It is a well-known fact in the innovation management literature that social network structures both constrain and facilitate ideation activities. Still, recent research shows how the impact of social networks varies across different contexts, e.g. depending on the complexity of knowledge involved or the outcome of the innovation. Although, at initial stages of the ideation process both the type of knowledge required and the final outcome is often unknown or at least unclear, while in contrast the problem type involved is known in advance. Based on an unique dataset of more than 30.000 ideas, submitted to an online platform in response to pre-defined problems, we introduce a distinction between broad versus narrow problems. We claim that different kind of cognitive problem structures will attract different kinds of resources and different types of networks will form around different problems. Multiple levels of analysis are in play, since we rely data on: 1) Different kinds of problems, 2) ideas and comments to these ideas, 3) the characteristics of individuals involved in the given ideation process and 4) the social network activities of these individuals. The project is still ongoing, but preliminary data analysis does tentatively indicate how narrow problem structures indeed are associated with different network structures compared to broad problems.