YEAR
2023
COMPANY
ROLE
Product Designer
Information Architecture
Wireframe
Prototyping
Visual Design
INDUSTRY
Technology
SUMMARY
This project tackled the challenge of improving Google Search's ability to handle conversational and complex queries. Users increasingly seek information through natural language and open-ended questions. The goal was to redesign the search experience to deliver more direct, informative, and helpful responses.
device
Desktop
PLATFORM
Web
💡 Summary
Understanding how people naturally ask questions is vital for improving search interfaces. Investing in language processing technology makes a big difference.
Results:
Increased click-through satisfaction for conversational searches by 22%.
Reduced "reformulated" queries (users needing to rephrase their search) for complex topics by 10%.
Improved user feedback on the clarity and directness of search results.
Overview
Problem
Google Search, while incredibly powerful, often presented fragmented results for conversational queries (like "What's the best sushi place close to my office?") and complex, open-ended questions (like "How do I build a basic website?").
Goal
Redesign Google Search to provide more direct, informative, and helpful responses to these increasingly common queries.
My Role
UI/UX Designer, embedded in a cross-functional team that included researchers, engineers, and product managers.
Process
Interviews
Conducted in-depth interviews with 12 Google Search users, focusing on their recent experiences with voice search and asking complex questions.
Search Log Analysis
Examined over 100,000 recent search queries, noting a 15% year-over-year increase in conversational style questions.
Competitive Analysis
Evaluated the query handling of Bing, DuckDuckGo, and Siri to identify potential areas where Google could differentiate itself.
Design
Wireframing
Designed mid-fi wireframes that detailed the flow of "Knowledge Panels" for complex searches and conversational results with integrated maps and business info.
Visual Design
Developed hi-fi mockups in Figma, ensuring that while new, the design still felt distinctly "Google" in its style.
Prototyping
Built clickable prototypes in Framer to demonstrate dynamic interaction with the search results.
User Testing
Usability Labs
Conducted 8 moderated sessions where participants performed searches in both the old and new interface, focusing on conversational tasks (finding restaurants) and complex tasks (researching basic coding).
A/B Testing
Launched the new design to 5% of users, monitoring click-through rates, dwell time, and recorded qualitative feedback surveys.
Results
Qualitative Feedback
Users expressed delight with the new "answer card" format for conversational queries, finding they got the info they needed faster.
Quantitative Metrics
Showed a 22% increase in click-through satisfaction for conversational searches and a 10% decrease in "reformulated" queries for complex topics.
Iteration
We removed a "Related Questions" section from complex search results due to user confusion and added clearer "Refine Search" options.
Learnings
NLP Advancements
Collaborating with our Natural Language Understanding team was critical for success, as their models were key to interpreting questions more accurately.
Data-Driven Design
The metrics from initial user research and A/B testing validated our design decisions.
The "Long Tail"
Designing for conversational search also helped users with less common niche queries get more relevant results.