Today’s rapidly changing and highly regulated business environments demand that organizations are agile in their decision making and data handling. At the same time, transparency in the decision making processes and in how they are adjusted is of critical importance as well. Our research focusses on obtaining transparency by not only documenting but also enforcing data governance policies and their resultant business and data rules by using a multi-level metadata approach. The multi-level approach makes a separation between different concerns: policy formulation, rule specification and enforcement. This separation does not only give more agility but also allows many different implementation architectures. The main types are described and evaluated.