The data governance “industry” thrives on a curious dichotomy. On the one hand, some service providers insist to clients that they need a data governance program, that they must create a data governance council and that they should immediately staff a collection of roles ranging from data governance council member to data steward. This has led to a promulgation of the ubiquitous stack of PowerPoint slides that grace each data governance manager’s desk. But the creation of an org chart does not guarantee that the quality and integrity of data will automatically improve.On the other hand, some tools vendors advocate use of their metadata, data quality and emerging data policy management tools as key components of a data governance activity. But again, acquiring tools will not automatically improve the usability of enterprise data.

Making data governance operational requires an operating model: well-defined processes that ensure execution of the procedures associated with data policies and enforce compliance with data consumer expectations. The operating model uses defined processes to map the roles in the org chart to the proper use of tools and techniques for observing agreed-to data policies.

When our company, Knowledge Integrity, works with customers to assemble an operating model, we typically focus on three levels of constituents: the executive level, the tactical level and the operational level:

• Executive level. The responsibilities of people at this level include communicating the value of data governance to business owners and resolving data management issues that cross lines of business.
• Tactical level. The responsibilities of people at this level include overseeing the quality of data that's shared across the enterprise (such as master data and reference data); overseeing creation of the business glossary, metadata repository, and shared business/data quality rules; and ensuring that processes and techniques that rely on acquired tools across the represented lines of business are integrated.
• Operational level. People at this level assume day-to-day responsibility for data usability, such as data quality assessments, implementation of data quality controls, responding to reported data issues, managing the use of shared reference data, master data, and business rules, and resolving issues to ensure data utility.

A well-conceived operating model provides an inventory of processes that bridge the gap between policy and implementation. The data policies that are defined by the data governance council (at the tactical or perhaps even the executive level) must be “translated” into methods for the data steward to employ tools such as data profiling, metadata management, or incident reporting and tracking to achieve the measurable compliance with defined expectations.

The real trick, though, is developing the operating model before mapping out the org chart or buying the tools. Having a well-defined operating model guides each of the participants in what they will be expected to do and how it is expected to be done, which will alleviate some of the uncertainty and confusion that often accompanies the initial attempts at data governance. In turn, your approach to governance will be more successful because there will be a plan for making data governance operational.