Traditional network centrality indicators answer the question "What characterizes a highly important node." The answers, however, do not generalize. They are rarely usefully accurate for the 99% of nodes which are not highly central. Even for highly central nodes, their relevance is dependent on network topology and sampling. We address a different question: "What characterizes node influence" and produce a metric which is accurate for every node in the network. Influence is defined from an epidemiological perspective as the expectation of force of infection produced by the node. The resultant value is strongly predictive of many epidemic outcomes over a wide range of network topologies, simulated and real. It also shows high rank correlation to walk-based centralities, indicating it captures their definitions of importance. The measure naturally extends to weighted and directed networks. A live demo lets you explore node influence and epidemic spread on the world airline network.