A campus data management process primer
- Over the course of five years, Life University transitioned from an institution with no trust in its data to an organization dependent on data
- To clean up its data, Life assigned data owners and documented data definitions
- Life created a data integrity group to increase stakeholder engagement
A conversation with John McGee, Vice President of Operations, Life University
John McGee doesn’t consider himself an IT guy. He possesses a technical background, to be sure, but after consulting with more than 1,500 organizations to help them reach their goals, he considers himself more of a “community builder.” Now, as vice president of operations for Life University, he is applying those “community building” skills to help revamp and revitalize that institution’s approach to data quality and analytics.
In this conversation with Ellucian, John shares his insights and practical techniques to help institutions thrive by using data and metrics to make accurate, more predictable decisions.
Q: Can you give us a little background on how Life University approached analytics before you arrived?
A: When I started at Life University about five years ago, they had not generated a reliable report in 10 years. They absolutely had no trust in their data. They had no concept of how to make it work, and had just abandoned it. So, when they had strategic conversations, what that meant is that they were in effect flipping a coin to make decisions. We've changed that over the last five years, and now actually are an organization that's dependent on data.
Q: How does Life University use its data differently now?
A: We obviously have informational data—such as the number of students we have in some age range, and so on. We do that just like everyone else. But we use the majority of our data now to help us run the institution, asking questions like, “Where are we going to be in a year? What if we change this situation by three percent, what does it look like?” We're using the data now to more than just inform us, but to actually guide us into where we want to end up.
Q: How did you get to that point?
A: Data is the smallest part of the institutional DNA. But each piece must have an owner. This was one of the things that institutionally we really struggled—no one owned data. Or worse, the department that was responsible for putting it in somehow owned it but didn't have any ownership of it. If you think about your enrollment department, they collect probably 80 percent of the initial data you collect on a student, yet they don't really care about anything other than whether the student got recruited and ended up in a seat. So, if that's the person who's responsible for a critical piece of information, they're probably not valuing it the way that you would value it. You have to know who owns it, the right person has to own it—or the right group—and the data must be defined.
Q: Can you explain what you mean by data definition?
A: Five years ago at Life University, I would go into a meeting and ask, “What's our enrollment?” And I literally would hear four different numbers. How did we get to four different numbers? Everyone would explain how they calculated it. So, we said, “You know what, we're not going to leave this room until we define “enrollment,” so I can have one definition that says, “How many people do we have registered attending classes at our institution? That number is X.” Or I can say, “How many people do we have registered at our institution who are degree-seeking?” Now that can be a different number, but it's defined. We always know what we're asking for.
Q: With so many people touching each piece of data, how do you keep everything in order and prevent chaos?
A: A data warehouse. People across dozens of business units enter and manage institutional data. A data warehouse is used to collect data in the most consumable manner possible. What that means is I don't want people thinking about the definition of a piece of data. I don't want anyone to think about that, and I certainly don't want them to interpret it. So, when we collect a piece of information, we create the definition of it and then we create the formula that meets that definition. And then we put it in the warehouse. So, when someone comes and pulls that piece of data, there's no interpretation. They don't have to write a formula. The data warehouse is not for bulk storage. It's designed for those critical pieces of data that you want to manipulate or manage in some way and then distribute to your end users. A data warehouse should be designed and developed to guide decision making, not just inform.
Q: What are some other tools or techniques institutions can use?
A: We also use a data definition document. Those are the rules for handling the data. Every piece of data has a line item. And our team talks about who owns the data, which department owns it, where it goes, and how to know if it's correct—I can't do that for every piece of data. If you put in an application to Life University and you put in your first name, I don't know if you're Bill or if you're Bob. But I do know that there has to be something in that field. Every line item has something that says it lives within these parameters. Sometimes we can be very specific—It has to be one of these four letters, for example. And then sometimes we have a range.
Q: How do you go about controlling your data?
A: We have a scrub module, and when, say, 350 pieces of data come out, we compare it against those definitions. And if it scrubs correctly, then it goes into the warehouse. If it doesn't scrub correctly, then we send an email over to the person whom we identify as the owner and we say, “Fix it.” Twice a day we pull every piece of data and we clean it. We now have tremendous confidence in the cleanliness of our data. It means what we intended it to mean. When we identified who owned it and what department owned it, we actually went into that department and documented the process by which it got put into the system. So, we understand how it comes to life in our organization. And where we identified conflict because the process was incorrect, we changed or challenged those processes. This is a lot more than just data.
Q: Can you expand on how you did that—how you managed to get all the stakeholders to buy in?
A: We have a data integrity group. These are the users who have a voice in keeping data integral. So, when we say, “Hey, guys, we're gonna change this piece of data, we're gonna change the process by which it's collected,” somebody's sitting downstream from that piece of data can say, “Wait a second, if you change it, it's going to mess me up." We bring those people to the table and we talk about the changes that we're going to make. The group must represent the entire institution. These are people who are decision makers in those departments.
Transparency is paramount. In our data integrity group, we have an unwritten rule: if there’s going to be a major change, we'll meet three times before we actually implement the change. We’re going to give everybody three bites at the apple to understand what we're saying before we actually make the change.
Q: Since you’ve improved the quality of data at Life University, what results have you seen?
A: The most gratifying result has been that we now can’t get data requests filled quick enough. We are building dashboards and data models constantly. Professionally speaking, the best part has been our ability to build predictive tools that allows the institution to look into the future—not just predict what might happen, but to build tools that allow us to tweak data assumptions to build the future we desire on paper, then translate that to deliverables.
For instance, if I want to increase enrollment by 7 percent over the next five years, then we need to understand how retention, graduation, and recruitment work together to give us that 7 percent increase. Our model may focus entirely on recruiting more students, or reducing retention, or any combination of the two. By manipulating the degree of change in each of those areas we can develop the balance that makes sense for us. We can then create a data model that aligns resources, time, and people needed to deploy the solution.
To learn more about the benefits of analytics and how to implement a program at your institution, visit our analytics-driven campus web page.