Chatbot
Challenge
Using our existing My Hanover Policy (MHP) product, the product team wanted to include a Chatbot tool to help meet a goal of enabling customer self-service set by our senior leadership team. The expectation for this tool is to use Gen AI to create a conversational experience for our customers.
Process
As part of a cross-functional team including product management and engineering, I represented the UX team as the lead UX designer on this initiative.
I worked with the product team on surveys to users to understand their interest or expectations from a chatbot. I worked with our UX Analyst to gather information on existing behaviors from our users to understand the typical questions funneled by our customer service teams while also evaluating the areas of our product that users frequently visited.
I also examined similar types of products being used by competitors or adjacent industries to understand what’s working or not with those experiences and make note of any trends. As we progressed through different phases of design, I created tests in Usertesting.com to validate design, functionality, and behaviors.
Deliverables
Using Figma, I created an initial user flow for the experience with sketches of the potential look and feel. After feedback from stakeholders on this initial work, I progressed to mockups in Figma and creating reusable components. With tweaks based on feedback from stakeholders, I created a prototype and tests in Usertesting.com to validate the design and functionality. For dev handoff, I provided them with the CSS and specs and Figma files. I reviewed the Chatbot in staging & production to ensure alignment with the designs.
Outcomes
The chatbot was rolled out as a soft launch to gauge reception from customers. Initial feedback has been positive where users can quickly find information with a decrease in customer drop off and an increase in engagement within MHP. We have noticed that users expectations for the Chatbot has been to expect it to respond as if it were Chat GPT. This validates our approach for Phase 2 where we expect to integrate with an LLM.
