Experts argue that adoption of healthy sanitation practices such as hand washing and latrine use requires focusing on the whole community rather than on individual behaviors. According to this view, one limiting factor for ending open defecation lies in the capacity of the community for collective action: each member of a community bears the private cost of contributing by washing hands and using latrines but benefits through better health outcomes depend on whether other community members also opt out from open defecation. We rely on a community-based intervention carried out in Mali as an illustrative example (Community Led Total Sanitation or CLTS). Using a series of experiments conducted in 121 villages and designed to measure the willingness of community members to contribute to a local public good, we investigate the process of participation in a collective action problem setting. Our focus is on two types of activities: (1) gathering of community members to encourage public discussion of the collective action problem and (2) facilitating the adoption of individual actions to attain the socially-preferred outcome. When the facilitator starts by introducing a topic and a group discussion follows, can the facilitator further improve outcomes? Will a group discussion that follows facilitation improve, reduce, or have no effect on collective action? We find evidence that cheap talk raises public good provision and that facilitation by a community member does not improve upon open discussion.