Conventional spatial network analysis relies on topological metrics that neglect visitor behavioral dimensions, leading to systematic mismatches between structural centrality and actual spatial vitality. This study proposed a revised node efficiency model that incorporated three visitor-centric parameters: space attractiveness, quantified by the type, frequency, and duration of on-site activities; a visitor inflow coefficient reflecting dynamic crowd aggregation; and path attraction impedance, accounting for the flow-retention effects along pedestrian routes. The model was validated using field observation data from 20 public space nodes at Xi’an Banpo International Art District. Comparative analysis against conventional node efficiency and degree centrality metrics revealed substantial discrepancies between topologically critical nodes and those that actually sustain visitor vitality. Two distinct types of supply-demand mismatches were identified, namely network-oriented potential spaces and activity-oriented potential spaces, and differentiated intervention strategies were proposed. These findings demonstrate that incorporating visitor activity characteristics reshapes the assessment of node importance, offering a human-centered diagnostic lens for public space analysis. This approach provides a methodological framework for evidence-based spatial intervention, with empirical validation in the context of industrial heritage renewal, and suggests promising directions for further testing in other urban public space typologies.