Networks are sets of nodes and edges (graphs) that model complex systems throughout science and engineering. To study networks, we usually design our algorithms and computations around nodes and edges, but it is really higher-order connectivity patterns captured by small subgraphs, or network motifs, that are the fundamental structures controlling and mediating the behavior of many complex systems. In this talk, I will generalize classical ideas in spectral graph clustering to account for higher-order structure in networks. This leads to better empirical results on several real-world datasets and also exhibits better numerical properties on some model problems. I will include applications from ecology, biology, transportation, and social networks.