The simulation of mobile networks requires mobility models that reproduce the movement of nodes in a realistic way. Although many such models exist, they are not well suited for model office scenarios in which movements are mostly caused by people meeting in known locations to discuss some issues. We present first steps towards a meeting model for office environments which is based on the real world movement data gathered after performing a one-week sensor network deployment in our department. We describe details about the processing of the acquired data, the construction, and the execution of the movement model. Finally, we discuss some of the lessons learned throughout the experiment.