Volatilidade de requisitos em ambientes dinâmicos e com múltiplos stakeholders: estratégias e boas práticas para apoiar o processo ágil de desenvolvimento de aplicação de software governamentais

Abstract

The increasing demand for quality across various aspects of software products heightens the complexity of processes surrounding the software lifecycle. Certain characteristics further amplify this complexity, such as the dynamic nature of the environment and the high number of stakeholders, which make functional requirements highly volatile. In this context, this study aims to investigate, both in the state of the art and practice, how the quality of a computational solution and its development process are impacted by requirement volatility in dynamic environments with multiple stakeholders. To achieve this, a state-of-the-art review was conducted to understand how the topic has been approached and the current level of investigations. Following this, a practical investigation was carried out in a real environment characterized as dynamic and involving multiple stakeholders. The case study focused on PNLD Avaliação, a software developed to support the Pedagogical Evaluation of the National Textbook and Teaching Material Program. In the first stage, a historical and documentary inspection of the software artifacts generated during the development lifecycle was conducted, allowing for a better understanding of the research object’s nuances, the identification of stakeholders and their roles, and the impact of periodic scope changes. Applying the Chi-Square statistical test confirmed that differences in the number of user stories altered, discontinued, and completed across the two studied periods were statistically significant and likely not due to random variation. Subsequently, in the second stage, semi-structured interviews with various project stakeholders revealed several broad challenges in managing dynamic environments and multiple stakeholders. Despite the use of agile methodologies, their effectiveness was limited by unclear documentation, flawed process definitions, team overload, and the absence of a formal mediator, which hindered alignment and the resolution of critical issues. Finally, in the third stage, the SUS method was applied alongside an open questionnaire to capture end-user perceptions of the studied system’s usability. Although the overall SUS score indicated a positive rating (73.2, five points above the minimum threshold), approximately 40% of users expressed general dissatisfaction with the system. For deeper insights, the open-ended questionnaire responses underwent Deductive Thematic Analysis, highlighting structural issues such as inadequate integration and redundancies that hampered productivity and user experience. The lack of iteration and feedback during development resulted in a complex interface requiring constant re-learning, frustrating users. Overall, this study emphasized how conflicting interests, poor communication among stakeholders, and the absence of process centralization around stakeholders add complexity to the software lifecycle, further weakening it in dynamic environments. Based on these findings, a set of best practices was proposed to support the management of software development projects operating in complex, dynamic environments with multiple stakeholders. Additionally, an agile development process was outlined, considering the nuances of established best practices to enhance workflows’ robustness and efficiency in similar environments.