Designing a Personalized Learning Recommendation System
Context
Eduspark an online learning platforms suffering from content overload, making it difficult for users to discover relevant courses.
As a Product Manager, I was tasked with designing a personalized learning recommendation system to improve:
- User engagement (+25%)
- Course completion rates (+30%)
- User satisfaction (+20%)
Problem statement
Learners struggle to discover relevant courses aligned with their goals and skill levels, leading to low engagement, high drop-offs, and poor learning outcomes.
Business objective
Implement a mobile-first gamification, personalized dashboards and behavioral nudges.
Goals
- +25% user engagement (DAU/MAU, session duration)
- +30% course completion rates
- +20% satisfaction (CSAT/NPS)
Research
College student
- Goals: Skill-building, internships
- Pain points: Overwhelmed, unclear path
Working professional
- Goals: Career growth, transitions
- Pain points: Time constraints, irrelevant content
User Journey
Key insight
The core shift in the journey is from “search-driven discovery” → “guided learning experience”, where the system proactively recommends the next best action instead of relying on users to figure it out.
HMW statements
“How might we” enable learners to discover and enroll in the most relevant courses tailored to their interests, skill levels, and career goals, so that we increase engagement, improve course completion rates, and enhance overall learning satisfaction?
North star metric
“Weekly Active Learners Completing at Least One Meaningful Learning Unit”
Assumptions
- Smart Onboarding & Personalization have High Reach & direct impact on engagement and completion
- Gamification has low to medium effort with strong confidence from known patterns
Solution design
- Smart Onboarding & Personalization
- Learning Experience & Content Delivery
- Engagement Systems—Gamification
- Dynamic Home Personalization
- Feedback Loop system
System design
Experimentation & validation strategy
- Implement A/B testing to experiment with personalized vs generic homepage
- Implement Learning paths for both personas and observe the metrics for 6 months
Tradeoffs
- Gamification depth: Use only badges & XPs, as it drives motivation without overwhelming users
- Personalized onboarding: Start rule based, with segmented course view
- Smart Notifications: Behavioral enrichment, 1/day cap with smart timing and user-adjustable settings
Prioritization
“I used RICE to prioritize onboarding, learning paths, and retention nudges as MVP features since they deliver the highest impact with lowest effort, while deferring advanced ML-based personalization to later phases.”
Visuals
In app feedback widget
With the expansion of Instant Resoution to broader consumer segment, the strategy was to capture user sentiment real time by providing an optional Share your feedback CTA
Outcomes
“If successful, the system is expected to drive the below uplift through higher conversion and retention.”
~22-25%
Weekly active learners
~28-32%
Course completion
20%
% Revenue increase
Key learnings
- Learned how to develop features & solution using prioritization framework like RICE
- While building this system, I realized that even the best recommendations don’t drive impact unless users are guided, motivated, and nudged to act.