We study a multiple-receiver Bayesian persuasion model in which the sender wants to implement a proposal and commits to a communication strategy which sends correlated messages to multiple receivers who have homogeneous beliefs and vote sincerely. Receivers are connected in a network and can perfectly observe their direct neighbors' messages. After updating their beliefs accordingly, receivers vote for or against the proposal. We characterize optimal communication on various network structures and find that the limited information spillovers in the model often do not prevent the sender from attaining maximum gain from persuasion. Our results highlight the importance of the network structure when designing optimal strategies, as voters are not necessarily better off with strictly more information. Surprisingly, the creation of new links may even benefit the sender.