In today's fast and ever changing business environments, it is very important to quickly adapt business processes to new business requirements, and to optimize their performance. From this viewpoint, business process performance prediction plays a key role in directing the business. Previous studies have shown that a network queuing model is able to predict the business process performance. In order to do this, the model requires details of the service?s inside behavior.

This paper presents a business process performance model called ?BizCast?, which enables us to estimate business process execution time under any given condition and not to require details of the service?s inside behavior. We especially focus on the overlap between business process instances to predict the performance based on the outside behavior. Prototype evaluation of the service model for supply chain services shows high accuracy, which is, at least, almost 0.9 correlation coefficient.