Purpose This paper aims to provide a process-driven scientific data quality (DQ) monitoring framework by information product map (IP-Map) in identifying the root causes of poor DQ issues so as to assure the quality of scientific data. Design/methodology/approach First, a general scientific data life cycle model is constructed based on eight classical models and 37 researchers’ experience. Then, the IP-Map is constructed to visualize the scientific data manufacturing process. After that, the potential deficiencies that may arise and DQ issues are examined from the aspects of process and data stakeholders. Finally, the corresponding strategies for improving scientific DQ are put forward. Findings The scientific data manufacturing process and data stakeholders’ responsibilities could be clearly visualized by the IP-Map. The proposed process-driven framework is helpful in clarifying the root causes of DQ vulnerabilities in scientific data. Research limitations/implications As for the implications for researchers, the process-driven framework proposed in this paper provides a better understanding of scientific DQ issues during implementing a research project as well as providing a useful method to analyse those DQ issues based on IP-Map approach from the aspects of process and data stakeholders. Practical implications The process-driven framework is beneficial for the research institutions, scientific data management centres and researchers to better manage the scientific data manufacturing process and solve the scientific DQ issues. Originality/value This research proposes a general scientific data life cycle model and further provides a process-driven scientific DQ monitoring framework for identifying the root causes of poor data issues from the aspects of process and stakeholders which have been ignored by existing information technology-driven solutions. This study is likely to lead to an improved approach to assuring the scientific DQ and is applicable in different research fields.