If data isn't managed well, your organization could be at significant risk. This list will help you spot areas where your data governance may fall short.
One thing we all can agree on is that hindsight is always very clear. When we're discussing and reviewing issues in technology systems and failed projects, we can usually make a data path to the problem. Data governance issues can exist in organizations of any size, but if you don't know much about data governance, that's an indication of a potential issue.
I recently attended Enterprise Data World, a conference around topics including Big Data. During that event, it was pretty clear that there is a priority to ensure that data of all sizes and profiles are managed well for organizations big and small. I came up with this list of 10 indications that you may have a data governance issue; hopefully you can hedge off any issue in your organization!
1: You have pockets of adoption
When it comes to data and its access, pockets of adoption may not cut it. If you hear this type of conversation, keep in mind that it takes only one problem spot to cause a data handling issue. Adopting data governance has to include the entire cycle and scope of the organization. The reality is that it takes just one system to improperly handle a piece of sensitive data and cause an issue.
2: There is no internal data dictionary or business glossary
This may seem oddly familiar to the TLA (three-letter acronym) learning process associated with new hires; the same applies to data. The notion of a data dictionary is usually implemented on database systems and enterprise applications. But as many systems are involved in today's complex web of IT systems, it becomes a priority to ensure that all data dictionaries and business glossaries are the same. How many times have you been in a situation where there may be a field or term in one department that is the same as another department? That's a different issue; but the principle is the same: It's a good idea to have one data dictionary for the organization and ensure that applications and their data profiles are modeled around that dictionary.
3: There is no data steward
If you can't answer the question of who (or what group) in an organization is in charge of the administration of the data in regard to definitions and ensuring that the data quality is correct, that's a sign of a data governance issue. The data steward isn't usually the person or group that purchases the hardware or does the backups, but instead is the one who is in charge of making sure that the right data is presented in the right places, that the data is correct, and that its lifecycle is managed correctly. Part of the lifecycle includes data archive and eventual deletion. Who addresses these topics in your organization?
4: There are multiple data stewards
Somewhat the opposite of the point above, it is altogether possible that two groups are staking claim to the role of being the data steward. While this is more an organizational problem, it may increase the risk of things falling through the cracks or being incorrectly governed. This is can be especially problematic in a situation involving the deletion of data that one group needs and another group doesn't need. Don't let that happen to your data!
5: Multiple systems access governed data
We live in a connected world; I don't have to tell you that. Interoperable systems play a big part of our application and infrastructure profiles today. While we do good things like use strong passwords and common authentication models, we may do bad things like not let all steps of the process take the requisite care of the data. This can include storage systems, file share permissions, lack of encryption in connected systems, or related technologies that we may not even think of, such as logging and command-line interfaces. This is especially relevant for administrative tools, such as remote command-line interfaces or debug logging systems for critical applications. Generically speaking, there can be logs or session data that may include credentials, data, and more kept on local PCs or other server systems.
6: Some issues are "too difficult to correct"
If it is too hard to fix a data issue, and it seems better to simply keep working around the issue, there may be big limits to that mindset. These types of technology situations can cripple business. Consider whether the business doubled or tripled. Would these workarounds still seem valid?
7: Operational limits are causing failures
If operational situations arise where organizations can't close the books quickly enough due to multiple systems and workarounds, data issues may be slowing the organization down. This can lead to other organizational failures around coordination, due to multiple systems in use.
To be fair, we live in a world today where organizations acquire and divest companies frequently. This organizational behavior makes these data situations more common, even if for retention and archival reasons.
8: Regulatory needs have changed
We also live in a world of constantly changing and evolving requirements for data subject to regulatory compliance. Financial services, insurance, medical data, and others know that this is a serious responsibility. If a data profile is in-scope for any regulatory or compliance requirement, it's important to know where the new boundaries are. This can mean additional costs to go through the compliance drills as well as any corrective actions -- but it's a reality of the businesses we are in.
9: Correction processes are too difficult
A sign of a mature data management empowers non-data stewards and other end users to start corrective action procedures for data. This includes incorrect classification, improper handling of certain data, and matching up like data that is duplicated. If this process is too complex and not intuitive, the users will not do it. It's that simple. The process doesn't necessarily need to be entirely completed by end users in the organization, but a work request to the data stewards of the organization to correct a situation can help greatly improve the overall data quality.
10: There's a lack of data stewardship flowHaving a clear protocol for how data will be handled by a data steward is a great starting place for effective data management. Three key areas include specifying stakeholders, enacting data stewards, and implementing technology to tend to data. A sample flowchart from Collibra shows an approach to data stewardship (Figure A).
You may not notice it at first but this flowchart includes "backward" flow items with conformance metrics and events such as violations and defects. This underscores the importance of the previous point: The company and organization as a whole must be able to implement corrections to data as it exists in the lifecycle.
Data quality is where it's at
There's no disagreement that data is a critical part of technology. The fact is, especially as data is exploding, if data isn't managed well there may be significant risk. The risks are many and you can surely identify your own. How can we make data better? Do any of these catch points exist in your organization? What other indications would you offer that may point to a data governance issue? Share your comments below.
- The 21st Century Data Center (ZDNet special report page)
- Executive Guide: The 21st century data center (free ebook)