Abstract Understanding convective aggregation is very important for understanding tropical climate and climate sensitivity. However, we still lack a full understanding of how aggregation evolves in the real world or what phenomena and scales are analogous to the self-aggregation observed in idealized models. In this study, we apply the moist static energy (MSE) variance budget framework to ERA5 reanalysis data to study the evolution of large-scale aggregation over tropical oceans at basinwide scales. Our novel phase space diagnostics focus on the variability of observed aggregation compared to most previous self-aggregation studies, which focus more on the aggregated mean state. We visualize observed aggregation to evolve anomalously around a mean state in a cyclical fashion forming aggregation–disaggregation cycles. We find horizontal advection of MSE to play the primary role in determining when the domain aggregates or disaggregates. In contrast, all advective, radiative, and surface flux feedbacks are found important for determining the magnitude of the aggregation anomalies. Surface fluxes and horizontal advection tend to dampen aggregation anomalies, while radiative fluxes and vertical advection tend to amplify aggregation anomalies. Looking deeper into the advection terms, we find that changes in vertical advection are dominated by enhanced low-level subsidence over the dry regions during the more aggregated states. This creates an anomalous drying tendency over the dry regions, which maintains aggregation anomalies. In contrast, horizontal advection changes are found to be dominated by increased moisture advection out of the moist columns with stronger aggregation. Significance Statement The purpose of this study is to characterize and understand the evolution of large-scale convective aggregation in the real world through reanalysis data. While most previous observational studies have focused on the evolution of clouds and cloud populations with aggregation, we focus on the energetics and the impact of aggregation on redistributing moisture throughout the domain. Our framework highlights that aggregation can be visualized as a continuously occurring cyclic feature at large scales in the tropics. Further, our work provides a deeper insight into the changes in large-scale circulation that accompany aggregation and characterizes the similarities and differences between the different regions in the tropics.