A data warehouse (DWH) is an integrated collection of worthwhile data for management support. Unfortunately operational databases – the main source of feeding the DWH with internal data – only provide data with poor management value when no transformation has taken place. The paper deals with this topic. It concentrates on a conceptual architecture of a DWH, in which the transformation machine is the main constituent part of the architecture. This machine supports sub-processes of filtering, harmonization, aggregation and enrichment and is maintained by controller and technician interfaces. Furthermore an access, load and meta data administration handles the secure and documented loading and accessing of relevant data and guarantees technical and businessoriented documentation of all transformation activities.