Finding hidden growth opportunities in your product | Albert Cheng (Duolingo, Grammarly, Chess.com)

Key Concepts

• Growth as Value Connection: The primary role of growth is to connect users to the value of a product.
• Explore and Exploit Framework: A methodology for finding and capitalizing on growth opportunities, balancing divergent idea generation with focused execution.
• User Retention: A critical metric for consumer subscription companies, often considered "gold."
• Freemium Business Model: A strategy where a core product is offered for free, with premium features available via subscription.
• AI in Growth: Leveraging artificial intelligence to accelerate experimentation, data analysis, and prototyping workflows.
• Gamification Pillars: A framework for habit formation and motivation, comprising the Core Loop, Metagame, and Profile.
• High Agency: A desirable trait in team members, characterized by speed of thought, action, and learning, often prioritized over deep experience.
• Company Stage Fit: The idea that individuals thrive best at specific company sizes or stages.
• Reputation: The long-term impact of small daily decisions and character on career and life opportunities.

Introduction to Albert Chen and His Growth Philosophy

Albert Chen is recognized as a leading consumer growth expert, having spearheaded growth and monetization at highly successful consumer products including Duolingo, Grammarly, and Chess.com. Earlier in his career at YouTube, he contributed to streaming and gaming features used by over 20 million people. Chen's unique approach to growth integrates marketing, data, strategy, and product management. He emphasizes that the core job of growth is to "connect users to the value of your product," countering the perception that growth is merely "metrics hacking." This holistic view focuses on understanding the user journey and how product value evolves over time for different user segments.

Chen's background as a serious piano player, who considered a music conservatory before pursuing engineering, offers an interesting parallel to his growth philosophy. He notes that both music and growth rely on consistent repetition, tight feedback loops, and resilience to mistakes, where learning occurs through continuous iteration. Both also require a blend of structural underpinning (e.g., growth models, metrics, experiments) and creativity (e.g., hypotheses, innovative solutions).

The Explore and Exploit Framework for Growth

Albert Chen extensively discusses the Explore and Exploit framework, which he learned from his engineering partner at Grammarly, Nurmal (and potentially originated from Brian Balfour).

• Explore Mode: Analogous to "finding the right mountain to climb," this phase involves divergent thinking, brainstorming, and trying a wide range of new ideas to uncover potential growth opportunities.
• Exploit Mode: Analogous to "focusing your resources on climbing that mountain effectively," this phase involves convergent thinking, doubling down on successful experiments, and expanding their impact across the organization.

Chen warns against spending too much time on one end of the spectrum: excessive exploration can lead to a scattershot approach without clear strategy, while over-exploitation can result in saturation, stagnation, and merely local maximization.

Real-World Application: Chess.com Game Review Chen provides a concrete example from Chess.com, where a PM named Dylan was tasked with improving engagement and retention for the "game review" feature (a virtual coach that analyzes chess games).

• Exploration: Initial observation revealed that 80% of users reviewed games after a win, which was counterintuitive to the initial assumption that users would review after losses to learn from mistakes. This insight highlighted a psychological preference for positive reinforcement.
• Intervention: For users who lost a game, the product experience was changed to surface "brilliant moves" and "best moves," accompanied by encouraging coach messages (e.g., "Losing is just part of learning, keep it up").
• Results: This change dramatically increased game reviews by 25%, subscriptions by 20%, and significantly boosted user retention.
• Exploitation: The insight was shared broadly across the company. Adjacent product managers (e.g., for puzzles) could then audit their own "cold patterns" and make them more positive by tweaking copy, button colors, or success ratings. This allowed the initial experiment's learning to be expanded tenfold across the organization.

Chen emphasizes that typical experiment win rates are around 30-50%, meaning many hypotheses fail. Both significant wins and losses are valuable for learning. To guide the oscillation between explore and exploit, Chen uses experiment explorer tools to identify patterns. A signal for over-exploitation is when more and more experiments fail to achieve statistical significance, indicating that the "juice has been squeezed" from that area, prompting a return to divergent brainstorming.

Leveraging AI in Growth and Product Development

Albert Chen highlights several ways AI is transforming growth work and product development:

• Text-to-SQL Capability: Chess.com developed a Slack bot that uses AI to answer ad-hoc data requests (e.g., "how many subscribers in South Africa?"). This automates first-pass answers, reducing the workload on data analysts and fostering a more data-informed culture by encouraging users to ask more questions without fear of bothering colleagues. This is an in-house solution not yet publicly released.
• AI Prototyping Tools: AI accelerates the journey from idea to representative solution. Product managers use tools like Vzero, designers use Figma Make, and engineers use Cursor, Cloud Code, and GitHub Copilot to quickly create prototypes for core product screens (e.g., onboarding, home screen, chessboard). This visualization makes bolder ideas more discussable and testable, significantly speeding up the exploration phase of growth.
• Workflow Integration Challenges: While AI tools are powerful for individual functions (marketing uses for translation, customer support uses Intercom Fin), Chen notes a current challenge in seamlessly bridging these "tinkering" efforts into a cohesive workflow, especially for larger companies that still involve handoffs between functions. He expects this interoperability to improve over time.
• Impact on Experimentation Cycle: AI tools, like ChatGPT, can summarize analysis documents and suggest new hypotheses, making the ideation and research phases of the experiment cycle much faster.

AI's Role in Chess.com's Product: Chess and AI have a long history, with AI chess engines like Stockfish now dramatically outperforming human grandmasters (e.g., Stockfish at ~3600 ELO vs. Magnus Carlsen at ~2800 ELO). Chess.com leverages this by:

• Game Review: Running powerful chess engines behind the scenes to evaluate every move, then translating complex analysis into plain, approachable language, often with audio (using LLMs for personality and speech).
• Customer-Centric Approach: Emphasizing that AI is applied to augment the human playing experience and provide value, not just for the sake of using the "new hot thing."
• LLMs vs. Chess Engines: Chen notes that LLMs themselves are currently poor at playing chess due to hallucination and a focus on pattern recognition over deep, precise calculation, highlighting the importance of applying the right AI technology for the specific problem.

Major Monetization Win at Grammarly

Albert Chen, working with Chief Product Officer Nome Leavinsky, identified a significant monetization opportunity at Grammarly, an AI-powered writing assistant with a freemium business model.

• Context: Over 90% of Grammarly users were on the free service. Free users primarily saw suggestions for spelling and grammar correctness. Paid features, such as tone improvement, clarity, and sentence rewrites, were typically only revealed after a user accepted all free suggestions and hit a paywall.
• Problem: The "lived product experience" for most free users was that Grammarly was merely a spelling and grammar checker, limiting their perception of its full power, especially during the generative AI transformation. Data showed that a very small percentage of free users accepted all suggestions; most picked and chose.
• Experiment: Grammarly decided to sample a number of different paid suggestions and intersperse them to free users across their writing. This provided a "limited taste" of the premium offering, acting as a real-time, capped "reverse free trial" (e.g., a few premium suggestions per day).
• Initial Concern: There was a rational concern that giving away premium features might reduce the incentive to subscribe.
• Result: This change led to a nearly doubling of upgrade rates. Free users began to perceive Grammarly as a much more powerful tool.
• Key Learning: For freemium products, the free offering should, to an extent, reflect the full potential of the product. Providing a taste of premium features can significantly drive conversion by showcasing greater value.

Success Factors for Consumer Subscription Products

Chen outlines critical elements for building successful consumer subscription businesses:

• User Retention is Gold: High retention is paramount. Low retention forces companies to aggressively monetize users on day one, which is significantly harder and shifts the business model.
• Word-of-Mouth Growth: Successful consumer subscription products often grow organically through word-of-mouth, especially those with a broad, mission-oriented value proposition (e.g., Duolingo for education, Chess.com for learning).
• Network Effects: Products with social features (Duolingo) or B2C2B plays (Grammarly, where free users influence team purchases) benefit from network effects.
• Freemium Sweet Spot: The core value proposition should be permanently free, with premium features layered on as a "sampling" or "taste."
• Retention Benchmarks:
  • New User Retention (D1): A 30-40% Day 1 retention rate is considered solid for a consumer app.
  • Existing User Retention: For products with daily frequency, the retention of existing, habitual users is the most crucial lever for long-term growth and compounding.
  • Grammarly Exception: For products not used proactively daily, activation, installation, and the "aha moment" are more critical than daily existing user retention.
• Cost-Efficiency: Many successful consumer subscription products start "scrappy" and cost-efficient, as it often takes time to find product-market fit and sustainable growth.
• Resurrected Users: For mature companies with a large base of inactive or sporadic users (potentially hundreds of millions), investing in an excellent "resurrected user" experience is crucial. Examples include Duolingo's social notifications to bring back users and placement tests for returning language learners.

Operating Models of Successful Consumer Subscription Companies

Chen contrasts the operational styles of Duolingo, Grammarly, and Chess.com:

• Duolingo: Characterized by a highly particular and structured product development approach, often referred to as the "Green Machine" playbook. They hire intelligent, energetic college graduates, provide advanced experimentation tooling, and prioritize "clock speed." The product experience can change multiple times a day for each user. Their core ethos is around motivation and habit building, with language learning serving as the primary vehicle.
• Grammarly: Evolved significantly, starting as a paid product for students, expanding to a freemium model for everyone, then focusing on professionals, and eventually developing a B2C2B/enterprise sales motion. Chen's role involved "product-led sales," identifying teams and companies for demand generation. The company is rapidly transforming with generative AI, aiming to become a broader productivity suite. A key insight was that the core product experience itself (frequency and quality of suggestions) was the primary driver of repeated activity, rather than solely growth team levers.
• Chess.com: Defined by its "super fanatical" culture around chess. The company is globally remote and hires individuals who deeply love and play chess, leading to constant "dogfooding" of the product and a vibrant internal energy for generating ideas. Their growth combines consistent growth experimentation with leveraging "big waves" like the pandemic, Queen's Gambit series, and the rise of streamers.

Chen notes that there's no single "right" way to succeed, as all three companies have achieved immense success with distinct approaches.

The Power of Brand and Community in Growth

Initially skeptical of marketing as a product person, Chen now views brand and community as "rocket fuel" for growth, working in tandem with experimentation.

• Duolingo's Duo the Owl: The character's personality, developed through push notifications and product experience, was powerfully leveraged by the marketing team on TikTok, YouTube, and social media. This fed into internet memes, and internal tracking showed that these brand channels contributed 20-30% of new users on some days.
• Chess.com's Organic Surge: For its first 15 years, Chess.com was largely under the radar. However, in the last five years, a combination of the pandemic, the Queen's Gambit series, and the rise of YouTube/Twitch streamers led to a massive surge in popularity. This demonstrates how growth experimentation (iterating on the product) combined with significant cultural waves (brand/community) can lead to exponential growth.

Experimentation Best Practices and Culture Shift

Chen offers several key practices for effective experimentation:

• Just Start Somewhere: He notes that 40% of product teams don't run experiments at all. For consumer products with scale and frequency, experimentation is crucial because consumer behavior is fickle, and even experienced power users can misjudge new user experiences.
• Tooling: While Duolingo and Chess.com have in-house experimentation tools (Duolingo's being a "huge accelerant"), Chen advises against building in-house from day one, recommending third-party tools like StatSig (which Grammarly used).
• Ambitious Goals: At Chess.com, the goal is to scale from ~50 experiments in 2023 to 1,000 experiments per year in the future. This ambitious target serves as a "north star" to drive conversations about the necessary systemic changes (e.g., involving lifecycle marketing, App Store optimization, content marketing, enabling no-code experimentation for specific screens, and improving observability). The true value is in the learning and accelerated shipping, not just hitting the number.
• Culture Shift: Shifting a company culture towards experimentation requires:
  • Critical CEO/Co-founder Support: Leaders like Eric and Danny at Chess.com must demonstrate mental flexibility and actively preach product-led growth and experimentation.
  • Celebrating Wins: Showcasing successful experiments (like the Chess.com game review example) motivates teams, validates the approach, and encourages broader application of learnings.
• System Over Individual Experiments: The overall experimentation system is more important than any single experiment. This includes:
  • Growth Model: A clear understanding of how the company grows and which channels to leverage.
  • Robust Instrumentation: Accurate product instrumentation is vital to avoid "wonky results" (e.g., a past horror story where user retention was configured backwards, making positive results appear negative).

Habit Formation and Motivation

Drawing from his experience at Duolingo and Chess.com, Chen discusses principles of habit formation and motivation, referencing Jorge Mazal's gamification pillars:

• Core Loop: The fundamental, repeatable action (e.g., a Duolingo lesson) that provides immediate rewards and reinforces a streak, often supported by push notifications. This loop must be tight and engaging.
• Metagame: Long-term motivators that provide a sense of progression and achievement (e.g., Duolingo's path, leaderboards, achievements).
• Profile: A reflection of a user's investment and progress over time, building a sense of ownership and accomplishment within the product.

Addressing Beginner Challenges at Chess.com: Chen highlights that over 75% of new Chess.com users identify as new or beginner. Data shows that less than a third win their first game, and losing a game results in 10% worse user retention. To counter the self-doubt and negative reinforcement beginners often face, Chess.com is experimenting with:

• Crafting a more delightful "learn how to play" experience instead of immediately dropping users into live games.
• Hiding user ratings for the first five games to prevent discouragement from plummeting scores.
• Offering alternative play options like coaches, friends, or bots.

Counterintuitive Lesson: Building Teams

Chen shares a counterintuitive lesson about team building:

• Beyond Deep Experience: While traditional hiring focuses on job descriptions, similar company experience, and deep domain knowledge, Chen has observed that the highest performers often possess high agency, clock speed, and energy, coupled with a strong commitment to the mission, even without extensive prior experience in the specific domain.
• Experience as a Crutch: In rapidly shifting environments, particularly with the advent of AI, deep experience can sometimes be a "crutch." Learned habits may need to be intentionally discarded, requiring a "beginner's mind."
• Focus on Speed of Learning: Chen prioritizes hiring individuals who respond, move, and learn quickly, believing that companies with the fastest speed of learning are most likely to survive and thrive.
• Identifying High Agency: These traits are often revealed through "soft signals" outside the formal interview process, such as the types of questions candidates ask, their engagement with the product, references, communication style, and energy during conversations.

Company Size and Personal Fit

Chen believes that everyone has a company stage where they "shine best." He describes his personal "goldilocks zone" as medium-sized companies (roughly 500-1,000 people, typically 10-20 years old, durable, and profitable).

• Big Tech (e.g., Google): Offers immense scale and robust tools but can be slower to move and ship products, which Chen found frustrating.
• Tiny Startups: Move incredibly fast but can be grueling due to the challenges of building awareness and acquiring users one by one.
• Medium-Sized Companies: Provide a balance, allowing Chen to contribute at scale while still getting into the details (reading experiment results, examining pixels) and executing at a daily/weekly pace rather than monthly or quarterly. These companies are often at key inflection points, offering dynamic environments.

Failed Corner: Chariot (Commuter Shuttles)

Chen recounts a significant failure from earlier in his career at a startup called Chariot, a commuter shuttle service in San Francisco.

• The Idea: Chariot, which offered reliable, fast, and affordable fixed-route shuttles, explored offering "dynamic routes" (Chariot Direct) similar to Uber/Lyft, aiming to improve vehicle utilization.
• The Failure: The initiative ultimately did not work out.
• Lessons Learned:
  1. Solution Searching for a Problem: The project was driven by a "wouldn't it be nice if" mentality rather than a clear, user-centric problem.
  2. Neglecting Multi-Sided Marketplace: Too much focus was placed on the rider app, neglecting the impact on drivers (who became confused and disgruntled) and the operations team, leading to a challenging overall product experience.
  3. Premature PR: Extensive public relations efforts were conducted before validating customer demand. This created sunk costs and pressure to succeed, making it harder to pivot or abandon the idea.

This experience profoundly influenced Chen, reinforcing the importance of rigorous experimentation and validation before significant public announcements or resource commitments.

Conclusion and Main Takeaways

Albert Chen's discussion underscores the dynamic and multifaceted nature of growth. His journey highlights the importance of:

• Authenticity and Continuous Learning: Drawing lessons from personal experiences and the successes/failures of others, acting as a "mental sponge" to absorb, test, and adapt new ideas.
• Strategic Frameworks: Utilizing structured approaches like the Explore and Exploit framework to systematically identify and capitalize on growth opportunities.
• Data-Driven Decision Making: Emphasizing experimentation at scale, robust instrumentation, and leveraging AI to accelerate insights and workflows.
• Holistic User Understanding: Connecting users to product value by understanding their evolving needs, motivations, and the psychological underpinnings of their behavior.
• Cultural Alignment: The critical role of leadership support and celebrating wins in fostering an experimentation-driven culture.
• Beyond Metrics: While metrics are important, growth ultimately serves the user and the product's mission, with brand and community acting as powerful accelerators.

Lightning Round Highlights

• Recommended Books:
  • Ogilvy on Advertising: A 40-year-old book packed with practical examples of copy and creative that compel user action, emphasizing effectiveness over cleverness.
  • Dark Squares (memoir by Danny Rensch, Chess.com co-founder): An "unbelievable story" about growing up in an abusive cult and being a chess prodigy.
• Favorite Product: The Breville Barista coffee machine, which provides a daily ritual and caffeine fix, especially after moving away from coffee shops.
• Life Motto: His mother's quote, "Nothing is more important than your reputation," which emphasizes how small daily decisions compound to open doors and how fragile reputations can be.
• Chess.com Skill: Albert's ELO rating is about 1800 for rapid games and 1500 for blitz, playing multiple times daily and taking bi-weekly lessons. His username is "goniners."