Many software vendors, analyst and journalist are overusing the term “Data Governance” in today’s complex business and IT environments. However, it has become one of the primary goals and drivers for data-related IT projects whilst at the same time being one of the most difficult to define, measure and quantify. What real meaning can we give to the concept of Data Governance? What are its importance, impact and meaning for the enterprise?
To try returning some meaning and context to Data Governance, let’s go back to the semantics through an analogy understandable by everyone, and insightful in the smallest detail.
Welcome to Data Land… If data are its citizens, the governance of such a country would aim at ensuring that these data co-exist in a peaceful way, stayed healthy, enriched themselves, were not living on top of each other, did not destroy each other in the case of conflict, and most importantly work together every year at improving the GDP of Data Land. This means creating value by the use and action of everyone. Of course, bioethics laws would prevent the cloning or duplication of its inhabitants… Data governance would then define itself as a framework which intends to ensure the efficient management of the data in the enterprise. Putting data under governance prevents its chaotic generation and use.
In Data Land, governance implies:
A territory to govern
The scope of influence of governance must be clearly defined, the border of its country clearly delimited.
What type of data are we talking about? The question of the perimeter is not a trivial one, and its impact on the projects and tools to be implemented is big. Master data, so critical that it commands a particular investment for its management, forms a first consistent set and its governance leads to MDM projects.
What about transactions data or social interactions data? More and more popular, full of intelligence for the enterprise, they do not fit into the normal referential bucket, but can benefit from data quality initiatives, with their own specific concerns (volume and volatility, for instance). Also, Data Land is not free from globalization. Though it is important to establish borders for security reasons, “common market” initiatives with neighboring countries (partners, data vendors, data pools) are increasing and aim at surpassing the scope of the traditional enterprise, in favor of the “exterprise”.
Any Data Land (for instance the master data one) must have a leader, a sponsor who conveys a vision and ensures the alignment of all the members of his government who, like in real life, may be tempted by the will of handling its governed data in an autonomous or selfish way. This executive sponsorship is an important success factor of projects related to data governance. Its absence, source of the famous government hitches, leads to systematic failure. The governor is often an executive (CIO, COO, CEO), with enough power and respected enough to impose a choice in case of blocking.
A government and its supervisory body:
Every country needs a team who define the detail in terms of strategy and laws to put in place to ensure it is functioning correctly. Data Land requires nothing less. The organizational changes and the setup of dedicated enterprise-wide teams are among the most advertised collateral for data governance projects. The Data Governance Council is tasked with defining the rules governing the data i.e. the law. The Data Stewards ensure compliance with the law and, if not enforced, will take action to ensure compliance. In order for the initiative to be a success and just like national governments, they should theoretically be independent of particular interests and business lobbies. They do however need to have an intimate knowledge of the data and its use in the enterprise processes. This is why they often come from the “civil society”, meaning they were members of the business teams before, with a mission of surpassing their previous assignment for the greater good.
Laws and institutional processes
The first objective of the abovementioned government is to establish the governance scheme, the set of rules that govern the best practices around creating, using, modifying and removing data. These laws are of multiple types. The ones that establish property titles (data owners), easement rights (data consumers) and security rules (data custodians). There are also the ones that define the boundaries, restrictions or more positively the data standards. These rules will be enforced and data controlled by the data stewards. As in the civil society, an efficient management of the data involves orderly empowerment of the actors (prevention) as well as systematic control (repression). The enforcement of the law and its corrective aspect may be supported by processes orchestrating multiple users, according to the scheme defined by the Governance Council.
So what about IT tools? They are the infrastructures of Data Land. Vehicles, road signs, and even if it is less fun, speed cameras. They are here to facilitate the application of the governance scheme, to give tools to the government, to enforce order and the respect of the law. In any circumstances, they can help with the definition of the scheme. Data governance is an initiative taken by the enterprise for the enterprise, independently of any IT solution which will have to adapt (if sufficiently flexible).
As with any country-based government, data governance has an ambition to manage the enterprise data landscape with perfect efficiency.
Ambitious ? Surely.
Critical ? Definitively.
Let’s then ensure that the way to this ideal will deliver value by itself. This is what the relevance of IT tools should be judged against.
Special thanks to David Jordan for translating the original article from French to English.