I designed an internal chatbot that supports Call Center Employees. It helps them quickly find answers to Customers' questions related to how the Benefits were and should be processed.
NOTE
Due to the sensitive nature of working with Protected Health Information (PHI). The work represented has been altered to avoid any violations.
Objective
Our goal was to reduce the time it took for a Member to have a question answered. Before, Call Center Employees used multiple apps to find relevant information depending on the task. We aimed to have one place to go that pulled from the source of truth. Ensuring the information was accurate and quick to get to.
This POC was to see how we could integrate the results of an Algorithm into the chatbot. So Call Center Employees can understand key indicators related to the customer without interrupting their current process.
Project Contributions
Created User Flows, Decision trees, Wireframes, and mockups for the project.
Users
Our primary Users were Call center Employees. They have a lot of time constraints in their job. That requires them to be efficient and clear in their communication. For example, they are not allowed to have more than 2 minutes of dead air on a call without checking in. All the while they could be looking at how a claim was processed, what a Member should be paying a Provider, Reporting an issue with how a Claim was processed, etc.
During discovery, I found three types of users would be impacted the most by this project.
Process
Kickoff Meeting
We met with the team in charge of the algorithm. To understand how it worked and what the results from the algorithm meant. There were two results we were expecting. One predicted how likely the Member was to disenroll from Florida Blue The other predicted how likely the Member was to go to the hospital in the next 90 days. Both are key indicators for the company. As the goal was to keep Members happy and healthy respectively. After having an understanding of the algorithm, I wanted to understand how our Call Center Employees handle sensitive situations.
A simplification of the output from the algorythm.
Business Expert Interviews
Sat down with some hand-picked experts from the teams that would be using the application. They usually try to understand why a Member is having an issue. So they can try to fix it themselves. If they can fix it. A ticket is created and researched by an appropriate team. If a Member is suspected to be not well. There is a team that reaches out and checks on the Member. In both situations, the communication is friendly and empathetic, never aggressive or pushy.
The user flow before the POC integration.
Breaking down a vertical Slice
To deliver a prototype quickly. I worked with a System Architect to focus on something we could deliver quickly. We decided to focus on delivering an example of how the system would handle a scenario where a Member was suspected of leaving Florida Blue soon. By doing this we wouldn’t boil the Ocean being stretched too thin to deliver anything of substance promptly. After all, we wanted to show what we could do. So we could get funding for a bigger project.
Building a user strategy
First, we wanted to understand if the Member had a problem. What was the problem, and how could it be fixed? Needing to gather information, we proposed a questionnaire that would quantify how the members felt about the company. Given the Chatbot is a research tool.
The process for any task is more or less a game of 20 questions. Starting with the basics, like is this a Benefit question, Claim question, or Authorization question. Drilling down until the Employee has the information they need. We wanted to put a message at the end of every loop or if the User had not taken an action for more than a minute. Giving enough time for the User to convey the needed information, popping up as a gentle reminder to the Employee. Lastly, we identified key information that Customers call in for that indicates they might be leaving. From the feedback we received from the Expert Interview and decided to integrate the questionnaire into the flow itself if the algorithm predicted the Customer was going to leave.
We also wanted to show the results directly to the employees so they could know how to handle their messaging on the call. As mentioned above, Employees are supposed to be friendly and empathic in these situations where normally they would be more neutral.
Points in the user flow to add elements to the UI.
Choosing our point of attack
I started looking at the existing user flows and tasks to identify key points. Where we could integrate our solution. Using our solution, we will update and add to the user flows to show how an Employee would go from step A to B. With the key points identified, I started to iron out what would be needed from a UI perspective. I borrowed questions from an existing Customer Metric system that had already been established. This would make reading the data easier as Management could use the system without needing to have new training. Ultimately, making it easier to integrate back into the algorithm itself. So it can learn from the feedback on the calls to validate its predictions.
Showing where this could go
With the foundation set, I started to create wireframes for the content. From the Wireframes, I built a mockup of the end-state of the application. From this, we walked our Business experts through the flows. Gathering feedback on the solution. Trying to understand if we had any gaps. After this, the dev team created a prototype to validate the doability of the solution. Once complete upper management reviewed the solution and we received the green light to test our solution live.
POC mockup click through.
Outcome
The work was broken into parts. Starting with displaying the statuses at the top of the UI. Then going from there into the other solutions.
Summary
This project taught me how to look at a process holistically. By creating a couple of points in the process that could have the most impact. Then building a solution quickly. Finally testing it to make sure our hypotheses were correct or not.
Related Case Study
Benefit Narrative, search
A walk-through of how I discovered and solved a communication gap between multiple departments internally.
About
From Tigers to Jaguars, I am more than a UX/UI designer. Discover the entire picture of who I am.