Semantic Search

🎯 The Challenge

Content editors struggle to find relevant content in large libraries when they don't know exact keywords or titles. Traditional keyword search misses semantically similar content.

🎨 What I Designed

Complete semantic search experience transforming how content editors discover and find relevant items.

• Semantic search integration into existing Content Inventory UI
• "Whisper/Suggest" component – prompt-engineered suggestions teaching users effective semantic query patterns
• Relevance explanation UI – helping editors understand why specific results appeared
• Discoverability solution – moved semantic search from hidden Innovation Lab to prominent position in main UI
• Hybrid search architecture – designed integration of semantic search with traditional filters
• Filter logic definition – established clear rules for intersection vs. union behavior

🔬 Research & Strategy

Applied Jobs To Be Done framework and collaborative design process.

• "Stacey wants to quickly find the best matching content item"
• "Stacey needs reliable search results – no need to double-check"
• "Stacey wants to understand why a result appeared"
• Facilitated 3-hour Design Studio workshop (Nielsen Norman Group methodology)
• Led internal user research sessions with Content Inventory power users
• Cross-functional collaboration with PM, engineering, and stakeholders

🧩 Key Design Challenges

Balancing multiple complex considerations:

• Paradigm Shift – Balancing semantic-first approach with users' established mental models from traditional keyword search
• Explainability for Non-Technical Users – Making AI relevance scoring understandable to content editors
• Unified Experience – Seamlessly integrating two search paradigms into a single, intuitive interface
• Trust & Reliability – Designing transparency features that build user confidence in AI-powered results

🌊 Impact