Data governance institutes a system of accountability when it comes to an organization’s most important data assets. Data providers have an obligation to meet the needs of data consumers when the benefit outweighs the cost. Because this obligation exists across the organization despite functional or geographic distinctions, we have to start with the right organizational structure.
First, we need a high-level authority that can make key decisions for data. Many companies set up a data council composed of a cross-functional team of senior managers and executives. As pointed out earlier, the cost of good data often needs to be borne by one group for the benefit of the entire organization. A strong authority is needed to make and approve these decisions. The council ultimately owns data policies.
Second, a day-to-day role of data stewardship needs to be established. Stewardship is different from ownership, and this distinction is critical in defining this role. Data is owned by the enterprise, not an individual. So the term data owner is at best misleading, and at worse detrimental, leading to data hoarding and silos. Stewardship means caring for something on behalf of a group, and that’s exactly what a data steward does for data. At a macro level, deriving authority from the Data Council, data stewards take care of data on behalf of the enterprise. At a micro level, data stewards represent the interests of those who benefit from data and hold those who provide data accountable.
To be effective, data stewards need to be equipped with carrots and sticks. A partner at a global SI once told me that data stewards are like the workers walking up and down the aisles at British football stadiums. They wear bright yellow jerseys so you know who they are. If you need to know where to buy a drink, they’ll direct you to a refreshment stand. But if you misbehave, they’ll escort you out of the stadium.
For business process owners who are the key stakeholders, the basis for carrots and sticks are both in place. A business process consumes data as input; data stewards can help the process owner obtain necessary data and ensure that the data meet quality standards and expectations. The same process owner also produces data for downstream processes; data stewards will hold the same process owner accountable for the output.
Some companies assign data stewardship responsibilities to individuals in the lines of business who have other jobs. Frequently, these individuals had already been unofficially performing data stewardship tasks, but the role was not formalized, and they have not been granted official authority. Some companies set up dedicated data organizations whose primary function is data stewardship. Either approach can work, depending on the company’s culture and business needs. It is not always necessary to add headcount.
We can draw some parallels to another type of important enterprise level asset: people. To properly care for human resources, we need an HR function. Companies that excel at HR, like Proctor and Gamble, recognize that human capital belongs to the enterprise. They are not owned by a specific business unit, function or line manager. And these companies have well-developed policies and processes for managing people and developing them to maximize their value to the enterprise. HR is responsible for setting these policies and processes, like hiring, promoting, compensating and career development. HR also enforces these policies.
If the HR function is people governance, then we need to do the same for data. Supported by authority and stewardship, data governance sets policies and processes for data regardless where it is created, used and stored. Then, it implements and enforces these policies and operationalizes the processes.
In my next blog, I’ll write about the technology needs of data governance.
This blog is part 6 of a 9-part series of blogs on the topic of Enterprise Data Governance. To read other posts from this series, please see below.