How to launch a scalable data governance program
Forty years ago, businesses employed people to make products and deliver services with highly tangible and visible processes and outcomes. Several sets of engineers or business analysts with years of tacit process knowledge cross-checked every business form before the information could cause any problem or error. In the last 40 years, however, enterprises have mixed complex information systems with refined processes and automation, employing people to make and manage things that are largely intangible, such as data. This shift has created a challenge for data governance and stewardship. Enterprises would benefit from asking, “Are we giving data the attention we once applied to waste factors, safety stock, corporate risk and customer service?” The answer for most is no. Data governance is the cycle of defining attributes necessary to maintain order, efficiency and control of the data in information systems. Data stewardship is the process of leveraging the attributes defined by data governance to maintain order, efficiency and control of the data. Every large enterprise around the world is currently grappling with these concepts, because they are the next stage-gate for enterprise performance in the information age. Downstream effects of improved data governance and stewardship are reduced cycle times, improved transactional readiness, more trustworthy insight from analytics, greater business agility and leaner, faster and better operations.
The chief difference between governance and stewardship is that governance is a broad term defining the overall, big picture strategy of data management, whereas stewardship refers to the specific measures taken to reinforce that big picture strategy. If the two terms were nesting dolls, stewardship would be found nestled inside the shell of governance.
Analysts and professional services organizations offer a lot of clever definitions of data governance, but those rarely provide a picture of what data governance should “look like” to an organization. The picture is often more important than the definition because some elements of the picture may already exist within an enterprise and a simple definition can’t account for their presence.
The "picture" of data governance is best represented as a matrix, where the three basic components are aligned to particular business objectives:
If an enterprise has all three of these components, then it is engaged in data governance. The degree of balance in a data governance program is derived from order, efficiency and control. These elements are an important measure of balance because data governance can be the flashpoint for common enterprise dilemmas, such as standardization vs. adaptation in operations and consequently data.
A simple exercise in painting the picture of data governance for your company involves laying out the components of organization, processes and business execution knowledge and reviewing what your company has or doesn’t have in those areas. Layer on top of those three components the objectives of order, efficiency and control to define the purpose of actions your company is taking in each area (see figure below).
There are a number of great business reasons to establish data governance, but not all data warrants the same level of rigor. The value of governing a specific field of data is established by decisions on two levels:
1. How closely does governing the field fit the primary mission of the governance program (i.e., the primary business driver)?
2. Will governing the field provide benefits to analytics, operations or compliance within the primary mission of the governance program?
ERP systems are overwhelmed with data that provides no clear business benefit, and in some cases, the data comprises artifacts from business activities decommissioned long ago. Successful organizations look at data governance with a cautious eye to governing too much at once. Keep the business drivers for governing data tight and the threshold for what to govern high, and you will save yourself some time at the white board.
Keeping a high threshold for what an organization needs to govern is a healthy way to start. The organization involved in data governance, presumably the data management organization, has to become proficient in making tough calls of what to govern, when to govern and how to govern. Keeping a high standard for what data to govern would allow this organization to grow into reasonable expectations and develop both processes and business execution knowledge that are crucial to implementing systems and rolling out improved data stewardship.
The value of specific data fields evolves over time as business operations and regulatory climate evolve. Maintaining a high threshold for what an organization decides to govern will pave the way for an iterative or incremental approach to the data governance cycle and return high value in the long run.
There are many types of data worthy of governance within an organization, including enterprise reporting, transactional data, reference data (sometimes referred to as structural data) and supplemental master data (sometimes referred to as conditional data). However, it is master data that is often at the heart of it all, exposing vulnerabilities in operations, analytics and compliance on a regular basis. The first step in getting an organization behind a data governance movement is clarifying the differences in types of data and generating feedback/consensus on the area(s) of need. Master data governance offers the greatest number of avenues to generate results, so this paper will focus on data governance for master data.
However, the skills necessary to get a data governance program off the ground are non-technical. The components in the previous chart require “soft skill” activities to provide definition, organization and policy development. Recruit team members who are skilled in business analysis, communication and planning to review these areas for the level of need and value. Stay focused on building a unique picture of data governance that your company needs. That picture will likely be a snapshot of business excellence, and expectations you set must be aligned with the business initiatives ahead.
Companies in the early process of defining data governance, either for a new domain or for the company as a whole, need to stay focused on delivering value with analyst skills that the team already possesses. Therefore, the first iteration of data governance and the tools used for establishing stewardship should be low on the technology spectrum.
The main reasons for this are as follows:
Information Value Management is particularly well-suited for new data governance programs because of it's robust data governance, policy and stewardship management capabilities.
David Woods, Partner
The cycle of data governance can sometimes be a challenging process to set in motion. The concept of data governance doesn’t come naturally to business teams that are frustrated with data, and IT advocates typically step into the picture armed with data quality reports that imply inadequate governance.
The best way to start educating an organization about data governance is by focusing on the business objectives ahead, rather than current or past failures. This is best executed when some background work could be referenced anecdotally as a weak point in operations, analytics or compliance. The data governance cycle is iterative, and a set script may be hard to stick to, as variations in the sequence are expected among projects. Organizations should focus on two levels when initiating the data governance cycle: master data management need and business case.
The master data management need is composed of knowing the answers to the following questions:
The business case does not always focus on an ROI calculation. Heavy calculations of ROI can be distracting and cause inflation of expectations and project scope. Organizations in the early stages of building data governance competencies should take manageable chunks of opportunity with obvious rewards. These obvious rewards are the three objectives discussed earlier: order, efficiency and control.
Eventually, the business case will require all of the following to be addressed at a summary level:
By focusing on what data matters and why,
Data governance is a requirement for succeeding in today's competitive landscape. What else can it help you accomplish?