Data teams are used in Dutch secondary education to support schools in data use for school improvement. Such teams are likely to be most effective when knowledge is shared between the data team members and brokered throughout the school. Social network structures may play an important role in this. The present longitudinal, mixed-methods study used social network analyses to display how network structures changed over time in eight schools with a data team. Furthermore, the log files and status reports of four teams were studied to explore the relation between the team’s progress and the striking changes in their networks. Results showed that some teams are substantially better at sharing and brokering their knowledge. Several differences in the data teams’ progress, for example in whether the data team members critically reflected on the quality of their data, were found to be related to the changes in their network structures.