Here are the key takeaways from Albert Cheng's talk on "Finding hidden growth opportunities in your product" (Duolingo, Grammarly, Chess.com):
- Explore–Exploit Framework: Growth teams must balance exploring new ideas (searching for new “mountains” to climb) with exploiting what’s already working. Too much exploration leads to scatter; too much exploitation can saturate and stagnate. Use insights from experiments across the product and expand successful tactics to adjacent areas for maximum impact.
- Rapid Experimentation: Teams like Albert's run upwards of 1,000 experiments a year—this is a cultural shift requiring founder/leadership support and fast, lightweight experimentation frameworks. Wins must be shared broadly and celebrated to propagate a culture of experimentation across the org.
- Human Psychology Insight: Sometimes, counterintuitive user behaviors happen, as with Chess.com’s game review feature: users are most likely to review after a win, not a loss. Surfacing positive feedback after failure dramatically boosted engagement and retention.
- Monetization Strategies: At Grammarly, offering samples of premium features to free users—rather than restricting purely to basic spellcheck—doubled rates of upgrade to paying plans. “Reverse trial” (feature-limited but ongoing) works better than short time-based trials for converting users.
- Retention Is Golden: For consumer subscription apps, user retention is key. Target daily retention rates of 30–40% for strong health. Mature products obtain most growth (~80%) from “resurrected users”—previously dormant accounts that return. Activation/onboarding is critical especially for products infrequently opened (like Grammarly).
- Gamification Pillars: Build habits and motivation by focusing on:
- The core loop (tight feedback and reward cycle for daily habit)
- The metagame (leaderboards, achievements for long-term motivation)
- The profile (visible investment and progression over time)
- Applying AI for Growth: Use AI tools (like data retrieval bots, prototyping tools with v0, Figma Make, Cursor, Claude Code) to answer ad hoc data questions, accelerate prototyping, and empower experimentation. The challenge lies in integrating these tools across functions to streamline actual shipping-to-production.
- Cultural Lessons: Building high-performing teams often means hiring for agency and fast learning—not always for deep experience. Especially as AI disrupts fields, old habits can be a liability; seek high-energy, adaptable personalities and foster beginner’s mindsets.
- Retention and Virality Tactics: Both through delighting users in unexpectedly positive ways and tapping into naturally viral product moments (like Duolingo’s streaks), growth comes from leaning into what organically works and amplifying it.
- Systems Matter: Robust experiment tracking, the right growth model, and correct instrumentation are foundational—misconfigurations can derail lessons. The company-wide system for running and learning from experiments is as important as the experiments themselves.