Building an effective data governance framework
A comprehensive collection of articles, videos and more, hand-picked by our editors
About 90% of managing a data governance program involves internal communication and dealing with people issues, says Danette McGilvray, president and principal of Granite Falls Consulting Inc. in Newark, Calif. In this interview with SearchDataManagement.com, McGilvray offers advice on what to do -- and what not to do -- in setting up and leading a governance initiative. One of her top keys to success for data governance managers: making sure you know the reasons why your organization is looking to create a data governance process, so you can ensure that the program addresses those issues.
Can you briefly explain what data governance is and what the potential benefits are for organizations?
Danette McGilvray: I look at data governance as being what outlines and enforces three important things: rules of engagement, decision rights and accountabilities as it relates to data and information. And we do this by organizing and implementing the right level of policies and procedures, structures, roles and responsibilities in order to effectively manage our information assets. And the right level is an important phrase there. Because fundamentally, data governance gives us venues for interaction and communication so that we can make effective decisions [about data], identify and resolve issues, and escalate those issues when necessary.
How important is it to get business units and departments involved in data governance programs?
McGilvray: The short answer is that it's very important. The caveat I'm going to make on that is people need to be involved appropriately. What you're trying to accomplish will determine exactly who and at what level will be involved in this, but it does definitely need to include the business. It's not truly data governance unless you have representation from [the standpoint of] how data is being created, used and updated, which is on the business side. But the technology folks need to have a say in it, too, as far as how data is stored and shared and the different things that we do to maintain and dispose of data.
Should the business side take the lead on data governance, or does that depend on the specifics of the organization?
McGilvray: I would say in an ideal world, the business side would take the lead because that means they really understand how they have responsibilities around the data. However, I'm very practical. If there are people on the technology side who are aware of the need [for data governance] and are willing to step up to the plate and lead that effort, then we should go with that. But we need to bring in the business side as well, as quickly as possible.
Do you need a formal data governance council to be successful?
McGilvray: I don't know that I would say you have to have a data governance council per se. What you do need, though, is an acknowledged group whose responsibility includes making the decisions, setting the priorities and resolving the issues related to the organization's data and information. If it ends up being a formal data governance council, that's fine. If you have a decision-making body that is already set up and that makes similar kinds of decisions, you can take advantage of that and whatever that group is called is fine.
What's the right size for a data governance council or a less formal group that's involved in a governance initiative? Can it become unwieldy if it gets too large?
McGilvray: Yes. And that's where people become afraid of data governance, because they think, "Oh no, it's just going to add another layer of bureaucracy and it will slow down what we're trying to do." When in reality, if we've done it right, it's going to help us be able to speed up our decision making and the decisions we make are going to be more solid and less likely to have ill effects in the company.
The number of people it takes really depends on how you're looking at the representation within your company. You could have a data governance council that might have 10 people. But I've also seen companies where they want representatives from all the different lines of business, and that can turn into 35, 45, 50 people. I think the best way to handle that is having a good process for setting an agenda. Really doing a good job at managing the agenda helps us be able to manage that large a number of people.
Do you find that most companies designate formal data stewards as part of their data governance efforts, or is data stewardship still an emerging concept?
McGilvray: I really like to emphasize the idea of data stewardship, which means that when data comes across your desk or is on the screen in front of you, you realize that not only am I taking care of the data for the sake of the transaction or the report or whatever I'm doing right now in my area, but I'm also responsible for making sure I take care of it on behalf of the rest of the company. That's a stewardship attitude. And everyone needs to have that attitude, really.
Having an actual title called steward, though, has evolved into something that's very common. What I see differing sometimes is at what level a steward is. An enterprise data steward might have the responsibility and authority to make decisions about data, but they might not be the people doing the hands-on work with the data. In some other companies, the people doing the hands-on work are considered the data stewards. And sometimes, people will use other words besides steward. To make it work, it's really important to use titles that will be accepted within your particular culture and environment.
What's the first thing that someone tasked with setting up and managing a data governance program should do?
McGilvray: The very first thing is to find out why they're being tasked to do that. One of the biggest reasons of failure is if people think, "I've heard about data governance. I don't know what it is, but we've got to have it." And I have heard of a few places that have set up data governance councils and then they're sitting there wondering what they're supposed to do. If someone is asking me to do data governance, I want to find out what are the issues that are causing us to need data governance. That's what is going to drive where you first focus your efforts, so you're turning out something real.
What are some other keys to success or best practices for managing a data governance program?
McGilvray: One of the biggest things is being able to articulate what data governance is to your organization. People often are surprised at how much communication is involved in making data governance work. It's an integral part of it, but people go into projects like this saying, "Yeah, yeah, I know. Yeah, I've heard I need to do communication." And they look at communication as being something that's not real work. But what they don't understand is the rest of it won't work unless we've got the right kind of communication.
There's also getting the data stewards to agree to the responsibilities that they would have, and getting their managers to agree to have them spend time [on a governance program]. Data governance is all about people, which is why the communication piece is so important. And don't give up. A governance program won't be built in a day or a month. It takes time and effort, so keep moving forward and be persistent.
On the flip side, can you talk about some of the common mistakes that organizations and data governance teams make on governance programs?
McGilvray: Not communicating and not taking the time that it takes to get buy-in. People need to be clear on what it is that you want them to do. "Am I there to make a decision about data? Am I taking some action? How much time do you think this is going to take?" And the fact of the matter is, people are already making decisions every day about data. They're often making them without full awareness of all the impacts and all the inputs they should have; what we're trying to do is make that whole process more effective. But a mistake would be to think that nobody is doing this.
Another misstep is not being aware of or preparing to handle change -- doing organizational change management. Data governance will initiate change, which often causes people to be fearful. You know what they say: "I'm OK with change -- just don't do anything different!" The easy answer to that one is make it real for them: How are we going to make your life better?
What about measuring the results or the impact of data governance programs? Is it important to prove their value?
McGilvray: Right off of the top, people need to come up with a way to be able to track the issues that they're addressing: How many issues have we resolved this month? What are we working on? But on top of that, what we really want to be able to see is the content of those issues: What did we do and what was the result of that? If your data governance program is focusing on some data quality type of work, for instance, and you put more detailed data quality metrics in place, that's a way to show that governance is helping increase data quality across the company.
Everybody's talking about "big data" these days. What kind of impact will the growing amounts of unstructured data in particular have on governance programs and initiatives?
McGilvray: I think it actually increases the need for governance because data is data, whether it's unstructured or it's sitting in structured databases. We still need to understand who has the authority to manage that data throughout its lifecycle. And more data means that we have more decisions to make around that data.