Operationalizing Data Governance
The Data Governance Silo
Breaking through silos within organizations is often a primary focus of modern data governance initiatives. This is a sensible objective, as it’s only when we collaborate across disciplines and functions that we can begin to gather a holistic view of our data and business landscapes, and ultimately make better-informed decisions. However, all too often, there is a lack of proper planning for this goal, and the governance team who started out with the best intentions ends up creating a new silo of their own as a result.
The “governance silo” can occur quite naturally if not carefully avoided, as the data stewards documenting and defining what they want to govern do not necessarily extend their reach and their goals beyond this area. If the governance team determines the key data elements (KDEs), documents their definitions, creates committees of stakeholders and sets up workflows to ensure these definitions are kept up to date, have they really implemented meaningful change within the organization? Generally, business users living their day-to-day lives do not go to a glossary tool to look up a regularly used concept—so what is the impact of this glossary?
We have already seen governance solutions mature to an extent. For instance, it’s now a standard requirement to include some kind of discovery capability as part of a governance solution. Of course, levels of automation and scalability vary by the type of solution (please click here to see more about the Informatica approach to automation and scale), but another now-common expectation is understanding that the physical location of data should not be an entirely manual process.
Another sign that governance has matured: we no longer want a single-faceted view, but instead want to move beyond the glossary of terms and their physical locations to understand the usage of these concepts, how they are used, for what purpose, whether or not this is aligned with company policy, etc. It’s important to accommodate this requirement for different viewpoints, without risking the consistency that enterprise governance models require to be successful (learn more here about Informatica’s approach to agile DG without overzealous customization).
And even when progress has been made, governance solutions with scanning capabilities and multiple viewpoints still must go beyond the documentation phase. To truly effect change, we need to go one step further into operationalizing governance.
Operationalizing governance is about more than just gaining understanding. It is about using that understanding to impact the way things are done. For instance, gathering together the appropriate stakeholders who can:
- Discover lineage changes and act upon them
- Execute data quality rules and step in when issues arise
- Monitor the privacy status of sensitive data and intervene as necessary
- Promote and enable access to governed data assets
Scanning your technical landscape once does not mean you can consider this box checked. Operational governance means being able to keep track of changes in metadata and ensure that you are able to take appropriate actions. If there is a change in a database—for instance, the deletion of a certain field—this could have ramifications for downstream systems and processes. It is critical that these kinds of changes can be detected by your discovery capability so that stakeholders can take the appropriate actions. Going beyond the data governance silo with intelligent integrations makes this possible.
Beyond just knowing what data you have and where it has come from, it is also critical for any business to understand the quality of their data. This will allow end users to make critical decisions like sourcing data from the best possible system, flagging inconsistencies and data issues, and remediating problems as they arise. Although data governance and data quality have always been interlinked, recent developments now allow us to operationalize governance and automate the creation of quality rules. Using natural language processing techniques, it is possible to put the creation of data quality rules into the hand of the business users who know best what those data quality rules should be and to then automate the creation of these rules across the enterprise (to see a demonstration of this, please click here).
Privacy is another key concern for most governance teams, particularly with the rise of regulations around personal data. It is no longer enough to know just where the personal data that your company handles is stored: it is now also essential that you are able to justify that storage policy, and to be able to show what kind of safeguards are in place for different levels of sensitive data. A key component of mature governance programs is being able to not only track the security of your data, but to alert relevant stakeholders when that security is insufficient is (to see how privacy information can be understood through the lens of governance, click here).
Shopping for Data
Another natural extension of governance is the desire to promote the use of the data that is well-governed, of good quality, and adequately protected. Analytics, always a key focus for companies across industries, has a common problem with discrepancies between results from different projects. If data scientists pose the same questions, but have used different data to establish their answers, it’s not surprising that they will get varied results. So the ability to direct your data consumers to the best data available for their particular purpose becomes the cherry on top of a well-governed estate. Once data is understood within context, you will have all the necessary information to guide those who want access to it, and be able to guide them through the process of requesting that access with permission from the appropriate stakeholders.
The use cases above relate to doing more than just documenting and defining with the understanding you gain from governance. The market has plenty of tools that will address these concerns individually, helping you to monitor governance, metadata changes, quality, privacy, or to shop for data—but these capabilities do not exist independently of one another. The end goal, as we have acknowledged, is to break down silos, so rather than addressing each of these questions in isolation, doesn’t it make sense to try to bring them all together? This is not to say that one tool will (or should) meet all these requirements, as it is not worth sacrificing depth for breadth, but intelligent integrations between these tools can allow you to pull together just enough of the relevant information to allow governance to go beyond the documentation silo and effect real change throughout an organization.
To find out more about how our automation and scale capabilities can reduce manual effort in your organization and introduce greater agility, register for the webinar, “Informatica Intelligent Differentiators Series: Scale and Automation,” or visit our Data Governance Standards: 4 Intelligent Differentiators web page.