Amid accelerating urbanization, the rapid growth of private car ownership has intensified commuting imbalance, energy consumption, and environmental governance costs. To address these challenges, customized bus (CB) services offer an effective means of optimizing urban passenger transport, easing congestion, and reducing emissions. This study integrates macro-level mobility demand data with micro-level travel behavior surveys to evaluate the city-scale carbon reduction potential of CB systems in Dalian and Shenyang. Using mobile signaling data, the study identifies commuting flows and distances, while revealed and stated preference surveys capture mode choice behaviors. A multinomial logit (MNL) model estimates modal shares and, based on the independence of irrelevant alternatives property, infers counterfactual mode distributions without CB services. Scenario analysis shows that introducing CB services shifts approximately 6.2% and 6.4% of commuters in Dalian and Shenyang, respectively, from other modes, demonstrating its potential for reducing urban transport carbon emissions.