Framework para aferir produtividade de desenvolvedores de software: métricas e estimativas

• Ano de defesa: 2025
• Autor(a) principal: Vargas, Daniel de
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Santa Maria; Brasil; UFSM; Programa de Pós-Graduação em Ciência da Computação; Centro de Tecnologia
• Palavras-chave em Português: Acurácia de estimativas; Desenvolvimento ágil; Avaliação de produtividade; Medidas e métricas; Estimation accuracy; Agile development; Productivity evaluation; Measures and metrics; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: http://repositorio.ufsm.br/handle/1/36723

In agile software development, evaluating developer productivity requires adopting measures that capture not only productivity but also the quality of the work delivered. At the same time, the accuracy of effort estimates plays a central role in planning, delivery predictability, and efficient use of resources. This work aims to improve the evaluation of software developer productivity from two complementary perspectives: (i) selecting appropriate measures and metrics that consider the quality of what is produced, and (ii) applying actions that enhance the accuracy of effort estimates within agile methodologies. In the first phase of the research, a systematic literature review with a defined protocol identified measures and metrics used in productivity assessment. The study mapped 117 measures and 88 metrics, later organized into categories, with the addition of a new category specifically addressing quality-related aspects. Results indicate that process-oriented measures dominate, particularly time per task and task size (e.g., Story Points and T-Shirt Sizing) and while indicators directly linked to quality remain limited. The second phase applied an action research study in an agile software development team, spanning 27 months and 33 development iterations. Two intervention cycles focused on improving estimate accuracy. Results show significant progress, with estimates increasingly closer to actual values and fewer significant deviations. Positive side effects also emerged, such as fewer unplanned tasks, reduced gaps between estimated and actual effort, and shorter planning meetings. These improvements highlight how simple interventions, such as accuracy reports and recurring feedback, foster more reliable forecasts and more consistent team productivity. Together, the two phases led to the proposal of a framework for evaluating developer productivity in agile teams, integrating productivity and quality metrics, continuous feedback, and iterative interventions to enhance both estimate accuracy and the recognition of work quality. For academia, the findings expand understanding of how to integrate quality into productivity metrics and demonstrate the role of feedback in improving estimation accuracy. For industry, the study provides practical evidence that accessible measures and periodic reports can help managers and teams increase predictability, reduce rework, and balance productivity with quality.