Time | 4 weeks
Role | Design + Research
Team | PM, Engineer, Researcher, Content Strategist
Transforming Expedia's Virtual Agent for Refund Eligibility Checks
Background
The onset of Covid-19 triggered widespread travel cancellations due to lockdowns and restrictions, leading to an influx of customers seeking refunds rather than credits for their canceled trips. Expedia witnessed an escalating number of customer service inquiries concerning refund eligibility. To alleviate call volumes and empower customers, Expedia's customer experience director sought the integration of a refund eligibility feature within the Virtual Agent chatbot.
Problem
Customers seeking refunds instead of flight credits inundated Expedia's customer support channels. This surge necessitated a self-service solution for customers to ascertain their refund eligibility without relying on live customer service agents.
Above: Refund Eligibility Conversation Flow with Back-end Processes
Target Audience
The design focus was on empowering customers seeking refunds through the Virtual Agent chatbot, enabling them to independently check their eligibility.
Roles and Collaboration
In my role as lead designer, I collaborated closely with a multidisciplinary, cross-functional team comprising a content strategist, a researcher, and an engineer. My responsibility included fostering collaborative efforts among team members to craft and test this new user experience.
Scope and Challenges
Designing for Expedia's Virtual Agent involved intricate dependencies across various user flows within the chatbot. Coordinating the integration of a new skill required alignment with multiple design teams managing different sections of the user journey. I ensured seamless integration of the refund eligibility skill within existing and forthcoming projects.
Above: Wire-Frames for Skill’s Flow Created with Figma
Process
Initiating discussions with the Product Manager, I outlined the requirements and constraints for integrating the refund eligibility skill into the chatbot's functionalities. Collaborating with the engineer, I mapped out the conversation flows and talked through backend processes that would need to be implemented.
Creation and Testing
Based on the conversational flows, I developed high-fidelity wireframes and collaborated with a content strategist to devise diverse chat text versions for refund scenarios. Prototypes for each eligibility scenario were built using Figma. With the researcher and content strategist, we refined the prototypes and devised a comprehensive test plan focusing on discoverability, content preferences, and overall usability.
Above: Video of Working Prototype for In-Eligibility Scenario
Usability Testing and Recommendations
A series of unmoderated usability tests and preference tests were conducted using Usertesting.com. The results were meticulously reviewed with the researcher, leading to the creation of a detailed slide deck summarizing test outcomes and recommendations for the way forward.
Above: Selection of Slides from Testing Results and Recommendations Presentation
Outcome and Implementation
The findings and design recommendations were presented to the PM, engineering, and customer service leadership teams. One critical insight revealed the necessity to integrate the skill into an earlier section of the user flow. Accordingly, we recommended another design team to incorporate our work into their skill, appearing earlier in the conversation flow. The agreement was reached, and I handed off user flows, design assets, content, and recommendations to the designated team for seamless integration.