The problem

When I joined Expedia we ran a smoke test to determine channel preference. The results showed a traveler preference for Chat. The first stage provided the chat option to travelers and an off the shelf solution for agents. In the second stage we created an agent tool to receive chat and phone calls.

As our chat channel matured we built a chat bot to automate common traveler scenarios. The logical next step was to provide the bot automation to our agents. Voyager Next was the project that initially brought the agents a virtual agent.

Mobile channel picker, call back and chat entry UX.

Two varriations of our chat bot experience.

The roles

Initially I was tasked with designing the customer chat UX. As the Voyager Next project began I partnered with Jesse Brown, Dan Saimo, and Chris Arredondo. We worked to flesh out the framework and visual design we called Orion. Once these guiding elements were set in motion I worked with our Product and Development team to bring it all to life.

Early idea for chat plus a pane for trip details.

Orion design with chat and Virtual Agent.

Research

Call centers visits for the chat roll out have provide rich research opportunities. During trips to the Philippines and El Salvador I have been able to gain an understanding of the agents job. Each trip included days of ethnographic research followed by 1:1 usability tests.

Some examples of research driven changes:
An early MVP decisions that the typing indicator wasn't needed was reversed. A desire to be able to help each other prompted exploration into agent to agent chat. Supervisors interviews lead to a tailored search for training chat agents.

The Jeepney is ubiquitous in the Philippines.

A three pane exploration for Agent to Agent Chat

A layered exploration for Agent to Agent Chat

A Covid-19 redesign for laptops as agents began to work from home.