A long-term appraisal of commuting change is crucial as it reflects resident’s responses to planned urban changes. Previous literature tends to tap on the mobile phone data (MPD) to estimate commuting flows, but its use is often limited to a representation rather than an explanation of the changing phenomenon. This research utilises a planning-support model that combines increasingly available MPD and conventional urban datasets to construct a well-calibrated commuting matrix in the base year 2010. Based on the sound outcomes through the cross-year model validation, counterfactual simulations have been conducted to help local planners and decision-makers understand what leads to the differences in commutes across categorical zones (i.e. centre, near suburbs, sub-centres, and far suburbs). The results show that both housing supply and jobs decentralisation in the city centre contribute to a shorter and more balanced commute, while the latter one has a larger effect. As a science of planning, the proposed model coupled with empirical evidence across years has the ability to deliver timely and situation-cogent messages regarding the success or failure of planned policy incentives.