Challenges to successful data governance deployment are numerous, but can be avoided or overcome through attention to best practices and industry standards
Introduction
Corporate interest in data governance often starts with recognition within data integration projects that there is a high variability in success and that this can be attributed, partially, to a lack of standardized approaches to data management. Historically data quality from data integration projects was expected to be perfect if the application code was perfect; quality was assumed to be “baked into the bugless code.” This, however, has been shown to be a false assumption; data quality predictors are much more complex than “correct application code.”
Business rules, local assumptions, lack of universal definitions or differentiating definitions, database errors and program failures all contribute over time to data quality problems. Further, a great number of decisions come into play by many stakeholders in their roles to create, move, and integrate data, all capable of affecting data quality. Part of the solution involves the deployment of a data governance program that sets out policies, standards and even simple common practices for the delivery of data integration projects as well as effort to more effectively control transaction systems change management risks to data quality.
Although IT governance has been baked into organizations for many years, data governance is relatively new on the corporate scene. As a maturing professional practice, data governance experts are emerging with some common agreement around the definition of data governance, core elements to be governed, and the activities to be undertaken for effective data governance.
It is important to understand some common challenges to effective data governance development and deployment, and know some key issues and suggestions on how to avoid or correct them.
Careful thought and creation of elements that are tailored to an enterprise view are keys to success in a long-term data governance program.
Challenges to Successful Data Governance Deployment:
- Business Is Oversold And Under-Informed – Historically, IT organizations have fragmented the controls around data placing data responsibility in many diverse organizations: data centers, integration architecture, Software as a Service (SaaS), identity management, code management groups, and data warehousing groups to name a few. Fragmentation of organization is a risk to successful Data Governance, and must be addressed through a formal charter. At the same time, business organizations are being oversold on Data Governance as the next silver bullet. Of course, there is no silver bullet to an ideal IT department since the designs, operations and staffing skills are a complex system with many contributing cost and reliability factors.
- Organization Immaturity And Lack Of Capability – It is remarkable that Data Governance has become a hot topic in IT backrooms and yet the basic maturity necessary for the engagement with data governance practice is often lacking. How can that be when the overarching IT governance has been in place, actually for decades? Data Governance is related to IT governance but is a much different practice. IT governance can be characterized by monitoring of service and service level agreements, and financial reporting and control, and the skill sets by workers in IT governance reflect these functions. Data governance requires a skilled set of workers who understand not only their job at hand but also how their performance is related to others management of data throughout the organization. Education and training is key to developing this maturity and should not be underestimated.
- Outsourcing Operations Without Data Governance – IT governance has historically concentrated on policy and service management, a focus that has formalized the relationship between organizations and IT, particularly in an outsourced world. What exactly are you outsourcing in the data realm? Generally database administration, database server farms, data in a cloud; in short data operations. Outsourcing has been carried on without separate Data Governance in organizations for years without perceived impact, so why is it needed now? Simply, over time standards and controls have been developed individually, not enforced or documented, and the repercussions become more evident as time goes on. Data Governance can establish controls as to how outsourcing operations are to operate, guiding principles, with standard security and backup/recovery, delivered at some service level. Data Governance in outsourced operations is bundled within larger IT governance in overall approaches with application support and contracts.
- Risk Management Is Separate From Data Governance – Drivers for Data Governance do not come from operations, but usually from the growing demand for data integration. Going through the mechanics of a data integration project will undoubtedly surface data quality issues. These issues not only have to be prioritized, but a framework becomes necessary to ensure recurrences do not happen. Potentially alarming circumstances become possible such as reputation loss or violation of regulations. Risks require evaluation and risk mitigation plans are needed. This is typically the scenario where the business can draw in the necessity for Data Governance.
- Positioning Of Governance Development In An IT (or Single Business Unit) Project – A local view or a project view of Data Governance is not likely to stand the test of time. There may be processes and influences that may be important to future extensibility that are not known within projects that will be road blocks to larger integration project. For instance, a simple ‘pick’ list, although standardized across the supply chain will not hold for multi-language deployment globally. A better idea would have been to develop a code table of numbers where each language equivalent could derive a translated pick list. A Data Governance initiative should be positioned as a program at the Enterprise level to engage all stakeholders in the formation and sustainment of the policies and standards for governing data as an enterprise resource.
- Primary Sponsorship Of Data Governance By IT – The Data Governance function needs to be supported, or led, or funded by the enterprise, not by the Information Technology (IT) department or by a single business unit. Doing so requires an education effort by data management specialists; business unit leaders have little expertise in data management functions and yet they are accountable and responsible for the data. Business should be mounting data policy and compliance programs to protect and manage their data assets. IT data management functions must engage their business partners and proactively offer to support the transition to educate business in the activities and tasks needed to get and keep quality data in the corporation.
- Organizational Resistance – The most frequent reason for non-execution of new policies and standards is that the people responsible for implementation do not understand the direction and the plan or do not want them or see them as having value. While developing Data Governance programs, implementation activities and their effects must be examined regularly so that policies, delivery, execution and structure are understood by everyone in the organization. Many organizations assign Data Governance / data stewardship responsibilities to individuals who are already committed 100% or more, so nothing will be done. Directives from the executive sponsors can help in dealing with resistance. However, the program itself must leverage existing staff in a smart way; streamlining processes and methods, taking away roadblocks and offering training for leveraging learning and good approaches.
- Assumption, Assumptions – Corporations assume their data is “good”, that best practices are being followed, that definitions for data are well understood and consistent, and that everyone is working from authoritative datasets. Any and all of these assumptions may be incorrect. Investigations, measurement, and continued assessment lead to clarity and a roadmap for improvement through a Data Governance compliance program.
Conclusion
Controls developed and deployed through corporate policy, national and international and sector based standards, process and compliance measurement locally and by governance have been desperately needed in all industry sectors. However, we know from recent unfortunate world economic events that these things are rarely self enforcing. Data Governance programs can add value in these efforts, both as a control function but also as an educating function. When data is seen as a corporate asset, the Data Governance proposition has self-evident value.
