Smarter customer service with machine learning
How do service providers decide which customers to serve first when information is incomplete?
In this video, Yueyang Zhong, Assistant Professor of Management Science and Operations at London Business School, explains her research on scheduling in environments like contact centers. Customers may signal a reason for calling, but managers rarely know in advance how urgent, complex, or costly each call might be. Even with perfect information, prioritizing is computationally hard—let alone when much of the information is missing.
Yueyang introduces Learn-Then-Schedule (LTS), a framework that uses random sampling to estimate query complexity, caller patience, and customer value. These insights are translated into importance scores that guide smarter, adaptive scheduling. By balancing exploration and exploitation, LTS reduces call abandonment, improves customer satisfaction, and enhances efficiency.
How do service providers decide which customers to serve first when information is incomplete?
In this video, Yueyang Zhong, Assistant Professor of Management Science and Operations at London Business School, explains her research on scheduling in environments like contact centers. Customers may signal a reason for calling, but managers rarely know in advance how urgent, complex, or costly each call might be. Even with perfect information, prioritizing is computationally hard—let alone when much of the information is missing.
Yueyang introduces Learn-Then-Schedule (LTS), a framework that uses random sampling to estimate query complexity, caller patience, and customer value. These insights are translated into importance scores that guide smarter, adaptive scheduling. By balancing exploration and exploitation, LTS reduces call abandonment, improves customer satisfaction, and enhances efficiency.