The challenge
When I joined the team, the acquisition numbers were worth celebrating. Marketing was pouring users into the funnel, and sign-ups were at an all-time high. But when I pulled the cohort retention data for Day 30, the celebration stopped.
65% of our learners were gone within a month, it's not bad in EdTech industry but insufficient for the company's growth targets.
The prevailing theory in the product team was a usability issue. The belief was that if we just made the interface cleaner and the buttons brighter, people would stay. I suspected the problem was deeper.
Understanding the learners
Before redesign anything, the deep-dive user research revealed that 70% of our active users were career-driven.
Key Insight: This segment wasn't motivated by "fun" or "badges" in the traditional sense. They were motivated by professional fear (fear of falling behind) and competency proof (need for validation).
The missed opportunity: The existing design treated them like casual hobbyists, offering generic progress bars that failed to map to real-world career value.
Top challenges from student survey
Over 1,500 students took the survey. Stated top challenges are common across all student persona/segments.
It’s hard to stick to goals
Not knowing how to solve problems
Low efficiency
Get bored
Not knowing what to do next
Can’t find learning materials
Being distracted by other mobile apps
No problem
Others
App crash
19%
13%
12%
12%
11%
10%
10%
4%
4%
4%
A few follow up interview conducted to understand the reason
These interviews were particularly helpful in uncovering the reasons behind user behaviours, providing qualitative data that added depth to our understanding.
The strategy and design
We moved away from ad-hoc feature updates (like V5 final-final) toward a hypothesis-driven growth model.
Hypothesis A: Users will fight harder to protect a digital asset they have built than to gain a new one. We needed to weaponise the "sunk cost fallacy."
Hypothesis B: For the career segment, time-to-competency is the only metric that matters. We needed to validate their skills immediately.
Mapping current systems
We mapped every screen in the old systems to understand user flows and pain points. This thorough mapping gave us a solid foundation for improvement and innovation.
Also we built a scalable design system to maintain consistency and usability.
The execution
Added "friction" to the onboarding process. Users now wager their time, setting a specific career goal (e.g., "Interview in English") and committing to a schedule.
Users who signed the commitment contract showed a 10% higher activation rate in week 1.
Updated the progress bar with a Streak Freeze and Wager system where users could bet virtual currency on their own consistency.
This shifted the daily trigger from "I should learn" (high friction) to "I don't want to lose my streak" (loss aversion).
Restructured the information architecture to align with career outcomes. Instead of "Level 4 Complete," the milestone became "Certified: Business Introduction."
These micro-certs were designed to be instantly shareable on LinkedIn, closing the loop between app effort and career reward.
The result
A design system was build the same time to maintain the consistency between B2B and B2C. By moving beyond user satisfaction and focusing on behavioural economics, the redesign delivered measurable business value:
📈
Retention uplift: Day 30 retention increased from 35% to 45%, significantly impacting the Life Time Value of the cohort.
💰
Trial-to-paid conversion: The commitment contract onboarding drove a 15% increase in trial conversions, directly contributing to annual recurring revenue.
🧠
Cultural Shift: The design team effectively deprecated vanity metrics (CSAT) in favour of north star retention metrics, aligning design directly with the P&L.
The challenge
When I joined the team, the acquisition numbers were worth celebrating. Marketing was pouring users into the funnel, and sign-ups were at an all-time high. But when I pulled the cohort retention data for Day 30, the celebration stopped.
65% of our learners were gone within a month, it's not bad in EdTech industry but insufficient for the company's growth targets.
The prevailing theory in the product team was a usability issue. The belief was that if we just made the interface cleaner and the buttons brighter, people would stay. I suspected the problem was deeper.
Understanding the learners
Before redesign anything, the deep-dive user research revealed that 70% of our active users were career-driven.
Key Insight: This segment wasn't motivated by "fun" or "badges" in the traditional sense. They were motivated by professional fear (fear of falling behind) and competency proof (need for validation).
The missed opportunity: The existing design treated them like casual hobbyists, offering generic progress bars that failed to map to real-world career value.
Top challenges from student survey
Over 1,500 students took the survey. Stated top challenges are common across all student persona/segments.
It’s hard to stick to goals
Not knowing how to solve problems
Low efficiency
Get bored
Not knowing what to do next
Can’t find learning materials
Being distracted by other mobile apps
No problem
Others
App crash
19%
13%
12%
12%
11%
10%
10%
4%
4%
4%
A few follow up interview conducted to understand the reason
These interviews were particularly helpful in uncovering the reasons behind user behaviours, providing qualitative data that added depth to our understanding.
The strategy and design
We moved away from ad-hoc feature updates (like V5 final-final) toward a hypothesis-driven growth model.
Hypothesis A: Users will fight harder to protect a digital asset they have built than to gain a new one. We needed to weaponise the "sunk cost fallacy."
Hypothesis B: For the career segment, time-to-competency is the only metric that matters. We needed to validate their skills immediately.
Mapping current systems
We mapped every screen in the old systems to understand user flows and pain points. This thorough mapping gave us a solid foundation for improvement and innovation.
Also we built a scalable design system to maintain consistency and usability.
The execution
Added "friction" to the onboarding process. Users now wager their time, setting a specific career goal (e.g., "Interview in English") and committing to a schedule.
Users who signed the commitment contract showed a 10% higher activation rate in week 1.
Updated the progress bar with a Streak Freeze and Wager system where users could bet virtual currency on their own consistency.
This shifted the daily trigger from "I should learn" (high friction) to "I don't want to lose my streak" (loss aversion).
Restructured the information architecture to align with career outcomes. Instead of "Level 4 Complete," the milestone became "Certified: Business Introduction."
These micro-certs were designed to be instantly shareable on LinkedIn, closing the loop between app effort and career reward.
The result
A design system was build the same time to maintain the consistency between B2B and B2C. By moving beyond user satisfaction and focusing on behavioural economics, the redesign delivered measurable business value:
📈
Retention uplift: Day 30 retention increased from 35% to 45%, significantly impacting the Life Time Value of the cohort.
💰
Trial-to-paid conversion: The commitment contract onboarding drove a 15% increase in trial conversions, directly contributing to annual recurring revenue.
🧠
Cultural Shift: The design team effectively deprecated vanity metrics (CSAT) in favour of north star retention metrics, aligning design directly with the P&L.
Optimising LTV for EF online learning platform
2022 - 2024
Huan Feng
©2026
huanfengme@gmail.com
Huan Feng
©2026
huanfengme@gmail.com
Huan Feng
©2026
huanfengme@gmail.com
Huan Feng
©2026
huanfengme@gmail.com