Data governance is an essential discipline in a context of increasing enterprise digitalization in all areas of the business. Data is at the heart of the digital transformation and data governance ensures its quality, integrity and appropriate use.
While its purpose is clear, there are many dimensions that data governance must address: strategy, policies and rules, organization and roles, responsibilities and competencies, and processes and technologies. On the one hand, it addresses how decisions about data are made by establishing strategy, prioritization and roadmaps. On the other hand, it provides a framework for Data Management practices and Data uses of algorithms or AI models in automated chains.
The polymorphic nature of the discipline, added to the breadth of the subjects to be managed, makes the task difficult. Aborted initiatives are not unusual. These failures have created, to say the least, a perception problem about data governance. In some cases, leaders have gone so far as to rename initiatives to avoid using the "governance term".
But the observed difficulties should not call into question the need to acquire the capacity to have trusted data, which is protected and adapted to uses. This need becomes all the more substantial as investments in analytics and AI accelerate. What lessons have been learned from past experiences? What can be retained as best practices?
1. Leadership and strategy : Successful data governance begins with a visionary and committed leadership. Data activities are driven by a data strategy which is itself aligned with business strategy.
2. Addressing business needs : For too long, data governance has been considered as too conceptual. To be successful, it must be operational and help meet business needs. In particular, it needs to serve both producers and consumers of data so that it can be ingrained in existing practices.
3. Integrated : Data governance activities must be integrated into existing business processes. Business processes, architecture instances, project methodologies, risk management frameworks, and data management activities related to the production, collection and use of data sets, KPIs and reports must all include data governance.
4. Shared Responsibilities : Data governance must be seen as a shared responsibility between business stakeholders, Data professionals and IT players. It is recommended to move from a vertical view of responsibilities to a horizontal view related to the data value chain. To effectively address the need for coordination between functional areas, Data Governance should establish an operational framework that defines responsibilities and interactions.
5. Based on principles : Guiding principles are the foundation for data governance activities. They help articulate corporate values and give meaning to the Data policies and rules that flow from them. This facilitates buy-in and provides additional leverage to address resistance.
6. Multi-level : Data governance must be able to operate at a local level, within a domain or sub-domain, as well as at a transversal level of the enterprise. It also often operates at intermediate levels.
7. Change management : In data governance programs, resistance to change has often been underestimated. However, when dealing with such a transformative initiative, active change management that impacts both individual behavior and collaboration patterns (IT/Data, Data/Business, etc.) is critical to success. The "people side of data" needs to be managed early in the process.
8. Iterative and Sustainable : Data governance should not be viewed as a project with a defined end, but rather as an ongoing iterative process that requires a medium / long term commitment. It underpins changing practices about how data is managed and used. Not necessarily by bringing about massive upheavals but by installing incremental changes that are adopted in a sustainable way by the company's stakeholders.
9. Based on technology : A recent IDC survey shows that excel files and other documentation are still among the most widely used methods when it comes to data governance. But it goes without saying that this tactical approach is inadequate in the face of the changes brought about by digital transformation. To succeed in a sustainable way and to deliver value data governance must be equipped.
10. Measured and driven : Good data governance has a positive financial result with a demonstrated return on investment. To successfully implement a measured approach, it is necessary to assess the initial situation, share the findings with decision-makers and plan for measurable improvements.
Shelemat DANIEL, Décembre 2020