Garbage in, garbage out. This motto has been true ever since punched cards and teletype terminals. Today’s sophisticated IT systems depend just as much on good quality data to bring value to their users, whether in accounting, production, or business intelligence. However, data doesn’t automatically format itself properly, any more than it proactively tells you where it’s hiding or how it should be used. No, data just is. If you want your business data to satisfy criteria of availability, usability, integrity, and security, you need a data governance strategy.
Data governance, in general, is an overarching strategy for organizations to insure the data they use is clean, accurate, usable, and secure. Data stakeholders from business units, the compliance department, and IT are best positioned to lead data governance, although the matter is important enough to warrant CEO attention too. Some organizations go as far as appointing a Data Governance Officer to take overall charge. The high-level goal is to have consistent, reliable data sets to evaluate enterprise performance and make management decisions.
Ad-hoc approaches are likely to come back to haunt you. Data governance has to become systematic, as big data multiplies in type and volume and users seek to answer more complex business questions. Typically, that means setting up standards and processes for acquiring and handling data, as well as procedures to make sure those processes are being followed. If you’re wondering whether it’s all worth it, the following five reasons may convince you.
Even business intelligence (BI) systems won’t look very smart if users cannot find the data needed to power them. In particular, self-service BI means that the data must be easy enough to locate and to use. After years of hearing about the sinfulness of organizational silos, it should be clear that even if individual departments “own” data, the governance of that data must be done in the same way across the organization. Authorization to use the data may be restricted, as in the case of sensitive customer data, but users should not ignore its existence when it could help them in their work.
Availability is also a matter of having appropriate data that is easy enough to use. With a trend nowadays to store unstructured data from different sources in non-relational databases or data lakes, it can be difficult to know what kind of data is being acquired and how to process it. Data governance is, therefore, a matter of first setting up data capture to acquire what your enterprise and its different departments need, rather than everything under the sun. Governance then also insures that data schemas are applied to organize data when it is stored, or that tools are available for users to process data, for example, to run business analytics from non-relational (NoSQL) databases.
When the CFO and the COO work from different sets of data and reach different conclusions about the same subjects, things are going to be difficult. The same is true at all other levels in an enterprise. Users must have access to consistent, reliable data so that comparisons make sense and conclusions can be checked. This is already a good reason for making sure that data governance is driven across the organization, by a team of executives, managers, and data stewards with the knowledge and authority to make sure the same rules are followed by all.
Global data governance initiatives may also grow out of attempts to improve data quality at departmental levels, where individual systems and databases were not planned for information sharing. The data governance team must deal with such situations, for instance, by harmonizing departmental information resources. Increased consistency in data means fewer arguments at executive level, less doubt about the validity of data being analyzed, and higher confidence in decision making.
The risks of data hoarding are the same as those of physical hoarding.