The paper studies a variant of Granovetter’s threshold model of conforming behavior. In this model, a network of agents with binary states (inactive/active) acts in discrete time under the influence of instigators (always active agents). Using special computational method based on algorithms for solving Boolean satisfiability problem, we can find dispositions of small number of instigators among initially inactive agents, which after several time moments force the majority of agents to the active state. Using a large number of networks, both randomly generated, and fragments of real world social networks, we search for the interconnection between static and dynamic characteristics of network vertices: whether or not static centrality measures influence the chances of the vertex to be picked as an instigator.