Effective data collection is crucial for the success of data science projects, ensuring that data is of high quality and sufficient quantity for practical applications. However, the process is often complex, time-consuming, and prone to yielding low-quality data if not well-organized. Despite its importance, standardized protocols for data collection are lacking, leading to inconsistencies across projects. This paper introduces a comprehensive data collection protocol aimed at streamlining the process and enhancing data quality. The protocol is exemplified through the ADRENALIN project, which focuses on developing advanced machine learning algorithms for smart control of heating and cooling systems. The case study demonstrates the protocol’s practical application, showcasing its effectiveness in overcoming common data collection challenges and ensuring reliable outcomes. By providing a structured approach, this protocol improves the consistency and comparability of datasets, facilitating better benchmarking and more accurate data-driven solutions, thus filling a critical gap in the literature and offering a valuable tool for researchers and practitioners.