Requirements analysis plays a key role within any development project to reduce the risk of failure. In business intelligence projects, Data Warehouses (DWs) gather heterogenous data from different data sources and provide a unified view of data to be used by decision makers using different visualization tools. Different studies recognized the importance of conducting DW design by users’ requirements in conventional DWs, which enriched DW design cycle by a first phase of requirements definition. The emergence of semantic web technologies like ontologies consolidated DW design and provided a new architecture of DWs: semantic Data Warehouses (SDW) defined as repositories of semantically integrated data. A new generation of SDWs emerged recently that considers external semantic resources shared on the web: Linked Open Data (LOD). LOD datasets complete internal sources by new and relevant information for decision making. In this context, different studies proposed solutions and approaches for designing SDW and integrating LOD datasets. Similarly to first studies of conventional DW design, these approaches focused on data issues (like integration and multidimensionnality) and ignored the importance of users’ requirements. This study learns from past experience and proposes requirements-driven apporach for designing SDW from internal and LOD datasets, by considering requirements incrementally. This approch is based on the formalization of SDW design cycle. This formalization is supported by an ontological framework defining design decisions of each requirement as traces in the ontology metamodel. Different experiments are conducted for showing the impact of incorporating exploratory requirements at the schema level and at the instance level of the SDW. SSB benchmark is adapted and used to represent internal sources. Two LOD datasets are used: Yago and DBpedia. The proposed approach is illustrated through a case study analyzing book sales transactions.