Sergey Nivens - Fotolia
- Scott Robinson, Kentucky Farm Bureau
Business intelligence is taking on an increasingly critical role in the enterprise because it increases the accuracy of conventional processes and delivers game-changing strategic insights across the board. The management of data that feeds into the BI processes is equally important. Systems that feed into BI processes are more complex than conventional systems, which can make data governance even more difficult and potentially overwhelming. For this reason, BI data governance is more important than ever.
Tougher than tough
BI is generally culled from large bodies of data that are deep in history, broad in scope and staged in a data warehouse or similar structure optimized for mining. To make that data useful, it has to be managed in much the same way as conventional data -- and then some. Therefore, it's important to keep in mind these BI data governance areas of concern:
- There will likely be more dependencies than usual, and they likely will be more complex. BI data tends to be multi-sourced, with one body of data supporting other bodies of data.
- Dirty and incomplete data is always a concern in any aggregate data structure. Complex dependencies make the cleaning and completion of data in a BI source all the more important. Downline trending patterns discovered in BI processes are less reliable when source data is spotty, and gaps render good associations unusable.
- A strong, multi-sourced BI process is likely to contain unstructured data from nonconventional sources. Social media data, for example, is playing an increasingly important role in marketing and sales intelligence. Accommodating this data within BI processes presents new and different challenges.
- Once you've cleaned dirty data and figured out what to do with the unstructured stuff, the challenge of data integration follows. Integration requires putting the resulting data into a form that multiple business processes can easily and unambiguously exploit.
Failure, or even suboptimal execution, in any one of these areas can result in a BI data environment that delivers spotty or skewed results. Typically, those weaknesses will be revealed across many of the BI processes being supported. Since the product of a BI process is to see probability-based trends, predictions or outcomes, poor data at any one stage of the process not only adds errors into the result, but also compounds those errors.
Lessons from the public sector
BI data governance is essential and generally needs to go beyond the usual call of duty in its scope and diligence. Fortunately, there's a good model to follow -- open data initiatives in the public sector.
The municipal governments of Denver; Louisville, Ky.; and Boston are just three of several U.S. cities that have begun offering publicly owned data sets to local citizens through cloud deployments. These initiatives are very successful, allowing citizens and companies to download large data sets for many purposes -- from property reclamation to more efficient transportation scheduling. And these rigorous governance requirements for this unusual type of data are applicable to BI practices in the private sector.
Publicly owned data has to be integrated to offer optimum utility for the greatest number of users -- a useful principle for BI in general, but essential for open data. This process requires trimming away columns of data that, combined with other columns, can violate privacy or security standards. Also required is the standardization and top-down structuring of metadata across the portal to create associations between data sets where users might find them advantageous. That may involve extra work, but it's a practice that BI in the private sector should emulate.
Government-style compliance and oversight procedures include several critical principles that can be applied to business BI practices.
A diverse governance team. Governance requires IT participation, upper management representation and hands-on users. Some team members should be permanent, while others can be ad hoc, participating in discussions when they're able to provide advice on specific process outcomes.
Business-goal alignment. This process requires carefully monitored continuity. In municipal government, keeping the team's eye on the ball for the public good is the highest priority that would well serve private enterprises providing services to their customers.
Data journalism and end-user support. In open data deployments, governance teams in government and the private sector must ensure that there are a sufficient number of user case histories to support the best use of the data in analytics and reporting.
Continuous process improvement. As open data usage increases, governance teams must constantly refine and update their data management practices.
When it comes to governance of BI data, management in the public sector strives to do more with less, reach maximum efficiency and continuously improve. There can be no better formula for the private enterprises seeking to put BI to work.
The data governance process is being refined
Creating an efficient data governance framework is easier than it seems
Data stewards hold the key to data governance's success
Dig Deeper on Data governance strategy
7 steps to create a modern business intelligence strategy
IT managers look to deploy BI, analytics in the cloud
BI self-service tools need middle ground on use, governance
Data governance process taxed by self-service BI, big data