This paper designs an optimal mechanism to correct coordination failure. A planner wants her agents to coordinate on a cooperative action. Agents gather noisy private information regarding the underlying fundamental and decide whether to cooperate or not. The global game literature uniquely identifies the chance of coordination failure when the coordination risk is concentrated at one point in time. We analyze the case when the planner diffuses the coordination risk over time. The planner approaches the agents sequentially -only a proportion of agents at a time and advancing further only when the coordination failure has been averted so far. The public information of survival works as a coordination device and helps in mitigating the coordination risk. We show that if the planner can diffuse the coordination risk enough, then she can achieve the first best as the unique equilibrium outcome. However, if the planner has only limited power to diffuse the coordination risk, multiple equilibria can arise. A maxmin planner should diffuse the coordination risk as much as possible. We also show that if some groups are more reluctant to cooperate than others, a max-min planner should approach the more reluctant groups first. Our mechanism is robust to various generalizations and can be applied to a wide range of coordination games. * We are extremely grateful to Douglas Gale and David Pearce for their substantial guidance and support throughout this project. We thank Viral Acharya, Jess Benhabib, Itay Goldstein, Zhiguo He, Laurent Mathevet, Cecilia Parlatore, Ennio Stacchetti, Laura Veldkamp, Neil Wallace for helpful comments and suggestions.