This article originally appeared on the BeyeNETWORK.
There is no single definition for “data governance” that is accepted industry-wide. Perhaps the most straightforward definition is:
"Data governance is the organizing framework for establishing strategy, objectives, and policies for corporate data.” 1
Like its definition, there is no single way to design a data governance program. Each must be tailored to the specific needs and culture of the corporation it will serve – hence the need for best practices to follow.
Data governance is built on a foundation composed of the five elements depicted in Figure 1.
Figure 1: The Five Foundational Elements of Data Governance
These five foundational elements can be described as follows:
- Transparency is about clear and open communications regarding all aspects of governance. Transparency supports data governance by making sure all its activities are conducted in the open.
- Measurability provides the verification that governance is performing as expected. Without measurability, progress cannot be conclusively demonstrated.
- Alignment ensures that a data governance program aligns with the needs of the business, supports in-process and planned initiatives, and ultimately ties back to corporate strategy.
- Decision rights are the hallmark of all data governance programs. It’s not enough to simply convene a council. Members of the data governance council should be accountable for delivering measurable business value to the organization, with each council member charged with a particular area of decision-making.
- Oversight is what data governance is all about. Data governance oversees data and all aspects of data.
Each foundational element described has its own structure, which, in turn, is supported by one or more of the best practices described in the following section.
The Driving Forces Behind Data Governance
Four significant forces in the market have converged creating significant pressure on business executives. The first force is the amount of data being accumulated. Every estimate for the amount of information generated annually is trending up dramatically. The Advisory Council is projecting raw data set sizes to “increase in the range of 30 to 50 percent per year for the next few years.”2 Estimates by most other industry watchers are in the same range. The increase is so dramatic that it has been called an “information tsunami.”3 Data governance focuses on managing this proliferation.
The second force is regulatory oversight and compliance. Increased regulatory requirements, corporate embarrassments over the release of incorrect information and the general perception that the state of data is out of control have caused executives to sit up and take notice. The result has been a rapid increase in the deployment of data governance as a way of satisfying auditors and regulators.
The third force at work is the need for better data security. The need to safeguard corporate information from misuse, exposure and theft has prompted many companies to adopt data governance as a solution. In addition to establishing how an enterprise will define and reconcile its data, data governance places auditable controls on data access.
The fourth force is an emerging understanding among executives of the promise of information uses for business advantage. As an asset, cash can be used once. Data can be used over and over again. Moreover, there is a newfound realization that the only way data can be used and reused is if it is reliable. The way to control the reliability of data is through data governance.
With the convergence of these four forces, business leaders are beginning to ask how corporate data can be controlled in a process-driven, business-sanctioned way. They are looking beyond IT for the support, cleansing, and certification of corporate data. They are looking to data governance to provide solutions.
Until recently, there were no recognized best practices for data governance programs. This has at least in part been the result of the view that companies’ cultural, political and organizational issues tended to be company-specific and, therefore, each data governance program had to be totally unique. As more and more companies have initiated data governance programs, however, a number of common themes for data governance have emerged. Because these themes are common to virtually all highly functioning data governance programs, they should be viewed as “best practices” to be followed wherever applicable.
A best practice is defined as “an idea that asserts that there is a technique, method, process, activity, incentive or reward that is more effective at delivering a particular outcome than any other technique, method, process, etc.”
This description is exactly on point. The key is that a best practice is more effective at delivering a particular outcome than any other. Best practices are all about effective delivery. Data governance, or almost anything else for that matter, can be done without incorporating best practices; however, the delivery is vastly more effective if best practices are followed.
Best practices evolve over time. While the best practices presented in this article are not likely to fall out of favor, new ones will be added as more and more data governance programs are developed. Additionally, best practices rarely stand alone. Rather, taken together, they provide an interlocking web of capabilities. You can use these to design your data governance program the right way.
The “Sweet 16”
The 16 best practices in data governance can be divided into two categories: Developmental Best Practices and Operational Best Practices.
- Developmental Best Practices – those practices that are applicable primarily to the process of getting the data governance organization up and running.
- Operational Best Practices – those practices that are applicable to the day-to-day operations of a Data Governance Organization after it has been implemented.
The remainder of this article will discuss the seven developmental best practices. Part 2 in next month’s newsletter will describe the remaining nine operational best practices.
Developmental Best Practices
Developmental best practices for data governance involve the initial design and setup of the data governance program. The following best practices represent those that we have found should be adopted as data governance is launched.
Intentionally design your data governance program.
Data governance does not just happen, it must be intentionally designed.
The process of data governance design is for a core team of business and IT people to build commitment, socialize the vision, cement relationships, and think through governance processes and decision-making scenarios. Effective governance is the result of deliberate, iterative thinking and planning, not simply announcing that a governance council has been formed.
Design in cross-functionality and cooperativeness.
The full potential of data governance cannot be realized unless both business and IT actively work together to “make it happen.” Without open and active cooperation and participation from both IT and the business community, data governance risks becoming a siloed initiative or, worse, a one-time only effort. Yes, it can be done without full cooperation from both sides; and, yes, one side or the other alone can do it, but it can never reach its full potential unless both sides are actively involved.
Perhaps the most effective way of getting business and IT to cooperate on a data governance program is to establish a shared vision of data governance and an understanding of what it can do for both organizations. Without this shared vision, it is hard to keep both sides on task and on target.
The cross-functional and cooperative nature of data governance must not be lost in the transition from the developmental phase to the operational phase of data governance. It must be fostered throughout its life. This then is both a developmental as well as an operational best practice.
Find a burning business issue.
Don’t assume, as many have, that data quality is the only problem to be addressed by data governance. In all likelihood, data quality will be one issue on a list of many. It may or may not, however, be the highest priority item. Allow business to gauge the impact of the various goals of data governance, exposing the most serious pain points, and address them in a prioritized fashion.
Make sure data governance fits your culture.
Governance must deal with the social system or behavioral side of decision-making – relationships, interactions, attitudes, thinking, and learning. For this reason, corporate culture has as much to do with how data governance is designed as does anything else.
It takes time to change the way people think, work and institutionalize new processes. This will only reasonably happen if the data governance program aligns with – and doesn’t try to change – the culture of the organization.
Start from the bottom up, not the top down.
Starting data governance by forming a data governance steering committee and expecting the people involved to create an enterprise data governance program is fraught with problems and politics. While going from the top down may work, experience shows that data governance programs are less likely to be successful if they begin as cross-function committees charged with “defining” data governance. The reason for this is simple: there is too much tacit knowledge and entrenched culture involved in getting a data governance program off to a solid start. Executives simply do not have the time, the temperament and, in many instances, the detailed knowledge to create the support mechanisms necessary to make governance work – particularly if it doesn’t immediately serve their specific needs.
What has been shown to work is a bottom-up approach. Start by focusing on a carefully targeted business initiative like business intelligence or a single data subject area such as customer or finance data. This careful targeting of data governance reduces the overall scope of the initial effort and focuses on a single area of the company. After the initial target has been selected, the process of designing and documenting a functional framework for governance, the roles and responsibilities involved and a minimum set of policies and procedures necessary to support the functioning of a data governance organization can begin. Only when the functional framework is in place and ready to go should people be recruited for the various defined roles. Once the targeted program has been implemented and proven, it can be expanded to other areas following an incremental approach.
Establish clear communications.
Clear, well-defined and well-designed communications must be one of the cornerstones upon which the efforts to build a data governance program rest. A proactive communications plan should be thought out very early in the data governance development process. Distributing a weekly status report is simply not enough. Implementing data governance means that the way things are done will be changing. Change makes many people nervous. It is incumbent upon those designing the data governance program to ensure that stakeholders are comfortable with what is being proposed. The best way of doing so is through communications targeted to address concerns – describe how things will be done and establish a common vocabulary.
As an example, a series of short, crisply articulated education pieces should be distributed on subjects like:
- What data governance does (and what it does not do).
- The role of the data steward.
- The benefits of data governance.
- What is data management?
- What is data quality and how do we institute it?
- How will data governance affect my job?
The list of topics could go on, but should be tailored to address concerns identified by stakeholders.
Communication on the subject of data governance should not stop the moment the organization is up and running. Once the data governance program is operational, the status and direction of the organization should be communicated to all stakeholders on a regular basis. Make sure stakeholders know that governance is working, and tell stories about accomplishments and progress.
Define the scope.
It’s important to establish and communicate what data governance will do – and what it won’t do. Establish the initial scope of the program by selecting a single department or a single master data type to use as a “pilot” or “proof of concept” with which to get governance started. Once started, it will be much easier to justify expanding governance to other areas because there will be a record of accomplishment at which to point in justifying the expansion. Additionally, starting with a focused initial area allows for learning, making adjustments and correcting mistakes under less than an enterprise-wide spotlight.
Whether you are beginning with a departmental effort, like business intelligence governance, or a single subject area data governance program, it is of the utmost importance that the scope of the effort be well defined and widely broadcast. Nothing will more quickly put a damper on an otherwise well executed data governance program than having stakeholders with expectations the program was not designed to fulfill in the first place.
Part of the scope of all data governance initiatives must be the explicit statement that data governance is a program, not a project. Once initiated, data governance ideally evolves as an enterprise-wide capability upon which multiple organizations can rely.
Part of the process of setting the scope of the data governance program is setting the goals for the program. The goals define what the data governance organization expects to accomplish in a given time frame. These goals must have four characteristics:
They must be complimentary to the goals of the enterprise as a whole. It does no good to settle on goals for data governance that are at odds with those set for the corporation.
- They must be meaningful. Goals selected must be applicable to the data governance organization and must be selected to demonstrate the progress of the organization.
- They must be defendable. Setting a goal of “make the users happy” may be laudable, but it is very hard to quantify and, therefore, hard to measure. Clearly, a user survey could be distributed with the goal of ferreting out a “user happiness quotient.” A far more quantifiable goal might be, “Reduce by 25% the number of data-related help desk calls.” True, the two metrics do not track exactly the same thing, but it is arguably better to have a defensible hard number that is slightly off the mark but generally applicable than a soft number which may or may not be reproducible and requires a lot more defending than a hard number.
- They must be measurable. Goals must be items that can be measured – if not by hard numbers, than at least by soft numbers. Nothing positive will come of selecting a goal that by its nature can not be measured – no matter how desirable the improvement might be.
In Part 2 of this article, we’ll move from developmental best practices to the operational best practices that will serve the data governance program for the long-term.
- Dyché, Jill and Evan Levy, Customer Data Integration.
- The Advisory Council – February 22, 2007.
- Strategic Management Resources, Ltd. "Storage Trends into the 21st Century."