Facebook posts compete for human attention in a zero-sum game. The effectiveness of one post in grabbing the population’s attention comes at the cost of another, due to humans’ finite attention. This makes it challenging for government organizations to engage with their citizens through this bustling medium. In a large-scale longitudinal study we investigate what makes Facebook posts popular (seen by many) and effective (commented, liked, or shared by many) in a non-profit context: the official Facebook page of a mid-sized city. We model the competition dynamics that shape the fate of Facebook posts using Structural Equation Modelling – independent of their semantics. Our analysis reveals that for this community audience demographics, the timing of posts, and the media type of post are significant factors, and argue that our method can be applied to other similar pages to determine which factors can lead to improved government-citizen communication. Finally, we argue towards more actionable research to be conducted in the context of social media use in e-government.