The Microsoft Support Virtual Agent is a chatbot that provides help to Microsoft customers with their products. It covers a wide variety of topics, and it’s accessible both online and through the Get Help app in Windows.
I was the conversational designer and author for this project.
I had to create a cohesive conversation flow (or dialog) that gets users to their solution as quickly as possible with disambiguation and accounts for several different variables.
Users often come to the Virtual Agent upset, and the experience needed to show empathy to the user’s situation without appearing overly emotional or chatty. The overall goal was to solve the user’s issue and get them back on track as quickly as possible.
I used a whiteboard to draw out several iterations of the conversation flow, ultimately choosing this flow based on how it accounts for a vast majority of user issues upfront and helps guide users to their specific resolution as quickly as possible. I wrote and revised the content, including several rounds of feedback from other authors and stakeholders.
Before releasing this content, I conducted several A|B tests to ensure that these changes improved key metrics and made changes accordingly.
In the first month after publishing this content, my conversational experience design helped nearly 75,000 users fix their printer issues, and it accounted for a 20% increase in customer success from the previous content on this topic. This equates to thousands of dollars saved in support costs, and it continues to be a model for other writers on how to incorporate disambiguation and conversation looping to improve the user experience.