How to get started with data governance
Among all the data-related opportunities and concerns of this year, one that is demanding increased attention is information governance. Governance is a concern because the growing business appetite for data and analytics puts pressure on systems to deliver data rapidly, whether it is well governed or not, and because proliferation of complex, diverse data from multiple sources makes governance harder to achieve. But this challenge is also an opportunity because proper governance can increase confidence in data, thus making it more useful and valuable to the business.
In its predictions for 2015, Gartner, Inc. comments that “The rise of data discovery, access to multistructured data, data preparation tools and smart capabilities will further democratize access to analytics and stress the need for governance.” With more and more data coming from external sources, it’s easy to lose sight of who created it, how it has changed and what level of confidence in the data is appropriate.
If the need for information governance is growing, what does that actually mean that an organization should do? And what does governing data entail?
Forrester, Inc. refers to five key data governance domains: quality, reference, life-cycle management, security/privacy and metadata management. Other definitions include master data management, and a closely related domain is data integration. However it is defined, data governance addresses a broad spectrum of needs related to controlling and protecting data, checking its adherence to standards and increasing its availability and usability to the organization. So it’s important for the organization itself to consider questions like:
- What are our data standards?
- To whom should the data be available?
- For what purposes should it be usable?
- Do all potential data users share the same definition of the data?
- What is the value of data to our organization?
Since these are not questions with easy answers, and since there are typically multiple data stakeholders across multiple divisions and departments, data governance is an organizational issue as well as a process and technology issue. But the breadth of the issue doesn’t mean that it’s too much to tackle when so many other issues are also on the table. What it does mean is that a methodical approach can help any organization get started down a path to governance that has been followed by other organizations with success.
Some have started by addressing data privacy and protection, others by improving data quality and qualification for its purpose, others by managing the data lifecycle and others by tackling whatever was the burning issue for a particular industry at a point in time. They have moved ahead to the next issue at a measured pace—not swallowing the elephant but still making progress and increasing the availability, protection and value of the data along the way.
Some companies have chosen to focus on defining a common business language as an early step toward information governance. Anyone who has ever asked six people in the company to define “customer” can appreciate why that basic step is worth taking. Without shared definitions, it’s hard for anyone to understand the best reports or analytics on customer behavior. With 17 different definitions of “employee”—an instance we have seen in real organizations—it’s hard to be sure that employment practices are appropriate for all.
The IBM InfoSphere Information Governance Catalog encourages a standardized approach to discovering your IT assets and defining a common business language. It can help you create a well-documented information blueprint for aligning business requirements and reference architectures, establish a common business language shared by LOB and IT, and manage and explore data lineage to support compliance.
A catalog certainly isn’t the only way to get started with data governance, but it’s a good step to consider, whether it’s your first or a step along the way, as you plan your data governance journey. Pack your bags and get started!
For further reading: