The Scope of Data Governance
Richard Hines, analytics consultant and educator, discusses the scope of data governance -- including the importance of data governance and best practices for success.
- By Upside Staff
- September 1, 2023
In this recent “Speaking of Data” podcast, Richard Hines discussed the scope of data governance. Hines is a 25-year veteran of data and analytics, holding a variety of positions as consultant and educator, as well as heading AI initiatives for a large energy company. [Editor’s note: Speaker quotations have been edited for length and clarity.]
“Data governance is more important than it ever has been,” Hines began. “Early on, people weren’t fully bought into the value of data and governance, but no one doubts the value of data anymore.” Hines went on to explain how governance helps organizations achieve that value.
“Governance helps in two ways,” he said. “There’s an offensive perspective and a defensive one.”
- Defensive: focused on keeping data safe from data breaches and compliant with state, federal, and international regulations
- Offensive: primarily deals with data monetization -- keeping costs down or driving revenue up
“Companies often focus on the defensive aspect -- with good reason,” Hines explained. “However, if they want to achieve the offensive side of that portfolio, they will need to have a very strong data asset and data governance is the process by which they get that high-quality data asset.”
However, governance programs are facing new challenges compared to where they were.
“The explosion of big data about 10 years ago really threw a wrench into most programs at the time,” Hines said. “Unstructured data, such as audio logs from call centers, can prove to be very difficult to govern.”
Another challenge, according to Hines, is that companies don’t always own their applications anymore. With software-as-a-service, organizations are essentially outsourcing their applications, which has a big impact on governance. “Sometimes changes that need to be made to an application to improve quality are very hard to get done by SaaS providers,” he explained.
Hines also brought up the impact of IoT data as part of the massive data sets organizations must contend with today. “I’ve worked with sensors for many years -- going back to the mid-2000s -- and there are a lot of quality issues there.” He added that, although some IoT solutions are good at data quality aspects such as permissible values or top and bottom thresholds, others are not nearly as good.
On the more positive side, Hines noted several recent changes that may help improve the landscape.
“Some interesting things have come along recently,” he said. “For example, we have business glossary or data dictionary vendors that we didn’t use to have. The concept has been around, but organizations had to have them custom built. Now we have vendors in the space offering solutions.” Another innovation Hines mentioned is the broadening range of ways to curate and share data, from data fabrics and data lakehouses to data marketplaces.
Ultimately, the best way, as Hines sees it, to succeed with your data governance program is to start by focusing on where your problems are and work from there.
“In the classes I teach,” he said, “people generally want to start by talking about things such as roles and responsibilities, but that’s not where to begin.” Examples of potential problems Hines mentioned were things such as a recent data breach or predictive analytics that are not delivering their expected value.
Another important starting point Hines brought up was the importance of establishing a common language across all involved parties to discuss governance. “For instance,” he said, “if you talk about something such as data stewardship, you need to be sure everyone knows what you mean by that.”
Once these initial problems are successfully addressed, confidence in the governance program grows and it can be brought to bear on a new set of problems. This success also begins to change how the organization perceives their data problems. For example, getting into a new problem can reveal that there are deeper problems with things such as business processes or applications.
[Editor’s notes: You can listen to the complete podcast on demand. Mr. Hines will be leading two sessions at TDWI’s Orlando Conference about data governance; the first focuses on techniques for data profiling, assessment, and improvement (Sunday, November 5, 2023) and the other on managing data as an asset (Monday, November 6, 2023).]