Purpose Service oriented organizations have difficulties finding evidence based office design suited for new ways of working. This study provides CRE managers with information on factors that influence the activities at workplaces in an activity based office environment, and can be used for further development of office design. Design/methodology/approach Based on literature research a conceptual model is developed that shows the relations between the workplace characteristics, physical environment, organization characteristics, user characteristics, and the use of activity based workplaces. Hypotheses are proposed based on the model. Data is collected in 3 office environments in Belgium and the Netherlands, providing 90.890 observations. Specifically, a Bayesian belief network (BN) was used to derive and represent the causal relationships between all variables included. A major advantage of a BN is that the network structure takes direct and indirect relationships between the variables into account. The first network addresses the relationships between use of workplaces and CRE/organizational characteristics, without looking at user characteristics. The second network contains only data of occupied workplaces to study the influence of user characteristics. Both networks were used to test the hypothesis and predict the use of workplaces under (future) conditions. Findings According to the activity based philosophy, an employee performs an activity on the workplace that has the most suitable functional characteristics. Both networks, however, show more complex relations between the variables and expose relationships between environmental characteristics, workplace functionality and activity. The second network confirms that user characteristics influence use of the workplace as well. It can be concluded that a CRE manager should develop at least two types of workspace to support all activities in a service oriented organization.