The study delves into the main stages of the life cycle of software systems, emphasizing the integral role of web APIs within this framework. Notably, we focus on analyzing the structural components of development costs associated with these systems. Our findings reveal a considerable portion of the costs dedicated to sustaining the robust development of the software system. This underscores the importance of forecasting the intensity of various types of requests for system changes, enabling the timely fulfillment of customer requirements and accurate estimation of investment costs throughout the software system’s life cycle. To address this need, we propose a method for forecasting costs to support the sustainable development of a software system. This method relies on a hierarchical algorithmic clustering approach applied to retrospective weekly schedules of client requests for system changes. The clustering process is executed separately for each type of change, yielding distinct graphs. Representative graphs within the obtained clusters serve as key elements in predictive calculations. Determining a representative graph for each cluster, along with the corresponding cluster size, enables prediction by adding representative graphs in proportion to the cluster size. Maintaining the correct proportionality is achieved through the implementation of a greedy algorithm. The application of the developed forecasting method to support the sustainable development of the software system is illustrated using experimental data. This approach not only caters to current customer requirements but also ensures a comprehensive estimation of investment costs, offering valuable insights for optimizing the software system’s life cycle.