1. ¹Hewlett Packard Labs, Palo Alto, CA, U.S.A
2. ²University of Konstanz, Konstanz, Germany

Correspondence: Daniel A. Keim, Computer Science Institute, Universität Konstanz Box D78, Universitätsstr. 10, 78457 Konstanz, Germany. Tel: +49 7531 88 3161; Fax: +49 7531 88 3062; E-mail: keim@inf.uni-konstanz.de

Received 25 February 2005; Revised 29 November 2005; Accepted 3 January 2006; Published online 10 April 2006.

Abstract

Business operations involve many factors and relationships and are modeled as complex business process workflows. The execution of these business processes generates vast volumes of complex data. The operational data are instances of the process flow, taking different paths through the process. The goal is to use the complex information to analyze and improve operations and to optimize the process flow. In this paper, we introduce a new visualization technique, called VisImpact that turns raw operational business data into valuable information. VisImpact reduces data complexity by analyzing operational data and abstracting the most critical factors, called impact factors, which influence business operations. The analysis may identify single nodes of the business flow graph as important factors but it may also determine aggregations of nodes to be important. Moreover, the analysis may find that single nodes have certain data values associated with them which have an influence on some business metrics or resource usage parameters. The impact factors are presented as nodes in a symmetric circular graph, providing insight into core business operations and relationships. A cause–effect mechanism is built in to determine 'good' and 'bad' operational behavior and to take action accordingly. We have applied VisImpact to real-world applications, fraud analysis and service contract analysis, to show the power of VisImpact for finding relationships among the most important impact factors and for immediate identification of anomalies. The VisImpact system provides a highly interactive interface including drilldown capabilities down to transaction levels to allow multilevel views of business dynamics.