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

Oct 5, 2025 Episode Page ↗
Overview

Guest Albert Chang, a top consumer growth mind, discusses his explore and exploit framework, key growth wins at Duolingo, Grammarly, and Chess.com, and how AI accelerates growth work. He emphasizes user retention, building habits, and the power of brand and community.

At a Glance
25 Insights
1h 25m Duration
14 Topics
4 Concepts

Deep Dive Analysis

Albert Cheng's Background and Growth Philosophy

Explore and Exploit Framework for Growth

Applying Explore-Exploit: Chess.com Game Review Example

Leveraging AI for Growth: Text-to-SQL and Prototyping

Grammarly's Biggest Monetization Win: Reverse Trials

Retention and Growth in Consumer Subscription Products

Differences in Operating Models: Duolingo, Grammarly, Chess.com

The Role of Brand and Community in Growth

AI's Impact on Chess and Growth Workflows

Best Practices for Experimentation at Scale

Lessons from a Failed Product: Chariot Direct

Hiring for High Agency and Learning Speed

Finding Your Ideal Company Stage

Personal Insights and Life Motto

Explore and Exploit Framework

This framework involves two modes: 'explore' for finding new growth opportunities (like finding the right mountain to climb) and 'exploit' for focusing resources to scale those opportunities effectively (climbing that mountain). It helps teams oscillate between divergent thinking and focused execution.

Reverse Trials (Monetization)

A monetization strategy for freemium products where free users are given a limited, interspersed taste of premium features within their regular usage, rather than a time-based full trial. This approach helps users perceive the product as more powerful and can significantly boost upgrade rates.

Gamification Pillars

A model for building long-term user engagement, consisting of three components: the core loop (daily habits and immediate rewards), the metagame (long-term achievements and goals like leaderboards), and the profile (a reflection of a user's investment and progress over time).

High Agency (Hiring)

A desirable trait in hires, characterized by high 'clock speed' (thinking and moving fast), energy, and a rapid learning ability. This is often prioritized over deep prior experience, especially in rapidly evolving fields like AI, where learned habits may need to be discarded.

?
What is the biggest missing piece people don't get about building a successful consumer subscription product?

User retention is paramount; without strong retention, companies rely too heavily on getting users to pay on day one, which is a much harder business model.

?
How can a freemium product like Grammarly significantly increase upgrade rates?

By sampling various paid suggestions and interspersing them to free users, providing a limited taste of the premium offering, which makes the product seem more powerful and encourages upgrades.

?
What does good new user retention look like for a consumer app?

A Day 1 (D1) retention rate of around 30-40% is considered solid for a consumer app, indicating a healthy initial engagement.

?
How do mature consumer products sustain growth beyond new user acquisition?

For mature products, a significant portion (around 80%) of active users are existing users, and reactivated or 'resurrected' users are a similar size to new users, making their re-engagement a crucial growth lever.

?
How has AI impacted the game of chess?

AI chess engines are dramatically better than top human grandmasters (e.g., 3600 ELO vs. 2800 ELO), but this has opened up new creativity, strategies, and appreciation for the game, with tools like game review augmenting the human playing experience.

?
How can companies shift to a culture of experimentation?

It requires strong leadership support, celebrating early wins to motivate teams, and demonstrating how experimentation leads to faster learning, shipping, and metric movement across the organization.

?
What are common pitfalls when building new products, especially in marketplace businesses?

Pitfalls include solution-searching for a problem, failing to consider all user types (e.g., drivers in a marketplace alongside riders), and doing PR before validating customer demand, which can lead to sunk costs.

?
What company stage is ideal for Albert Cheng's working style?

Albert finds his 'goldilocks zone' in medium-sized companies (around 500-1000 people, often 10-20 years old) where he can contribute at scale while still getting into details and executing at a daily/weekly pace.

1. Connect Users to Value

Frame growth as the job of connecting users to the value of your product, rather than merely metrics hacking, to ensure a holistic and durable approach to company growth.

2. Prioritize User Retention

Focus on high user retention for consumer subscription products, as it reduces the pressure to monetize users immediately and is crucial for long-term, sustainable growth.

3. Sample Premium Features

For freemium products, offer a limited ’taste’ of premium features to free users, interspersing them into their experience to showcase the product’s full power and significantly boost upgrade rates.

4. Apply Explore & Exploit

Utilize the explore and exploit framework at a micro (insight) level: explore to find unexpected learnings, then exploit by applying that insight broadly across related product areas to maximize impact.

5. Reframe Negative User Feedback

When users experience negative outcomes (e.g., losing a game), reframe feedback to highlight positive aspects (e.g., brilliant moves) and offer encouragement, which can significantly improve engagement and retention.

6. Share Experiment Learnings Broadly

Disseminate successful (or even unsuccessful) experiment insights across the entire company, enabling other teams to apply the learnings to their own product areas and multiply the overall impact.

7. Hire High Agency Individuals

Prioritize hiring individuals with high agency, ‘clock speed,’ and energy, as their ability to learn and adapt quickly can be more valuable than deep, specific experience, especially in rapidly evolving fields like AI.

8. Leverage Brand & Community

Combine consistent growth experimentation with leveraging brand, community, and cultural moments (e.g., social media trends) to create ‘rocket fuel’ for growth, driving massive, sudden increases in user acquisition.

9. Use AI for Data Analysis

Implement AI-powered text-to-SQL tools (e.g., Slack bots) to automate ad-hoc data queries, making data more accessible and fostering a more data-informed culture by reducing friction for asking questions.

10. Accelerate Prototyping with AI

Utilize AI prototyping tools (e.g., V0, Figma Make) to quickly generate foundational screens and product flows, accelerating the ideation-to-test cycle and making bold ideas more discussable and testable.

11. AI for User Value, Not Hype

When integrating AI, prioritize customer value and apply the right technology for specific features, rather than blindly adopting the latest hyped AI trend without clear user benefit.

12. Start Experimenting Immediately

Begin experimentation without delay, even with simple A/B tests or third-party tools, to establish a practice of continuous learning and iteration, especially for consumer products with scale and frequency.

13. Set Ambitious Experiment Goals

Set ambitious experimentation goals (e.g., 1,000 experiments/year) not just to hit a number, but to drive conversations about necessary systemic changes (e.g., no-code tools, broader team involvement) to achieve such scale and impact.

14. Cultivate Experimentation Culture

To shift company culture towards experimentation, secure strong leadership buy-in and active advocacy from the top, and consistently celebrate experiment wins and learnings to energize and motivate teams.

15. Build a Robust Experiment System

Invest in a robust experimentation system, including a clear growth model and thorough product instrumentation, as the underlying system is more critical for long-term success than any single experiment.

16. Track Screenshots for Virality

Temporarily track in-app screenshots to identify organic ‘hotspots’ where users are already sharing content, then enhance those moments with delightful experiences (e.g., illustrators, animators) to amplify virality.

17. Gamification: Core Loop, Metagame, Profile

Design products for habit formation using a three-pillar gamification model: a tight ‘core loop’ (daily action, reward, streak), an engaging ‘metagame’ (leaderboards, long-term goals), and a meaningful ‘profile’ (reflection of user investment).

18. Support Learning Product Beginners

For learning-oriented products, intentionally design beginner experiences to mitigate self-doubt and negative reinforcement, guiding new users through initial challenges (e.g., hiding ratings, offering guided play).

19. Avoid Solution-Searching for Problems

When building new products or features, always start with a clear user problem to solve, rather than developing a solution and then searching for a problem it might fit.

20. Consider All Marketplace Users

In marketplace or multi-sided businesses, ensure product development considers the experience and needs of all user types (e.g., riders, drivers, operations) to avoid negative impacts on one group undermining the overall product.

21. Validate Before Public Relations

Avoid extensive public relations or marketing efforts for new features or products before validating strong customer demand, as premature PR can lead to sunk costs and pressure to continue with an unvalidated idea.

22. Cultivate a Strong Reputation

Cultivate a strong professional reputation by consistently making ethical decisions and treating people well, as these small actions compound over time to open doors and create opportunities.

23. Monitor Experiment Saturation

Monitor experiment results for declining statistical significance; if many experiments in an area yield insignificant results, it signals saturation and a need to shift back to exploratory, divergent thinking.

24. Target 30-40% D1 Retention

Aim for a Day 1 (D1) retention rate of 30-40% for consumer apps, as this indicates a solid foundation for growth and user acquisition.

25. Invest in Resurrection Experience

For mature products with a large base of dormant users, invest in crafting an excellent ‘resurrection’ experience and novel re-engagement strategies to bring them back, as reactivated users can be a significant growth component.

User retention is gold for consumer subscription companies.

Albert Cheng

Sometimes experience can be a crutch, especially in this world where the grounds are shifting so fast with AI. A lot of your learned habits actually need to be intentionally discarded.

Albert Cheng

The job is to connect users to the value of your product.

Albert Cheng

When you do find a thing that breaks through the noise, and it could actually be a hugely losing experiment, too. Those are also super valuable, right? Surfacing those across the company.

Albert Cheng

If you do too much exploration, you can have your team feel a little bit too scattershot, just trying a hundred different random ideas. What's the through line? What's the strategy? How do you pattern match successes across them?

Albert Cheng

If you do too much in exploitation, which is often the MO of growth teams, it can lead to this like saturation and stagnation where you're just locally maximizing a thing.

Albert Cheng

I think doing it before you have validation that customers definitely want the thing is quite risky. It can lead to a lot of sunk costs, once you get it out because you're, you're just, you know, you need to see it through. You want to see it succeed.

Albert Cheng

Gamification Pillars for Long-Term Learning Journey

Albert Cheng, referencing Jorge Mazal
  1. Establish a tight 'core loop' for daily habits (e.g., complete a lesson, get rewards, extend streak, receive push notification).
  2. Provide a 'metagame' for long-term motivation through achievements, leaderboards, or a clear path to strive for.
  3. Allow users to build a 'profile' reflecting their investment and progress within the product experience.

Experimentation Best Practices for Teams

Albert Cheng
  1. Start somewhere: Begin by running A/B tests, even with simple third-party tools or in-house solutions, to get into the practice.
  2. Understand your growth model: Have a clear understanding of how your company grows and which channels you will leverage.
  3. Instrument your product thoroughly: Ensure comprehensive tracking of product events to avoid wonky results and ensure data accuracy.
Over 90%
Percentage of Grammarly users on the free service This highlights the importance of freemium monetization strategies.
Nearly doubled
Increase in Grammarly upgrade rates Achieved by sampling paid suggestions to free users.
80%
Percentage of Chess.com users reviewing games after a win Counterintuitive, as initial assumption was users would review after losses to learn from mistakes.
25%
Increase in Chess.com game reviews Resulted from changing the loss review experience to highlight brilliant/best moves and provide encouragement.
20%
Increase in Chess.com subscriptions Resulted from changing the loss review experience to highlight brilliant/best moves and provide encouragement.
30-50%
Typical experiment win rate for growth teams Many hypotheses turn out not to be true in consumer products.
About 50
Chess.com experiments run in 2023 Company practically didn't experiment at all prior to 2023.
About 250
Chess.com experiments on pace for current year Significant increase from the previous year.
1,000
Target for Chess.com experiments per year An ambitious target set to drive conversations about necessary changes and enablers, not just the number itself.
Over 75%
Percentage of new Chess.com users classifying as new or beginner Highlights the need for tailored onboarding experiences for novice players.
Less than a third
Win rate for new/beginner Chess.com users in their first game Indicates a challenging initial experience for new players.
10% worse
Impact of losing a game on user retention in Chess.com Retention is 10% lower for users who lose a game compared to those who win.
About 1,000 people
Duolingo company size Example of a successful medium-sized company.
$12 billion
Duolingo company valuation As of the time of the podcast.
Around 1000-1500
Average chess player ELO rating For comparison with top players and AI engines.
Around 2800
Top Grandmaster ELO rating (Magnus Carlsen) Represents the peak of human chess skill.
Around 3600
Stockfish engine ELO rating Illustrates the dramatic superiority of AI chess engines over humans.
40%
Percentage of product teams not running experimentation According to an Atlassian state of product report.
About 1800
Albert Cheng's Chess.com rapid rating His personal rating for 10-minute games.
About 1500
Albert Cheng's Chess.com blitz rating His personal rating for 3-minute games.