In this paper, an approach for the site-level monitoring of building energy consumption is introduced based on properties of the Koopman operator: the Koopman operator is an infinite-dimensional, linear operator that captures nonlinear, finite-dimensional dynamics. Using properties of the operator, site-level building meter data can be decomposed into spatial-temporal modes. Through use of these modes, characteristics of building utility consumption can be studied at varying time-scales. Compared to conventional analysis, often performed one meter at a time, the proposed approach allows a facility manager to quickly determine the overall behavior of multiple buildings, and whether they are operating efficiently and according to programmed schedules. An implementation of this approach is demonstrated as it has been applied in the actual management of buildings at a university campus using measured sub-hourly meter data which has been collected over a one year time span. Spatial visualization of results allows data to analyzed in a condensed form compared to conventional methods.