Purpose: The technological developments have implied that companies store increasingly more data. However, data quality maintenance work is often neglected, and poor quality business data constitute a significant cost factor for many companies. This paper argues that perfect data quality should not be the goal, but instead the data quality should be improved to only a certain level. The paper focuses on how to identify the optimal data quality level. Design/methodology/approach: The paper starts with a review of data quality literature. On this basis, the paper proposes a definition of the optimal data maintenance effort and a classification of costs inflicted by poor quality data. These propositions are investigated by a case study. Findings: The paper proposes: (1) a definition of the optimal data maintenance effort and (2) a classification of costs inflicted by poor quality data. A case study illustrates the usefulness of these propositions. Research limitations/implications: The paper provides definitions in relation to the costs of poor quality data and the data quality maintenance effort. Future research may build on these definitions. To further develop the contributions of the paper, more studies are needed. Practical implications: As illustrated by the case study, the definitions provided by this paper can be used for determining the right data maintenance effort and costs inflicted by poor quality data. In many companies, such insights may lead to significant savings. Originality/value: The paper provides a clarification of what are the costs of poor quality data and defines the relation to data quality maintenance effort. This represents an original contribution of value to future research and practice.