Composing an experiment or a credible simulation model to search for a new optimized design may be costly in terms of both money and time as the design has not been realized yet. To alle- viate this problem, we propose using Designable generative adversarial network (DGAN) to quickly suggest a new effective design. DGAN is derived from the Generative adversarial net- work (GAN) with the addition of an inverted generative network structure. DGAN not only augments input data but also yields the corresponding design variable data for the augmented data. Assuming a product of an analogous domain was designed in advance, its performance data is utilized with DGAN in order to adapt to a new design domain and to bring out a design that attains similar performance. The effectiveness of the proposed methodology is investigated with problems designed to incorporate equations and simulation models.