Case study · Entertainment · 16 Weeks
Role
Lead Product Designer
Platform
Mobile (iOS & Android)
Duration
2 Years (25+ iterations)
Open full case study deck

Starting point
Personalization didn’t exist when I joined the team. Finding something to watch became increasingly difficult as the catalog grew to over 100k+ titles. Every user saw the same experience, regardless of their interests, creating an opportunity for us to make content discovery more relevant while helping the business increase engagement and viewing time.

Phase 1 : Proving the value
Rather than redesigning the experience all at once, I identified key moments throughout the journey where we could collect meaningful signals. Each experiment helped us validate assumptions while gradually improving recommendation quality.

Design explorations
I explored three interaction patterns to optimize onboarding for speed, focus, and recognition.
• Pills — Fast and playful, but difficult to scan.
• List — Familiar and detailed, but slower to browse.
• Grid — Fast to scan, recognizable, and scalable.


Phase 2 : The Framework
Our initial experiments proved personalization worked, but they also exposed a limitation—we were asking users to define themselves only once. The bigger opportunity was building a system that continuously learned from their evolving behavior.









This incremental approach increased completion from 66% to 79% while growing preference signals from 3 to 10+ selections. Each iteration built on previous learnings, refining both the interaction model and the quality of signals captured.
Takeaway
Personalization evolved through a series of carefully measured experiments rather than a single solution.
Results
Personalization became a core pillar of Tubi's discovery experience, reducing time to relevant content through continuous experimentation and iterative improvements. The initiative earned executive support, expanded beyond onboarding into multiple product surfaces, and established a scalable foundation for future personalization.


Explore more selected work.
Back to work