Systems and methods for clustering messages are provided. In some aspects, a method includes receiving a set of communications. The set of communications includes a set of messages. Each communication in at least a portion of communications in the set of communications is associated with a set of contacts. The method also includes determining a first contact list based on the set of communications. The method also includes determining a first subset of the set of messages, each message in the first subset of the set of messages being associated with a set of contacts including at least a first threshold number of contacts in the first contact list. The method also includes providing an indication of the first subset of the set of messages.