior by temporarily tracking user screenshots. They found hotspots like streak milestones and funny challenges, then invested in designers to make these moments even more shareable.
Before 'crowdsourcing' was a term, Luis von Ahn built games to solve problems computers couldn't. His ESP Game tricked millions of players into labeling images for free, providing crucial training data for early image recognition AI by turning a tedious task into a fun, competitive experience.
When introducing a new skill like user interviews, initially focus on quantity over quality. Creating a competition for the "most interviews" helps people put in the reps needed to build muscle memory. This vanity metric should be temporary and replaced with quality-focused measures once the habit is formed.
Instead of a passive, open-ended affiliate program, create concentrated launch windows (e.g., one week) with a public leaderboard and prizes. This injects competition and urgency, motivating affiliates to push far harder than they would in a standard, always-on program.
When Duolingo paused its "unhinged" owl mascot social media strategy, daily active user growth saw its smallest increase in years. This direct correlation demonstrates that for some consumer apps, the social media team can be as crucial for growth as the engineering team, justifying top-tier compensation.
While rewards can remind people of expectations, they are poor at building skills. Research shows a strong negative correlation between using external rewards (e.g., money) and developing intrinsic motivation. The more you motivate externally, the more you may weaken internal drive.
Labs are incentivized to climb leaderboards like LM Arena, which reward flashy, engaging, but often inaccurate responses. This focus on "dopamine instead of truth" creates models optimized for tabloids, not for advancing humanity by solving hard problems.
