TY - JOUR
T1 - Transfer rate prediction at self-service customer support platforms in insurance contact centers
AU - Andrade, Rodrigo
AU - Moazeni, Somayeh
N1 - Publisher Copyright:
© 2022
PY - 2023/2
Y1 - 2023/2
N2 - Customer support constitutes the face of a brand and plays a critical role in shaping customers retention and loyalty. To facilitate customer contacts and to reduce the corresponding operating costs, organizations resort to self-service interaction channels. However, dealing with self-service customer support platforms has been identified to be one of the most frustrating aspects of a poor customer contact experience. This paper builds on real data from contact centers of a major U.S. insurance company, covering about 10 million calls, to study transfer rate at the interactive voice-response platform. We develop an empirical model to analyze features that exert significant impact on the likelihood that a call arrived to the platform is eventually transferred to a live agent. We find evidence that caller types, specific caller intents, trying distinct channels, and payment attempts can help predicting the transfer outcome. Our study confirms the importance of a caller's location attribute and evinces that the size and composition of the set of effective features and their directions vary substantially across different states. Our results provide managers with several actionable insights to enhance the customer contact experience. For example, our feature selection analysis enables designing personalized interactive voice-response systems based on customer-contact features. Identifying those calls expected to be transferred at any time can guide recruiting and scheduling a workforce with appropriate skill sets. In addition, the transfer rate is a key input variable for various decision making problems arising in the management of call center operations.
AB - Customer support constitutes the face of a brand and plays a critical role in shaping customers retention and loyalty. To facilitate customer contacts and to reduce the corresponding operating costs, organizations resort to self-service interaction channels. However, dealing with self-service customer support platforms has been identified to be one of the most frustrating aspects of a poor customer contact experience. This paper builds on real data from contact centers of a major U.S. insurance company, covering about 10 million calls, to study transfer rate at the interactive voice-response platform. We develop an empirical model to analyze features that exert significant impact on the likelihood that a call arrived to the platform is eventually transferred to a live agent. We find evidence that caller types, specific caller intents, trying distinct channels, and payment attempts can help predicting the transfer outcome. Our study confirms the importance of a caller's location attribute and evinces that the size and composition of the set of effective features and their directions vary substantially across different states. Our results provide managers with several actionable insights to enhance the customer contact experience. For example, our feature selection analysis enables designing personalized interactive voice-response systems based on customer-contact features. Identifying those calls expected to be transferred at any time can guide recruiting and scheduling a workforce with appropriate skill sets. In addition, the transfer rate is a key input variable for various decision making problems arising in the management of call center operations.
KW - Big data
KW - Business analytics
KW - Call transfer rate
KW - Contact center
KW - Feature selection
KW - Machine learning
KW - Self-service customer support platform
UR - http://www.scopus.com/inward/record.url?scp=85138020085&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85138020085&partnerID=8YFLogxK
U2 - 10.1016/j.eswa.2022.118701
DO - 10.1016/j.eswa.2022.118701
M3 - Article
AN - SCOPUS:85138020085
SN - 0957-4174
VL - 212
JO - Expert Systems with Applications
JF - Expert Systems with Applications
M1 - 118701
ER -