Dynamic pricing of electricity is receiving growing attention as a key enabler of the smart grid and a solution for promoting grid resilience to increasing penetration of renewable sources. Despite various dynamic pricing programs are contemplated and actively implemented by utilities in many developed economies, corporate decision makers are not aware of the extent to which their businesses can adapt to the programs and its associated risks and opportunities. This study investigates the demand response of commercial and industrial businesses to the Korean critical peak pricing pilot which was implemented in the winter of 2013. The demand responses of 802 businesses covering 34 commercial and industrial categories are evaluated and characterized by constructing their individual baseline loads. The results resoundingly confirm substantial differences in demand response across and within the segments, with the magnitude of response closely correlated with the firms' expenditure shares of electricity. It is thus suggested that conventional business classification is not sufficient enough to characterize firms' demand response to dynamic pricing plans, calling for firm-level, data-based approaches to consumer segmentation and electricity services. This study also identifies two important types of response behavior to which due attention has not been paid: shortterm learning effect - that is, the demand responsiveness increasing with CPP event experiences and - seemingly inefficient behavior of preparatory and recovering reductions in off-peak hours neighboring critical peak hours.