Poster
Client Context
The team’s senior design project cooperates with Georgia Department of Human Services (DHS). Headquartered in downtown Atlanta, the DHS is a government agency that provides a variety of human services to encourage safety and well-being for all eligible Georgia residents: approximately 2 million people. Currently, the DHS has 3 divisions: Child Support Services, Family & Children Services, and Aging Services. The call centers for the divisions answer a variety of questions for Georgia residents regarding the impacts of medical benefits, social programs, and effects of new initiative changes.
In July 2023, DHS expects Congress to suspend continuous Medicaid healthcare benefits for millions of Georgians that went into effect during the COVID-19 pandemic. DHS is anticipating high call volume as customers adjust and react to the policy changes. Our client is seeking a solution that will minimize hold times, minimize abandoned calls, and decrease the amount of time with which their team is able to route a customer to the correct department and serve them, overall decreasing the amount of congestion in the system.
For the system, the DHS call center utilizes an interactive voice response (IVR) phone tree, consisting of nodes that represent various DHS question menus and branches that lead to a specific call type. There are currently seven major branches available to customers, including DFCS OFI, child support, aging services, and child welfare, among others. To manage call volume, the call center employs a 100% callback system that places customers in a queue for a callback from a representative rather than connecting them directly. The customers are served by dedicated agents or agile agents; dedicated agents serve only one call type and agile agents serve multiple. If a customer calls in during a period of high volume, they may receive a courtesy disconnect, which occurs when the call center reaches maximum capacity or the customer's issue has been resolved. Customers may only receive a callback once per day per phone number; if exceeded, they will receive a courtesy disconnect. The call center continually reviews and adjusts its rejection processes based on factors such as average speed of answer, average handling time, and call back list, with the aim of providing efficient and effective customer service.
Project Objective
In September 2022, a new initiative regarding cash assistance was implemented to assist those below a certain earning. At that time, there was a spike in the courtesy disconnect rate. The DHS is planning to have another initiative change regarding Medicaid in July 2023. During the pandemic, many residents had locked in benefits regarding Medicaid, however, the pandemic era benefit program is expiring, and the DHS anticipates that there will once again be a spike in calls from residents asking if they will lose their benefits. With this, the goal is to reduce the courtesy disconnect rate, especially after a new initiative change to serve more Georgia residents.
With call volume and courtesy disconnects expected to increase significantly in July, the DHS needed to create a plan to serve more people. To achieve this, the team analyzed data from January 2022 to March 2023 and identified three opportunities for improvement. The first opportunity involves prioritizing call types according to the agent type best suited to handling them. As the system has 18 different call types with both agile and dedicated agents, the team recommends giving priority to the dedicated agents serving the same call type as an agile agent. The second opportunity involves switching agents from low utilization call types to high utilization call types. Specifically, we calculated the arrival rate for each call type in 2022 and compared it with the calculator’s result to determine the difference between current allocation and optimized allocation of agents. Finally, the team proposes relaxing the threshold to make more agents available for serving additional customers. The team believed that finding the best service rate factor for the threshold would allow for more calls to be answered. The team's objective was to create valid simulation models and test these different aspects
Design Strategy
To understand the complex interactions of call types, agile workers, and dynamic threshold calculations, the team constructed a model of the system in Simio. Then, by perturbing the call volume, the team tested how the system may react in July due to the expected increase in call volume and what steps would be required to minimize disconnects. To validate the simulation model and ensure it's accuracy, we compared the simulation output against the actual disconnect rates from January 2023 to March 2023. The result was a 1% difference between the real-world disconnect rates and the simulated disconnect rates. With the model validated, we could then test each of the different opportunities: prioritization, threshold relaxation, and reallocation of workers. The simulation allowed us to view the impacts of changes on the existing system, but the team wanted DHS to be able to make dynamic changes to their workforce to ensure efficient and effective planning before initiatives are enacted. The team developed an excel calculator to provide the DHS with dynamic workforce planning that pulls from one of the existing autogenerated reports; as this report is updated with new call types or new agent types, the calculator will work to provide the DHS with specific numbers of agents needed to works towards a steady state while keeping an average wait time of 30 minutes.
Deliverables
To utilize the excel deliverable, users simply need to input a report, which is automatically generated by the DHS system. The calculator will then analyze the data and generate an informative report that can aid DHS in understanding the factors such as the number of workers in each queue and the expected wait time for customer callbacks. Leveraging coded M/G/c logic, the Excel calculator subsequently provides recommendations on the required number of agents for each type and calculated wait times. It's important to note that the calculator does not account for the minimum 5% CD rate resulting from threshold calculations. With that information, DHS can have a better understanding of the system.
First, the team tested prioritizing customers based on their call types by dedicated agents serving them first, followed by agile agents who can serve customers that can also be served by dedicated agents. This approach can help reduce wait times and enhance customer satisfaction. By changing priority, the courtesy disconnects rate became 19.79% (originally 19.92%) which decreased by 0.13% which was determined as insignificant and would not cause much difference.
Another strategy the team proposed is to relax the threshold, given that the minimum number of agents needed is much lower than the actual number of agents, which the team assumed that they have enough agents to serve more people. To test this idea, the team relaxed the threshold by decreasing the service rate factor to 0.8, 0.6, 0.4, and 0.2. This strategy had an inversely proportional relationship between wait time and courtesy disconnects. Overall, the team suggests that the DHS slowly makes changes to the rejection logic to determine what is best.
Finally, the team explored the benefits of moving workers between groups to reduce courtesy disconnect rate. Specifically, the team moved agents from low-utilized groups to high-utilized groups and transferred "higher-level" workers to "lower-level" groups, which did not require additional training since “higher-level” workers are usually agile workers which can serve multiple call types.
The team further recommends improving the call center by having the DHS collect and record number of agents in each agent type, instead of recording the number of agents in each call type. By doing so, the DHS can gain a better understanding of how to rearrange their agents or hire more workers based on the results provided by our calculator. Additionally, the team recommends that DHS use our agent calculator to conduct more workforce planning to increase their efficiency and develop a training/hiring plan to overcome future challenges. The DHS will be able to react accordingly when the teams' initiatives are implemented and reach their goal of decreasing courtesy disconnects, keeping their promise and commitment to the people of Georgia.
Value and Impact
Because the other two strategies weren't as impactful for a single day, the team implemented the improvement by switching agents and hiring more workers. Instead of focusing on 100% of people that are affected will call in, the team used 50% increase in MA since it is a better representation of reality. If 50% of people affected call in, the courtesy disconnect rate increases to 50.46% without any adjustment to the current system. After we tested the improvement of reallocating agents without hiring more agents, the courtesy disconnect rate decreased to 36.33% which decreased by 14.13%. This decrease in CD rates can allow the DHS to answer an additional 19,005 calls and serve more residents. By hiring 15 and 30 more workers, the courtesy disconnect rate decreased to 31.43% and 24.85%, respectively.
Project Information
Student Team
Daniel Anderson
Henry Wallace
Bochun Guo
Aashvi Patel
Shu Zhu
Zhiyi Li
Luqi Xu
Zhonghui Shen
Faculty Advisor
Faculty Evaluator
