Data Governance

Key ideas#

Governance vs. management#

DMBOK separates oversight from execution: - Data governance is about doing the right things: setting policies, standards and architecture, and resolving and escalating issues. DMBOK likens it to the legislative and judicial branches of government, which make and interpret the rules. - Data management is about doing things right: protecting, maintaining and administering data day to day, like an executive branch.

The data-centric organization#

A data-centric organization values data as an asset and manages it through its whole life cycle, starting at design. Data stops being a by-product of processes and applications; high-quality data becomes a goal of the business processes themselves. DMBOK warns that people often confuse data with IT. Managing data is not the same as managing IT, and culture (politics, unclear ownership, budget competition, legacy systems) can block the shift.

Operating models#

• Centralized: one governance organization oversees all data in all subject areas.
• Replicated: every business unit adopts the same model and standards but runs governance on its own.
• Federated: one body coordinates several business units, keeping definitions and standards consistent without centralizing the work.

Governance bodies#

• Data governance steering committee: senior executives from across the organization; approves funding.
• Data governance council (DGC): runs governance initiatives such as policies and metrics, and handles issue escalation.
• Data governance office (DGO): maintains enterprise-wide data definitions and standards.
• Data stewardship teams: focus on one subject area or project; business and technical stewards plus data analysts.
• Local governance committees: for divisions or departments in large organizations. DMBOK advises small organizations to avoid them.

Data stewards#

A data steward manages data on behalf of the organization and its stakeholders, not for personal purposes. DMBOK distinguishes several types: - Chief data stewards chair governance bodies or act as chief data officer (CDO). - Executive data stewards are senior managers on the governance council. - Enterprise data stewards oversee a data domain across business functions. - Business data stewards are subject matter experts accountable for a subset of data, working directly with stakeholders. - A data owner is a business data steward with confirmed authority to approve decisions about their data. - Technical data stewards are IT specialists: integration specialists, database administrators, data quality analysts and metadata administrators. - Coordinating data stewards lead steward teams and represent them across levels, especially in large organizations.

DMBOK repeats an observation from its first edition: good stewards are often found rather than made. Many organizations already have people doing steward work informally; recognize and formalize the role instead of inventing it. Stewards can also be trained.

Stewards focus on four things: key metadata, including the business glossary; documenting rules and standards, including business rules hidden in processes; managing data quality issues; and making sure projects follow governance policies.

Policies vs. standards#

Policies say what to do and not do: the principles for creating, acquiring, securing and using data and keeping it accurate. Standards and procedures say how. Keep policies few, short and precise; operational documents refer to them.