The Third Pillar of Data Culture: Data Governance

There’s an old saying that if you ask ten economists a question, you’ll get eleven opinions. The same could be said about data governance: ask ten experts to define the term, and you’ll get eleven definitions and perhaps twelve frameworks.

However it’s defined, data governance is among the hottest topics in data management. Five of the top 10 webinars hosted by Dataversity in 2020 were about data governance, including the #1 webinar, “Taking an Active Approach to Data Governance” (which, incidentally, was hosted by Alation and featured Alation customer Riot Games and data governance expert Bob Seiner, whose insights we’ll discuss later in this post).

Data governance is critical to building a data-driven organization and one of the three pillars of data culture. This is the final post in a four-part series discussing data culture. The other posts define data culture and explain why you need one, and then drill down into the first two pillars of data culture: data search & discovery and data literacy.

How to Define Data Governance

Before settling on a definition, let’s examine some of the popular definitions in use today. If you’re like me, these wordy descriptions can be hard to read (let alone parse!), so for a quantitative summary skip to the comma-counts in bold. Some popular definitions from respected sources on traditional data governance include:

• “The exercise of authority, control, and shared decision making (planning, monitoring, and enforcement) over the management of data assets.” — DAMA, a leading Data Management organization— 4 commas; 2 ‘and’s

• “The organization and implementation of policies, procedures, structure, roles, and responsibilities which outline and enforce rules of engagement, decision rights, and accountabilities for the effective management of information assets.” — John Ladley, author of the eponymous book on traditional data governance — 6 commas; 4 ‘and’s

Why Traditional Data Governance is No Longer Enough

The problem with traditional data governance is that it’s often as heavy a lift as the definition sounds. If your eyes glaze over reading those definitions, imagine how exhausting it would be to actually go through those checklists for each of tens of millions of data assets! Specifically, traditional data governance initiatives often:

• Take a boil-the-ocean approach, trying to solve every problem for every constituency

• Require the proverbial busload of consultants and large consulting budgets to implement

• Emphasize compliance at the expense of access, thereby alienating would-be data consumers

• Detach the governance system from systems used to consume data, thereby decreasing its operational relevance

• End up spinning out big-bang projects that too often spiral out of control and fail to deliver on business objectives.

• Create fight-or-flight responses, with “data owners” lashing out at those who don’t use assets as anticipated and their colleagues growing fearful of using data at all.

That’s why many organizations are not satisfied with the traditional approach to data governance and seek a different approach.

Importance of a Non-Invasive Data Governance Approach

Data governance expert Bob Seiner outlined such an approach in his book Non-Invasive Data Governance. Seiner defines data governance as, “the formal execution and enforcement of authority over the management of data” (1 ‘and’, 0 commas, 10x better!). Most notably, Seiner inverts the traditional data governance approach and explains that:

• Data governance doesn’t have to take a command-and-control approach.

• Organizations are governing data already, simply informally. All they need to do is to put structure around what they are already doing.

• Organizations don’t need to spend a lot of money to get data governance. They can improve data quality, security and risk management without the need for an expensive big-bang project.

• What organizations actually govern is data-consumer behavior, and not the data itself.

• Data governance need not be framed as a huge challenge and can be undertaken using an evolutionary, not a revolutionary, approach.

Many organizations have found a non-invasive approach to data governance costs less, happens faster, and is more effective than the traditional, big-bang methodology.

The Role of a Data Catalog in Data Governance

While data catalogs have their roots in data search & discovery, they have expanded over the past several years into data intelligence platforms that also support data governance, cloud data migration, digital transformation, and other applications.

A data catalog serves as a platform for major data governance initiatives by:

• Offering common governance functions, such as policy management, business glossaries, and workflow & approvals

• Using guided navigation to guide users to the most appropriate data assets — and warning them against using potentially problematic ones.

• Integrating with specialized governance-oriented systems, including data privacy and advanced data lineage systems

• Leveraging machine learning to improve the effectiveness of data governance programs, for example, by identifying data sets most in need of curation, as well as the stewards most qualified to do so (These consumption analytics help leaders manage the overall governance program.)

At its best, governance is actively woven into a person’s workflow. Traditional governance, by contrast, threw a heavy tome of legalese at people and demanded they memorize it and change their ways. But that is not how people work! Simply demanding that everyone exit their workflow to tick an external compliance checkbox will not make it happen, but it will piss people off. If you want rules to be followed, you’ve got to make them obvious to people where they matter, like a flashing speed-limit sign in a school zone.

In fact, navigating a data landscape legally is like driving. We need stop lights, speed signs, and warnings to guide our behavior — and keep everyone safe. Just imagine the chaos and crashes that would ensue if those signals disappeared, and to drive safely, you had to read a book and memorize every speed limit and one-way street direction in your region!

Traditional data governance failed for this reason. Active governance, by contrast, greenlights the right behavior and flashes a red warning for those in danger of crossing the line. It meets people where they are, and guides them to where they need to be — just in time.

​​How Does Data Governance Support Data Culture?

Here are just a few ways: Data Governance…

• Establishes a Common Language. Like the aforementioned apocryphal economists, ten people at a company may define one word 15 different ways. What does “profit” actually mean from a data and business perspective? A business glossary houses this key information. Active governance forms the framework by which terms and metrics are defined, creating semantic consistency, so everyone speaks the same language. This is a powerful means of aligning business and data teams around strategy and success metrics—which is critical for a data culture

• Creates Shared Processes. Sharing a common framework for “how we work with data” builds trust between people and the data they share, forming a set of cultural norms

• Elevates the Experts. In an enterprise of thousands of data users, it can be impossible to figure out or keep track of who knows what. For this reason, the loudest voice or HIPPO (highest paid person’s opinion) often determines the course of action, without consideration of relevant data. Active governance locates who should give input to key decisions by identifying experts based on usage. These folks’ expertise is valuable when it’s visible. Active data governance is about getting the people who are actually the most knowledgeable about a given dataset in touch with the folks who most need to use that data correctly — for a democratic, federated, and empowering culture.

In civil society, both the government’s laws and the population’s culture shape individuals’ behavior — reducing chaos and danger and thereby enabling progress. Similarly with data in organizations, data governance and data culture support and augment one another to tame the mess and facilitate impact.

Series Conclusion

A data culture is an organizational culture of data-driven decision-making. Nearly 80% of organizations today have a strategic initiative to become more data driven. CDOs consistently rank data culture at or near the top of their strategic priority list. In this four-part blog series, we have defined data culture, outlined benefits of building one, and then taken a deeper look into three pillars of data culture: data search & discovery, data literacy, and data governance.