Lenny's Podcast: Product | Career | Growth

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

October 5, 2025

Key Takeaways Copied to clipboard!

  • Growth is fundamentally about connecting users to the value of the product, countering the perception that it is purely metrics hacking. 
  • The Explore-Exploit framework should be applied at the micro, insight level, where successful experiments should be broadly shared to maximize impact across the organization before returning to exploration. 
  • For freemium consumer subscription products, the free offering should reflect the best of the product's capabilities (sampling premium features) to showcase value, as this can dramatically increase upgrade rates, as seen with Grammarly's 'reverse trial' approach. 
  • The goal of setting an ambitious experimentation target, like Chess.com's goal of 1,000 experiments a year, is less about hitting the number and more about sparking conversations that reveal necessary systemic changes across marketing, engineering, and product. 
  • Successful gamification, as seen at Duolingo, relies on nailing three pillars: a tight core loop (daily habit), a motivating metagame (long-term goals), and a valuable user profile (reflection of investment). 
  • When hiring, prioritizing high agency—characterized by fast clock speed, energy, and quick learning—can be more critical than deep, specific domain experience, especially in rapidly changing environments like the current AI landscape. 

Segments

Growth Mentality and Piano
Copied to clipboard!
(00:00:00)
  • Key Takeaway: Music practice and growth work both rely on consistent repetition, tight feedback loops, and resilience to constant mistakes.
  • Summary: Growth is defined as connecting users to product value, contrasting with pure metrics hacking. Albert Cheng draws a parallel between serious piano practice and growth work, noting both require consistent repetition and resilience to frequent mistakes. Both disciplines also benefit from a structural underpinning (growth model/music theory) combined with daily creativity and flow.
Explore vs. Exploit Framework
Copied to clipboard!
(00:09:40)
  • Key Takeaway: Exploration finds the right growth mountain, while exploitation focuses resources on climbing it effectively, requiring oscillation between the two modes.
  • Summary: The Explore vs. Exploit framework involves finding the right opportunity (explore) and then maximizing gains from that opportunity (exploit). A key insight from Chess.com’s Game Review feature showed that users review games after wins 80% of the time, contrary to the initial hypothesis that they review after losses. Flipping the post-loss experience to highlight brilliant moves and offer encouragement dramatically increased engagement and subscriptions.
AI in Data Analysis
Copied to clipboard!
(00:16:38)
  • Key Takeaway: Implementing text-to-SQL Slack bots democratizes data access by allowing non-analysts to ask ad-hoc questions, leading to an explosion of data-informed inquiries.
  • Summary: Chess.com is training Slack bots to handle first-pass answers for ad-hoc data requests, making the company more data-informed. This lowers the barrier for asking questions that users might otherwise feel embarrassed or hesitant to pose to a human analyst. This capability significantly speeds up the data-driven decision-making cycle.
AI Prototyping Stack
Copied to clipboard!
(00:18:31)
  • Key Takeaway: AI prototyping tools like VZero and Figma Make accelerate the transition from idea to testable artifact, shortening the time required for design and review handoffs.
  • Summary: Product managers are using AI prototyping tools to quickly generate representative solutions for core product screens like onboarding and the home screen. These AI-generated prototypes serve as a starting point for the rest of the company to iterate upon, making ideas more discussable and testable sooner. The current challenge is seamlessly bridging these rapid prototyping outputs into the production workflow across different functional teams.
Grammarly Monetization Win
Copied to clipboard!
(00:20:57)
  • Key Takeaway: Sampling paid suggestions interspersed with free suggestions for free users nearly doubled Grammarly’s upgrade rates by showcasing the product’s full power in real-time.
  • Summary: The core challenge at Grammarly was that free users perceived the product only for spelling and grammar, missing the value of paid features like tone and clarity suggestions. The team implemented a ‘reverse trial’ by interspersing a limited number of paid suggestions daily for free users. This strategy proved that giving a taste of premium features did not cannibalize subscriptions but instead made users see Grammarly as a much more powerful tool.
Freemium Strategy and Trials
Copied to clipboard!
(00:26:00)
  • Key Takeaway: Freemium models suit mission-oriented products that rely heavily on word-of-mouth growth, where the core value proposition should remain permanently free.
  • Summary: Freemium subscription models align well with missions focused on widespread adoption (like education or writing assistance) and products that benefit from network effects. For B2B features with high user investment (like CRM lock-in), reverse trials are powerful because users feel the pain of losing access. For many consumer products, however, standard time-based free trials remain the norm.
Consumer Subscription Retention
Copied to clipboard!
(00:28:07)
  • Key Takeaway: Solid D1 retention for a consumer app is around 30-40%, but for daily-frequency products, the retention rate of existing habitual users is the most critical compounding lever.
  • Summary: The biggest missing piece in building successful consumer subscription products is often user retention; without it, the pressure to convert users on day one becomes unsustainable. For apps with daily frequency, the retention rate of the established user base compounds significantly over time. Grammarly is an exception where the initial installation and ‘aha moment’ are critical because usage is event-driven rather than proactively opened daily.
Mature Product Growth Levers
Copied to clipboard!
(00:32:36)
  • Key Takeaway: For mature, high-frequency products like Chess.com, resurrected or existing users can account for 80% of active usage, making the user experience for dormant users a major growth lever.
  • Summary: In mature products, the active user base is heavily weighted toward existing or resurrected users, not just new sign-ups. Strategies must focus on excellent resurrection experiences, such as using social notifications to prompt return or re-evaluating proficiency levels upon return to place users correctly. This focus on existing users yields a much higher ROI than solely focusing on new user acquisition.
Operating Differences: Duolingo vs. Grammarly vs. Chess.com
Copied to clipboard!
(00:34:43)
  • Key Takeaway: Successful companies thrive with distinct operating philosophies: Duolingo uses rigid structure for high-velocity experimentation, Grammarly blends consumer and product-led sales, and Chess.com is fanatically focused on its core domain.
  • Summary: Duolingo operates with a highly structured, methodical playbook (‘The Green Machine’) that enables extremely high clock speed, changing the user experience multiple times daily through consistent experimentation. Grammarly’s growth job involved product-led sales, uncovering B2B patterns from its massive self-serve user base, while its core retention is driven by the quality of in-line suggestions. Chess.com is defined by its 100% fanaticism for chess, leading to constant internal dogfooding and an environment where AI augmentation (like Stockfish) is applied only where it directly serves the core user experience.
Brand, Marketing, and Experimentation Synergy
Copied to clipboard!
(00:41:48)
  • Key Takeaway: Marketing virality (like Duolingo’s Duo the Owl memes) and data-driven growth experimentation are not mutually exclusive but act as rocket fuel when combined.
  • Summary: The success of Duolingo’s brand personality, leveraged through TikTok and memes, directly translated into significant new user registrations tracked through ‘How did you hear about us’ data. Similarly, Chess.com benefited from external cultural waves (pandemic, Queen’s Gambit) that quadrupled overnight registrations. Growth experimentation provides the steady iteration, while external waves provide massive, exploitable spikes in volume.
AI Impact on Growth Work
Copied to clipboard!
(00:51:30)
  • Key Takeaway: AI accelerates the experimentation cycle by speeding up ideation, research synthesis, and prototyping, making the ’explore’ phase more accessible.
  • Summary: AI tools like ChatGPT significantly speed up the manual writing and synthesis of analysis documents, allowing growth teams to quickly extract insights and generate new specifications. The ability to visualize bolder ideas faster through AI prototyping (using tools like VZero) lowers the barrier to exploration. This acceleration makes it easier to oscillate between exploring new concepts and exploiting proven wins.
Experimentation Best Practices
Copied to clipboard!
(00:55:25)
  • Key Takeaway: Teams, especially on consumer products with scale, must overcome inertia and start running A/B tests immediately, as internal bias often leads to missed opportunities.
  • Summary: A significant percentage of product teams do not run experimentation, which is a major missed opportunity for consumer products with sufficient scale and frequency. Product teams naturally become power users, leading to a disconnect from the brand-new user experience, making experimentation essential to uncover hidden opportunities. The advice is to start small, perhaps using third-party tools, to build the practice of testing before attempting to build complex in-house systems.
Experimentation Tooling and Practice
Copied to clipboard!
(00:56:18)
  • Key Takeaway: Companies should prioritize running A/B tests quickly using third-party tools before committing to building in-house experimentation platforms.
  • Summary: Forgetting the new user experience leads to missed opportunities, necessitating the practice of running experiments, starting small (‘crawl, then walk, then run’). StatSig was used at Grammarly, while Duolingo and Chess.com utilize in-house tools due to their scale and specific needs. Building in-house tooling is generally not recommended from day one.
Chess.com Experimentation Goal
Copied to clipboard!
(00:57:22)
  • Key Takeaway: Setting an ambitious, self-made target like 1,000 annual experiments forces conversations that uncover necessary systemic changes across all departments.
  • Summary: Chess.com is shifting from practically no experimentation to an ambitious goal of 1,000 tests annually, up from 50 last year. Hitting the number is less important than the resulting insights, such as needing to enable experimentation in lifecycle marketing, app store optimization, and engineering for no-code configuration changes. This cultural shift required strong encouragement and alignment from the CEO and co-founders.
Motivating Users and Habit Formation
Copied to clipboard!
(01:04:42)
  • Key Takeaway: Effective habit formation in products like Duolingo is achieved by mastering the interplay between the core loop, the metagame, and the user profile.
  • Summary: Gamification success hinges on three pillars: the core loop (e.g., lesson, reward, streak extension), the metagame (long-term goals like leaderboards), and the profile (reflection of investment). For beginners at Chess.com, initial losses severely impact retention, prompting experiments like hiding ratings or crafting specialized ’learn to play’ experiences to guide users past early self-doubt.
Counterintuitive Team Building Lesson
Copied to clipboard!
(01:08:03)
  • Key Takeaway: In fast-moving environments, hiring for high agency—speed of thought, action, and learning—is often more valuable than hiring for deep, pre-existing industry experience.
  • Summary: High performers often possess high agency, energy, and clock speed, even without direct prior experience in the specific domain. In the age of AI, learned habits must sometimes be intentionally discarded, favoring those who can learn and move quickly. High agency can be observed through soft signals like the quality of questions asked and energy brought into interviews.
Company Size Preference
Copied to clipboard!
(01:10:38)
  • Key Takeaway: Medium-sized companies (500-1,000 people, 10-20 years old) offer a personal ‘Goldilocks zone’ balancing large-scale contribution with the ability to execute at a daily/weekly pace.
  • Summary: Big tech offers scale and best practices but can move slowly, while tiny startups offer speed but are grueling due to constant recruiting and user acquisition struggles. The preferred medium stage allows for contributing at scale while maintaining the ability to look at pixels and read experiment results daily. These durable, ideally profitable companies are often at key inflection points.
Failure Story: Chariot Direct
Copied to clipboard!
(01:13:44)
  • Key Takeaway: Launching a feature based on a ‘wouldn’t it be nice if’ solution without validating the user problem, while ignoring driver/operations impact and over-relying on pre-launch PR, leads to failure.
  • Summary: The attempt to introduce dynamic routes (‘Chariot Direct’) failed because it was a solution searching for a problem rather than solving a validated user need. The team focused too heavily on the rider app experience, neglecting the confusion and burden placed on drivers and operations staff. Promoting a feature heavily via PR before customer validation creates a sunk cost fallacy.
Lightning Round Insights
Copied to clipboard!
(01:18:42)
  • Key Takeaway: Reputation is paramount because small, consistent decisions about character compound to open surprising opportunities, while advertising success relies on compelling action, not just clever creative.
  • Summary: Recommended books included ‘Snuggle Puppy!’ (for kids), ‘Ogilvy on Advertising’ (for compelling action), and ‘Dark Squares’ (a memoir by Chess.com’s co-founder). Albert’s favorite product is his Breville Barista Express, and his life motto is that reputation is the most important asset, built by consistently doing the right thing.