Cognitive Behavioral Therapy for Insomnia (CBT-I) is a well-established treatment for chronic insomnia, but its effectiveness varies widely, particularly among shift workers who face irregular sleep-wake schedules and elevated stress levels. In this single-arm open trial, we investigated heterogeneous response patterns to CBT-I and their predictors in a cohort of 50 shift-working firefighters (mean age 34.8 ± 7.7 years; 88% male) who completed a five-week intervention. The primary outcomes were the Insomnia Severity Index and sleep diary–based measures, including sleep onset latency, total sleep time, and self-reported sleep quality. Sleep quality was assessed longitudinally, and growth mixture modeling was used to identify latent trajectories of treatment response. Three distinct response patterns were identified: steady-poor (24%), steady-moderate (50%), and improving (26%). This classification, based on subjective reports, was further examined using a mathematical model of alertness that is widely used in sleep and circadian research. The model was applied to wearable data to derive physiological response patterns. Multinomial logistic regression revealed that poor response was associated with engagement in emergency duty, longer sleep onset latency, and higher insomnia severity, while improved response was associated with shorter shift work duration, lower alcohol intake, and a 3-day shift cycle. These findings highlight that sleep quality trajectories during the intervention are highly heterogeneous and are associated with both clinical and occupational factors, suggesting that tailored intervention strategies may be needed to enhance treatment effectiveness in high-risk occupational groups.