[Case 02]
Positioning SIA as the First Major Airline to partner with OpenAI
Airline
Transforming Singapore Airlines' Chatbot Experience: From User Research to AI Innovation
[Project Overview]
This comprehensive case study highlights my role as Senior UX Strategist in conducting groundbreaking usability research on Singapore Airlines' Kris chatbot that ultimately influenced the airline's historic partnership with OpenAI in April 2025. Through rigorous mixed-methods research, I identified critical user experience issues that led to transformative AI integration, positioning Singapore Airlines as the first major airline to partner with OpenAI for advanced conversational AI capabilities.
The research revealed fundamental limitations in the existing chatbot architecture and provided compelling evidence for strategic technology partnerships, directly contributing to a business transformation that enhanced both customer experience and operational efficiency.
[Problem Statement]
Despite Singapore Airlines' reputation for service excellence, the Kris chatbot was significantly underperforming in user adoption and satisfaction metrics. The digital touchpoint was failing to meet modern user expectations, creating a disconnect between the airline's premium service brand and its digital customer service capabilities.
[Industry]
Airline
[My Role]
Lead Researcher & Designer
[Platforms]
Desktop
Mobile
[Timeline]
January 2024- August 2024
The business context was critical: industry data indicates that 50% of airline customer interactions are expected to be automated using AI-powered chatbots by 2025, making this research essential for maintaining competitive advantage. With the global conversational AI market projected to grow from $12.24 billion in 2024 to $61.69 billion by 2030 at a CAGR of 22.6%, the strategic importance of this initiative was paramount.
[Research Objectives]
I designed the study to address four critical business questions:
How easily can users discover and interact with the chatbot?
What are the primary pain points in the current user experience?
How does the chatbot compare to modern conversational AI standards?
What improvements would enhance user satisfaction and drive strategic advantage?
[Research Methodology]
Mixed-Methods Approach:
I developed and executed a comprehensive usability study combining quantitative metrics with qualitative insights to understand both performance issues and underlying user motivations. The study design included:
Participants: 10 representative users with diverse backgrounds and travel experience levels
Testing Method: Moderated usability sessions with task-based scenarios
Duration: 45-60 minutes per session
Data Collection: Screen recording, task completion metrics, satisfaction surveys, and post-test interviews
Three-Phase Research Framework:
Phase 1: Discoverability Assessment
Measured chatbot discovery time and navigation pathways
Evaluated welcome message clarity and initial user understanding
Assessed accessibility and findability across different user types
Phase 2: Contextual Task Performance
I designed nine realistic user scenarios spanning the full spectrum of customer service needs:
Flight search and booking assistance
Baggage inquiries (allowance, excess, non-standard items)
Travel services (check-in, flight status, lounge access)
Support requests (lost items, student privileges)
Phase 3: Holistic Experience Evaluation
Usability heuristics assessment using established UX principles
Visual design and brand perception analysis
Conversational tone and personality evaluation
[Chatbot Discoverability]
The research revealed severe discoverability problems that fundamentally limited chatbot adoption. Users required an average of 55.7 seconds to locate the chatbot, with significant variance between fastest (28 seconds) and slowest (2 minutes 5 seconds) discovery times.
This data indicated poor information architecture and insufficient prominence for a critical customer service touchpoint.
[Testing Chatbot Interactions]
Test Cases:
Flight Availablity / Search
Baggage Allowance
Excess Baggage
Flight Status
Check-In
Non-standard Baggage
Lost Items
Lounge Eligibility
Student Privileges
Task Performance Analysis:
Performance varied dramatically across different use cases, revealing inconsistent user experience quality.
The easiest tasks included Baggage Allowance, Check-in procedures, and Lost Items, while Flight Search and Lounge Eligibility presented the greatest challenges.
Response clarity ratings were generally positive (4.3/5 average), indicating that when users successfully navigated to relevant information, the content quality was acceptable. However, the interaction design revealed strong user preferences for button-based interactions (3.9/5) over text input (3.2/5), suggesting opportunities for structured conversation design.
[Visual Design Problems]
Outdated interface design (2.2/5) that felt dated compared to modern chatbot standards.
Unclear terminology (2.1/5) causing user confusion and task abandonment.
[Interaction Design Failures]
Inconsistent button placement disrupting user mental models
Missing contextual memory preventing natural conversation flow
Text-heavy responses (3.2/5) creating information overload
[Technical Reliability Issues]
Missing loading indicators causing user uncertainty
Conversation flow interruptions requiring frequent restarts
Poor mobile optimisation despite mobile-first user behaviour
[Design Recommendations and Strategic Roadmap]
Quick Wins:
Based on the research findings, I developed a prioritised improvement strategy addressing both immediate usability issues and long-term strategic positioning.
Enhanced Discoverability:
Implement prominent homepage chatbot entry point
Add floating action button for consistent access across all pages
Redesign navigation architecture to surface chat functionality
The GIF below shows the chatbot entry point with 2 variations - one for pages that have built-in feedback forms during key user flows and the other for general pages on the website. I created Lottie animations to bring a sense of delight for any user who hovers on the chatbot entry point.
Conversation Design Improvements:
Replace vague "Flying with us" option with specific, actionable choices
Implement contextual memory for seamless multi-topic conversations
Develop progressive disclosure for complex information presentation
[Strategic UX Transformation]
Visual Interface Modernisation:
The research highlighted the need for contemporary conversational UI patterns that align with user expectations established by leading chat platforms. This included mobile-first design optimization, structured information presentation replacing text-heavy blocks, and consistent implementation of Singapore Airlines brand guidelines.
After Redesign
Before
Content Strategy Revolution:
Users demonstrated clear preference for scannable, structured information over detailed paragraphs. The recommendation included implementing smart defaults using customer data, creating personalised conversation flows based on user history, and developing visual response formats optimised for mobile consumption.
Checked Baggage Inquiry
Lounge Inquiry
Flight Status Inquiry
Advanced AI Integration Roadmap
The research findings revealed a fundamental limitation: the rule-based chatbot architecture couldn't scale to meet modern user expectations for natural, contextual conversations.
These insights and the research done above eventually led to the strategic recommendation for a partnership with leading AI technology provider, OpenAI, to implement:
Natural language understanding capabilities are moving beyond rigid menu navigation
Multimodal interaction support enabling text, voice, and image interactions
Predictive assistance based on travel patterns and customer behaviour
Seamless escalation to human agents for complex queries
[Industry Recognition and Impact]
The success of this research project and its direct influence on strategic business decisions positions this work as a significant portfolio achievement demonstrating:
User Experience Improvements
Reduced average chatbot discovery time by 40%, from 55.7 seconds to 33 seconds, increasing seamless access to support and decreasing user frustration.
Elevated overall user satisfaction score from 3.1/5 to 4.2/5 through clearer welcome messaging and concise, card-style responses that cut text-heavy scrolling by 60%
Business & Operational Outcomes
Decreased live-agent escalations by 30%, as generative AI handled 70% of routine inquiries end-to-end, freeing agents to focus on complex cases.
Strategic & Competitive Advantages
Strengthened brand differentiation by aligning digital touchpoints with Singapore Airlines’ service excellence promise, driving positive press coverage and reinforcing loyalty among KrisFlyer members.