Finding hidden growth opportunities in your product | Albert Cheng (Duolingo, Grammarly, Chess.com)
Executive Summary
Albert Cheng discusses consumer-product growth at Duolingo, Grammarly and Chess.com: freemium, retention, reactivation, metric ownership and experimentation. Selected third-party transcript coverage excludes AI and chess-engine discussion and the lightning round. Company examples and numbers are his accounts, unverified here. Untimed editorial topics are not video chapters.
Key Takeaways
Keep a real free core
Cheng connects freemium to broad access and word-of-mouth growth. His preference is a permanently free core value proposition with a taste of premium features layered on top. Free users can also introduce a product to coworkers. This is his experience, not a universal pricing prescription.
Choose trials for the product’s behavior
He says reverse trials can work where users invest in setup and build a reason to stay, such as some B2B products. Consumer products can behave differently. The choice should follow how value and continued use develop rather than copy a successful company's paywall.
Match retention work to actual usage
For a daily-use product, habitual users and their retention can become a major growth driver. Grammarly taught him a different lesson: installation and activation mattered because users did not proactively open it daily. One retention playbook does not fit every product.
Give returning users a good re-entry
Cheng describes mature products accumulating inactive and sporadic users. At Duolingo, returning users could reassess their level rather than resume an unsuitable old lesson. Improving re-entry can be useful alongside new-user acquisition; his examples and performance claims are not independently verified here.
Let the right team own the metric
He staffed a growth team to improve Grammarly's repeated use, then concluded the core product team had more influence through suggestion quality. He shifted the work to that team. Adding a growth team is not progress if it gets between the problem and the people best placed to solve it.
Start experimenting before building a platform
Cheng encourages teams to begin with a manageable test and available tools. In-house systems helped some large companies, but he does not recommend building one from day one. The aim is to reduce wrong assumptions about users, not to recreate another company's infrastructure.
Use an experiment goal to reveal constraints
He describes Chess.com's ambitious experiment-count target as a way to discuss tooling, ownership and which teams could test. A thousand tests without learning or impact would waste time. Celebrate useful results and faster learning rather than rewarding volume alone.
The Conversation Breakdown
Keep a real free core
⏱ Untimed topicCheng connects freemium to broad access and word-of-mouth growth. His preference is a permanently free core value proposition with a taste of premium features layered on top. Free users can also introduce a product to coworkers. This is his experience, not a universal pricing prescription.
Choose trials for the product’s behavior
⏱ Untimed topicHe says reverse trials can work where users invest in setup and build a reason to stay, such as some B2B products. Consumer products can behave differently. The choice should follow how value and continued use develop rather than copy a successful company's paywall.
Match retention work to actual usage
⏱ Untimed topicFor a daily-use product, habitual users and their retention can become a major growth driver. Grammarly taught him a different lesson: installation and activation mattered because users did not proactively open it daily. One retention playbook does not fit every product.
Give returning users a good re-entry
⏱ Untimed topicCheng describes mature products accumulating inactive and sporadic users. At Duolingo, returning users could reassess their level rather than resume an unsuitable old lesson. Improving re-entry can be useful alongside new-user acquisition; his examples and performance claims are not independently verified here.
Let the right team own the metric
⏱ Untimed topicHe staffed a growth team to improve Grammarly's repeated use, then concluded the core product team had more influence through suggestion quality. He shifted the work to that team. Adding a growth team is not progress if it gets between the problem and the people best placed to solve it.
Start experimenting before building a platform
⏱ Untimed topicCheng encourages teams to begin with a manageable test and available tools. In-house systems helped some large companies, but he does not recommend building one from day one. The aim is to reduce wrong assumptions about users, not to recreate another company's infrastructure.
Use an experiment goal to reveal constraints
⏱ Untimed topicHe describes Chess.com's ambitious experiment-count target as a way to discuss tooling, ownership and which teams could test. A thousand tests without learning or impact would waste time. Celebrate useful results and faster learning rather than rewarding volume alone.
Core Frameworks & Ideas
Free core plus premium sampling
Keep the core value accessible while letting users experience premium benefits. Whether this fits depends on the product and economics.
Metric-owner check
Ask which part of the product actually changes the metric and give that team ownership.
Experiment goal as constraint finder
Use an ambitious target to expose tooling and team needs, then judge the work by learning and impact.
Actionable Next Steps
- ✓Audit whether your free tier delivers the core value users came for.
- ✓Compare new, habitual and returning users before choosing retention work.
- ✓Ask whether a growth team helps or blocks the actual metric owner.
- ✓Run one manageable experiment and write what it taught you, not only whether it won.
Notable Quotes
“And so it was an interesting learning in that I staffed up a growth team, tried to work on this metric, and then I realized actually I'm just getting in the way.”
“Obviously Duolingo is an experimentation machine, and so it's been a huge accelerant to have our own thing specifically tailored to be excellent at that.”
“People have to see it working and when it works, the metrics move and you learn faster and you ship faster, and that's a great environment to be part of.”
The One Thing to Remember
Growth work should connect users to value and help the right team learn, not merely add tests or another team.