In the initial stage of EV vehicle architecture development, quantitative performance development standards or goals are needed to develop driving performance concept that reflects customer needs. Currently, there are insufficient quantitative and detailed performance standards to suggest the architectural direction related to driving performance in the concept stage. In the concept stage, it is also necessary to establish performance goals for each vehicle class/type. In other words, in the concept stage it is necessary to differentiate the target through the application of weights for each performance. Additionally, an objective performance target that meets the vehicle segments is required. To satisfy these architectural development requirements, a concept development process was proposed in this study. In the initial stage of development, various architectures are reviewed, and activities are conducted to select the optimal architecture. In this study, a series of processes from architecture creation, performance review and optimal architecture selection were systematically implemented based on a lowinducing physical performance model. In addition, based on the selected optimal architecture, a virtual driving performance evaluation environment from the perspective of overall performances was established. By developing a structured concept model, it was possible to maintain consistency of analysis model and data management and to explore quickly various architectures. Through this, it was intended to promote data-based exploration. Finally, through the virtual driving performance environment, it was possible to check the virtual performances and at the same time, to review the product group target setting by using the benchmarking data of a competitor vehicle. By introducing the concept of driving performance index based on test data into the virtual environment, predicting the driving performances of the concept vehicle was achieved and this for early phases of the design.