Real-time customer probing and decision support in a call center

Abstract

In customer-care contact centers, agents are sometimes tasked with selling products to customers. The state-of-the-art guided selling solutions augment available historical data about customers with real-time information collected during the call that is fed into the real-time product recommendation models to generate offers tailored to the customer's needs and preferences. However, the technology does not address the question of how to scientifically structure and design this real-time information gathering activity. We propose a scientific approach to question (probe) customers via a sequence of questions optimally selected for individual customers. We describe the design of a complete product recommendation tool that makes use of our customer probing approach and is deployed in the call center of a global bank to assist agents with cross-selling and upselling. Carefully controlled experiments show our solution helping agents achieve close to a 20% improvement in sales revenues and reducing the performance gap between novice and experienced agents.