There are a lot of different opinions out there about how to implement data governance. Some offer a more "realistic" approach, while others tout "best practices," and then there are various recommendations between the two.
I suspect the reason there are so many approaches is because there are so many different types of organizations. Traditional, hierarchical organizations will probably deal well with a top-down approach that focuses on best practices and a "do-it-right-the-first-time" approach, while less formal organizations might benefit from a grass-roots, cooperative approach. As data consultant Jim Harris and Forrester analyst Rob Karel smartly pointed out, there's no need to be extremist about it.
There is one thing, however, that holds true regardless of your approach: The business ultimately needs to own data quality and, therefore, data governance. And with more CFOs concerned with the impact of bad data, I suspect the business will soon be on board with data quality initiatives.
Often, when data quality experts talk about the business involvement, they talk about finding a business sponsor and data stewards. But if business is going to own the process, it will need to devote more people and resources to data quality than that, contends Kelle O'Neal, founder and managing partner of First San Francisco Partners, in a recent Beye Network article.
O'Neal defines the business sponsor as the person who spearheads data governance, while the data stewards are "the conduit between IT and the business and are accountable for data and data management process definition and for data quality levels for specific subject areas."
But beyond these roles, she recommends the business commit to the following governance "jobs":
- Business Leadership: This is a steering committee group, composed of someone from each line of business that provides direction, vision and oversight for the governance effort. This is the first group that's formed when data governance becomes a focus, she writes. This group then forms a governing body for data governance called the data governance working group (DGWG).
- Data Governance Working Group: This group "creates, maintains, implements, and monitors policies, processes, and standards to ensure data reliability and sustainability," and should include representatives from across the business. Think of the DGWG as an arbitrator of all data governance decisions - if there's disagreement within the business units, the DGWG's decision overrides them.
- Data Governance Lead: This position sounds like a project manager for governance. It's a business person; although, she notes there should also be an IT equivalent. She defines this role as someone who "serves as gatekeeper of data governance best practices, acts independently across business units and IT, and drives the development and ownership of the data governance roadmap."
- Data Governance Coordinator: An administrative role for handling the nitty-gritty tasks of the data governance organization.
- Business Steward Leads (BSLs): These work with the data stewards to "understand information quality needs and ensure best practices are followed." They also serve as the go-to sources for defining the "use of key data elements and improve business processes," she writes.
- Data Owners: As the name suggests, this person owns and defines the data at the business-unit level. Owners also control access to the data. "Data owners are responsible for data element accuracy, integrity, timeliness and availability," she writes.
O'Neal's recommendations strike me as highly structured and requiring of a more top-down approach to data governance. In her bio, it says "O'Neal manages specialist data governance and data management consulting services to complex organizations," so I suspect she's more accustomed to dealing with structured companies. Certainly, her advice seems to fall more into line with that type of organization.
"Business and technology stakeholders must clearly understand their roles and what is required of them," O'Neal writes. "And because so much of the execution of a data governance initiative must be handled by the business, we recommend, as a best practice, that data governance be owned by the business from the beginning."
It sounds smart and I think it'd work well for many traditional, large companies. Still, I suspect this approach might be too cumbersome for many organizations that would, nonetheless, benefit from data governance.