How to develop a data governance framework

A data governance framework is a critical part of any analytics program.

Good data governance guidelines help organizations stay compliant with regulations while also giving employees a set of guardrails to securely, safely and confidently work with data to drive the decision-making process.

Developing a data governance framework that protects the organization while simultaneously giving employees more access to analytics tools, however, is not simple.

Alation, founded in 2012 and based in Redwood City, Calif., is a data management vendor whose data catalog platform helps organizations manage and access their data. Recently, the vendor added a data governance application, and during a webinar on Oct. 26 provided its advice for establishing a strong data governance framework.


Before putting that framework together, it’s important that organizations understand what they have — the IT personnel to develop and oversee data governance guidelines, the potential end users of data, and the data itself — according to Ellen Hudson-Snyder, professional services principal at Alation.

“When you’re thinking about a data governance program, some of the most important things to identify are what you have in terms of assets in your organization, particularly from a people and process perspective,” she said.

Meanwhile, when developing a set of data governance guidelines, it’s important that organizations not try to achieve too much right at the start.

Alation's Matt Sullivant (top), Myles Suer (middle) and Ellen Hudson-Snyder.
Alation’s Matt Sullivant (top), Myles Suer (middle) and Ellen Hudson-Snyder discuss data governance during a webinar.

Rather than begin with a set of strict guidelines that launch across the entire organization at once, they should start with a framework applied to a certain data set that will be used by a specific department for a specific analytics project and then build out from there, according to Matt Sullivant, principal product manager at Alation.

“As you’re establishing your framework and trying to figure out what you want from the data governance, you have milestones along the way,” he said. “You start off small with a small set of data and a small set of policies and then eventually you mature out to more robust processes and tackle additional data domains.”

Sullivant added that by starting small, there’s a better chance of success, and by showing success on a small scale there’s a better chance of both organizational leaders and potential end users of data seeing the value of a data governance framework.

“A lot of quick wins show the value of a data governance program, and then you can expand from there,” he said.

The seven steps

According to Alation, there are seven key steps to building a successful data governance framework:

  • Establish a mission and vision and create a set of policies, standards and glossaries.
  • Populate a data catalog with metadata that shows data lineage and analyze that metadata to discover what data is most popular and who are the top users of data.
  • Recognize and assign data stewards and empower those stewards to govern the organization’s data.
  • Curate data assets by describing different data sets and applying quality flags to the data sets so users can easily find the data they’ll find most useful.
  • Apply policies and controls so that not all data can be accessed by everyone within an organization and organizations can remain compliant with applicable regulations.
  • Drive community and collaboration to promote trusted data use.
  • Monitor and measure the entire data governance framework to determine policy conformance, create curation analysis, measure the usage and creation of data assets, and determine the quality of data.

“If you’re going to really do a process, you need to find out where the gaps are and where you need to make course corrections,” Myles Suer, director of solutions marketing at Alation, said regarding the final step. “That will feed back into the framework, and with machine learning and augmented intelligence you’ll end up with a situation where [the framework] gets smarter over time and takes less effort.”

He added that data governance is an ongoing process and data governance guidelines should evolve rather than be set in place one time and never revisited.

“Data governance can never be a one-and-done,” Suer said. “Data has to constantly be worked to stay relevant.”

You start off small with a small set of data and a small set of policies and then eventually you mature out to more robust processes and tackle additional data domains.
Matt SullivantPrincipal product manager, Alation


Ultimately, the goal of a data governance framework is to provide value.

It should prevent organizations from running afoul of regulatory agencies, thereby saving them from being fined. It should fuel more informed decisions that result in higher revenue, and it should automate mundane tasks to free up data workers to do deeper analysis.

“You can unlock [employees’] productivity, but also make sure they’re not accessing data that they shouldn’t,” Sullivant said. “Governance makes you sleep easier at night, but it doesn’t kill productivity.”

While Alation doesn’t have industrywide statistics on the value organizations can derive from the implementation of a data governance framework, Suer shared anecdotal information from some of the vendor’s customers.

One has seen a 25% increase in data inquiries now that employees who need data can find it themselves. Data workers at another used to have to wait 24 to 48 hours to get to access to data relevant to their jobs and that’s now down to five minutes, while data workers at still another used to have to wait one to two weeks to get needed data and that’s been reduced to a matter of hours.

“Data governance is sometimes viewed as a gate,” Hudson-Snyder said. “By enabling that, by providing people those guidelines and providing ways to monitor [activity], it helps remove governance as a gate.”