Technical Note—Waterfall and Agile Product Development Approaches: Disjunctive Stochastic Programming Formulations

The periodic selection of new product development (NPD) projects is a crucial operational decision. The main goals of start-up companies in NPD are to attain a reliable return level and deliver this return level fast. Achieving these goals is complicated because of uncertainties in projects’ returns and durations. We develop new disjunctive stochastic programming models that capture the above-mentioned NPD goals. The first stochastic model is static, representing the traditional waterfall product development process, whereas the second one is dynamic, representing the agile product development process. We design a reformulation method and a decomposition algorithm to solve a problem encountered by a U.S.-based software start-up company. Our results indicate counterintuitively that high reliability in attaining a targeted return may be achieved by investing in projects with a longer development time and higher risk. Furthermore, we show that if the capability to make dynamic decisions is overlooked while available, the time to attain the targeted return is overestimated.