Organizational improvisation (OI) has been gaining an increasing attention to respond to rapidly changing environments, and it needs close and proper management. However, most of the findings on Improvisation I information system development studies are based on variance-based models. Thus, Du et al. (2019) propose a process model that features a continuous iteration between improvisational search and build in ISD. Their four-phase model describes continuous and iterative methods of organizational improvisation to respond to opportunities and threats, presenting an excellent step-by-step guideline for information system development managers to refer to from a practical standpoint. Despite the contributions, their exploratory research is not immune to limitations. Although the authors explained that learning is working in their model, there is only a fragmentary explanation of how the learning process works. In this work, we explore the boundary condition of Du et al.'s (2019) proposed model for the OI process in ISD then suggest a new model for ISD by combining the evaluation and learning process model proposed by Beynon-Davies et al. (2004), which is based on Argyris and Schon (1978). We believe that the new framework will help us apprehend that organizational improvisation in ISD generates short-term learning and long-term learning through the evaluation and learning from a process model perspective.