We investigated the failure of an IS project within a global industrial company. The challenges in the complexity of product master data were one of the failure reasons. We detected a research gap in how to handle master data complexity in IS development, especially in data storage integrations. The tradi-tional approach is to define so-called golden records, “single versions of truth”, for each record, and then harmonize and cleanse data so that only or mainly golden record values will be used. We offer federative approach as an alternative to the golden record approach. According to this approach data interoperability is achieved by identifying shared attributes, by federating data on the basis of shared attributes’ metadata, and by developing IS functionalities to process the metadata and their cross-references. We compare the ontological stances of the approaches theoretically and with figures. We present the results of a case, where the federative approach was probed. Our study contributes to re-search by showing how to link data management to IS development to address the complexity of mas-ter data in data interoperability projects, by comparing the golden record and the federative ap-proaches, and by showing how the federative approach can be used in real-life contexts.