Maybe you are at the early stages of planning a data governance initiative at your company. Maybe you have already started implementing a program but progress has stalled. Perhaps you previously attempted to implement governance...but were ultimately unsuccessful.
Whatever your situation, understanding the pitfalls that have typically plagued organizations pursuing data governance will help you plan ahead to avoid them. Here are five common pitfalls you'll want to be on the lookout for.
1. Governing Data from within IT
Often, the need for data governance is first identified by the IT organization. IT, however, is generally not the primary user of the data, nor are they the creator of the data. In most cases, Business is the primary creator and user of data - and they are the one who must to fix the data when there are errors, duplications, etc. Therefore, an IT-led data governance initiative is likely to see limited success.
2. Governing Data in Silos
When data issues arise within an individual business unit or line of business, the tendency may be to address them within that unit. Implementing a data governance program confined to the individual unit may, indeed, satisfy their own, internal governance needs. But the problem arises when - and because - data is shared across different business groups. Where one group defines a given data element according to their own perspective and needs, another group may define the same element differently leading to inconsistent information across the enterprise and the potential for even more data problems.
3. Assuming Everyone Understands (and Appreciates) the Value of Data
While some stakeholders in an organization are highly involved in or at least aware of all that goes into fixing data errors and other issues and appreciate the value of that data to the organization, others may be immune to it. Only seeing the cleaned data, but not the resources consumed to get it to that state, they may have less appreciation for its value.
4. Using Meaningless Metrics
Drawing from the scenario in the prior pitfall, where one unit declares a program that results in a 30% reduction in data errors a success, another unit, unaware of the time and other resources that typically go into fixing data issues and the impact if not fixed, may view such a metric as useless (along with the investment made in the program). A metric that is meaningful
to the first group is meaningless
to the second. (Learn more about creating meaningful metrics here.)
5. Treating Data Governance as a Project
Most companies are project-driven. They identify what they want to accomplish, plan the approach, acquire funding and resources and plot a timeline with milestones. Then they execute. There is a beginning, a middle and an end. When pursuing a data governance initiative, companies will often approach it as they would any project.
However, once the data governance program is rolled out, allocated funding is used up, people shift their focus to other projects and the established policies, practices and standards governing data reach a point where they are no longer maintained or adjusted in response to organizational and business changes, the program falls apart.
The issue with treating data governance as a project is that it creates the expectation that there is a finite timeline with an endpoint, finite funding and finite participation. But data governance cannot be sustained without ongoing resources and support.
Stay tuned for a follow-up post where I'll turn to the flip side and discuss best practices that you'll want to adopt in order to avoid or overcome each of these pitfalls.