Data Governance is a Dirty Word: Understanding Why Data Governance Has Been Given A Bad Name
Why do we manage our data? More than ever before, data consumers expect data that meets their needs…every time. These consumers include our end customers, business process owners, data scientists, and many others. Moreover, now that the volume of data stored in any enterprise is growing exponentially, the pressure is on to turn it into new business insights, drive better decisions, and use it to understand and influence customer choices. Invariably the phrase ‘data governance’ is inserted into conversations about quality data, accessible data, compliant data policies, to name a few. Data governance should go a long way to meeting the need for reliable, valid and meaningful data to support our business priorities. However, the reality of a data governance program may not be quite what we expect. There are a few things we can focus on that help to avoid the hazards that give data governance a bad name.
First, we might think about existing business processes that have well-managed or governed data. Two that come to mind are financial data management and product development (think Product Lifecycle Management). To manage cash, assets, and obligations for our enterprise, we rely on economic data. Moreover, we have processes in place to ensure its quality, validity, and security. There are similar, well-defined business processes for product, recipe, and engineering data management that also provide quality and accuracy. From this perspective, governing data is not a new thing. However, the idea of establishing an enterprise data governance program that supports multiple initiatives and business processes, that is a new thing.
Let’s look at two of the main reasons it is hard to extend the same data management knowledge and capabilities that we use for financial and product data management into an extensive, and formal, data governance program.
Too much, too fast
The question ‘why we are doing this’ should be answered early in the development of a data governance program. All too often, this does not happen, and a program emerges with vague and far-reaching goals that include a large organizational structure, but not many specifics. These large programs typically engage a significant number of business team members, have too large a scope, too few targeted business-related benefits, and drive some very negative buzz in an organization. A lack of focus is especially true of technology-centered implementations where the ‘how’ to manage the data overshadows the ‘why.’
Start small and grow
One of the realities of a data governance program is that it is an exercise in organizational change. This exercise requires time, mindfulness, planning and education. It should start with a narrowly defined scope, and a plan for communicating how, specifically, the governance program will benefit the business. To achieve these objectives, the data governance program should initially focus on the most significant business initiative or most pressing organizational challenge. A few examples are:
- Managing Personally Identifiable Information (PII), where the scope includes IT security, compliance, legal, and current applications that store the data. This effort should be a working team of fewer than ten people.
- Modernizing Customer Relationship Management (CRM), where the scope includes sales, marketing, and the experience of existing customers. This should be a small working group, of business and technical resources, focusing on specific data integration, data quality, and data access challenges.
Then, once data governance skills and capabilities are honed, we can expand the program; which segues to the second reason that data governance gets a bad reputation.
Losing the audience
Many definitions can be found for data governance, but they all might be summarized as ‘establishing data policies that meet the needs of the enterprise and executing on those policies.’ Again, think of the data management policies that govern financial data. Financial data management includes business rules that are applied when the data is captured (whether it is keyed directly or processed through an external feed), checks and balances, accountability, data reconciliation, and data audit. A small group manages the data policies, and there is enterprise-wide responsibility for executing these policies.
When defining a formal data governance program, the requirement that data policies be enterprise level can feed the perception that a large, centralized organization is required. It is not uncommon for a new organization to be established, with new team members who are developing new capabilities to support effective data security, data quality, and data provisioning. This new data governance organization will extend to business data stewards and SMEs, who are, frankly, asked to participate in something that they don’t understand, and they are being asked by a team that is in learning mode. If this extended team is not actively defining or executing data policies that make sense (i.e., provide business value), then the audience is lost. Moreover, ‘data governance’ is remembered as a wasted effort.
Building the capability
A data governance organization needs to support a set of data management objectives that are tied to business value. To do this, a small organization at the onset may work best. The staffing should focus on at least two areas.
- Organization change. This is a ‘hands-on’ initiative. The data governance team should be capable of leading the implementation of new business processes for data management and communicating their value. The intent is not just influence or evangelism, it is onboarding, educating, process documentation and improvement, to name a few. It also includes technology, but that should always be secondary to the business
- Collaboration. To identify and execute data management policies across the enterprise, the team must be able to facilitate cross-silo communication, agreement, and If the scope is well-defined and supports a key business initiative, then participation will be guaranteed. However, buy-in is not guaranteed. The data governance team must include a leader who can facilitate new communication and drive collaborative decision making to support enterprise-wide buy-in.
Avoid thinking of a traditional, top-down, organizational model, and instead think about what you want to achieve, in the context of your business-driven data needs. Moreover, realize that you are implementing new business processes. If you can focus on a scope that demonstrates business value, manages organizational change, and communicates the positive impact of what you are doing, then data governance may be a big win and not a dirty word.