Social dominance hierarchies are embedded within multivariate and multiscale behavioural processes, where patterns of individual interactions form the basis of social bonds and give rise to higher-order emergent social properties. While the links between current dominance rank and social network position have been broadly established, less is known about the social precursors and consequences of rank reversals or the scale at which these processes are quantifiable. Here, we considered how longitudinal hierarchy dynamics relate to an individual’s unilayer and multilayer centrality in a large group of captive rhesus macaques. We considered two interrelated ideas: first, which social variables predict individuals’ subsequent ascension or descension in the dominance hierarchy; second, the influence of individual rank changes on their subsequent social centrality. Prior to ascending the hierarchy, we found that individuals were less socially central. This decreased centrality was indicative through lower unilayer subordination-signalling instrength, contact sitting strength, subordination-signalling authority scores, aggression authority scores, multilayer PageRank versatility (centrality across and within layers), and multilayer consensus rank. Positive rank change, however, was a poor predictor of subsequent network centrality. Prior to descending the hierarchy, individuals were characterized by lower grooming degree and subordination-signalling authority scores. Negative rank change was, however, a strong predictor of subsequent social position: individuals had lower grooming degree, lower aggression authority scores, and lower multilayer PageRank versatility. Overall, these findings suggest that rank change, regardless of direction, has social precursors and consequences in temporally adjacent 90-day periods. These dynamics can be captured at the local, individual scale (unilayer) and as emergent properties (authority scores, multilayer networks). Our findings highlight the individual and combined value of unilayer and multilayer approaches to better understand the dynamics of social structure.