Automated interactive sales processes

When we look at successful sales processes occurring in practice, we find they combine two techniques which have been studied separately in the literature. Recommender systems are used to suggest additional products or accessories to include in the bundle under consideration, and multi-issue negotiation focuses on optimizing the precise configuration of the bundle and its price. In this paper, we pursue the automation of such interactive sales processes. We present some key insights about, as well as a procedure for locating mutually beneficial alternatives to the bundle currently under negotiation. The essence of our approach lies in combining aggregate (anonymous) knowledge of customer preferences, learnt by the shop in interactions with previous customers, with current data about the ongoing negotiation process with the current customer. We present a memory- and a model-based method for online learning customer preferences and discuss their pros and cons. The performance of our system is illustrated using extensive computer experiments involving simulated customers with highly non-linear preferences which the system has no trouble learning.