Cognitive models of social anxiety disorder (SAD) provide theoretical frameworks that explain maintaining factors and guide treatment protocols. While cognitive behaviour therapy (CBT) is efficacious for SAD, mechanisms of change remain under-investigated. This study explored the synchrony of change among maintaining factors of SAD and investigated the timing and sequence of change. We analysed sessional data from 551 clients undertaking CBT for SAD in London (UK) primary care services. Clients typically received 14 sessions ( SD = 3.3, maximum = 22), had an average age of 27.8 ( SD = 6.4), and 57.4% was female, reflecting the clinical SAD population. We modelled velocity (rate of symptom improvement) and acceleration/deceleration (changes in rate of improvement) using Dynamic Exploratory Graph Analysis, a network-based analysis. This approach revealed synchronised change patterns, with peak change sequence plotting identifying critical periods of improvement. Results demonstrated synchronised change between self-referential processes and avoidance, forming a change component. A second change component showing internal synchronisation comprised worrying, somatic anxiety, depressive mood, and functional impairment. Limited synchrony was observed between these two components, suggesting distinct change processes. Peak change analysis indicated three phases: initial improvements (sessions 1–4); improvements in self-referential processes and avoidance (sessions 5–10), coinciding with behavioural experiments; and consolidation of change (sessions 10+) in second component factors. This study provides empirical validation of theoretical SAD models and illuminates temporal change dynamics. Understanding treatment trajectories of these change components helps clinicians select, sequence, and pace interventions so both components are effectively targeted.