Purpose - In present day society, focused on innovation, knowledge sharing (KS) is essential. CREM needs to provide accommodation designed for people to meet more often and share both tacit and explicit knowledge. Most workplace descriptions do not provide quantitative information on how the design actually stimulates KS. They cannot be implemented straight into a design nor convince general management in budget discussions. This paper tests the suitability of spatial network methodologies to provide this proof.Design/methodology/approach - After developing a conceptual model from literature, the model is tested with a case-study of one large research driven organisation in the Netherlands. For each possible dyad between 138 employees (= 9453 dyads), several KS indicators and workplace aspects are studied with statistical analyses.Findings - KS is clearly related to the allocation of people to the rooms and workplaces in a research building. Bumping into each other does not appear the reason for this relation. Up to distances of 22 meters dyads share average or higher amounts of knowledge. It is ambiguous whether the spatial network analysis methodology is relevant for measuring added value of the workplace for KS. Originality/value - This paper provides empirical evidence on a 'softer' added value of the workplace, which is scarce. Previous studies relied mainly on qualitative descriptions of the workplace, while this paper tests a methodology to quantify the workplace in measurable aspects and correlate them statistically to organisational outcome measures.