Transforming search from filters to intelligent content discovery
ROLE
Product Design Lead, leading strategy and end-to-end product experience
SCOPE
Evolving content discovery through AI-powered semantic search across multiple products
Context
TL:DR
Challenge
The existing search experience could no longer support the growing complexity of content
Users struggled when they:
Wanted to discover content rather than search for a specific item
Expected search to understand what they meant instead of matching keywords
Old search problem
Old search prototype
Research & insight
To better understand how search should evolve, I explored user workflows, analyzed existing behavior, researched AI search patterns, and collaborated closely with Product and Engineering
Insight 01
Users have different discovery needs
Precision SEARCH
User knows exactly what they need
↓
↓
DISCOVERY SEARCH
↓
↓
Insight 02
Content evolved beyond structured search
As content expanded, users needed to discover beyond predefined attributes and explore through context and meaning.
Before
Core sports content
Games, Highlights
Platform evolution
Content expanded beyond games
Press conferences, Interviews, Shows, News
What changed
Need for a smarter search model
Search needed to understand user intent.
Design decision
Decision 01
Create a hybrid search experience
Introducing AI-powered discovery while preserving existing workflows
The goal was not to replace filters with AI, but to create a more flexible search experience that supports different user needs.
Before
After
Decision 02
Designing a scalable search experience across products
Creating a consistent search model that can evolve with the platform
Search was a core capability across multiple products: Library, Mini Library, and Mobile.
I designed a scalable search pattern that could adapt to different contexts while maintaining consistency across the platform.
Search in studio & mobile
Exploration
Turning complex metadata into an intuitive search experience
Direction 01
The metadata behind Search
Search relies on a rich content model spanning people, teams, events, topics, and content types.
We explored how this structure could support a more intelligent search experience while remaining invisible to users.
Metadata panel
Direction 02
Connecting intent to metadata
Users don't think in metadata. They describe what they're looking for in their own language.
We explored how natural-language queries could be translated into the structured metadata Search already understands.
Final design
A smarter and more flexible search experience
The final experience combines natural language search with structured filtering, allowing users to move seamlessly between exploration and precision. The same search model now supports multiple products across the platform.
Preview search
Search by filters
Search by free text
Search by free text
Validation & Impact
Following launch, I analyzed adoption to understand how users interacted with the new experience
Faster content discovery: Users could reach relevant content more quickly.
More accurate results: Semantic understanding and richer metadata improved search relevance.
Validated hybrid experience: ~40% of users still used filters, validating the hybrid experience.