The quest for Data-Driven Decision Making in Higher Education
By Howard Miller, Chief Information Officer, UCLA Anderson School of Management
There’s a lot of recent rhetoric and postulation around the subject of data. Some industry experts have even stated that data is the new oil. It’s known in 2022 that companies like Google, Amazon, etc., collect a plethora of data about us as individuals. In many ways, higher education is really not that much different. We collect data on students, alumni, and employers as several prominent examples. Additionally, as IT, we have data in our ticketing systems on faculty, staff and students that we could hypothetically further analyze to understand trends and where we can improve and optimize existing processes.
I’ve been in my current role for a little more than 3 years at UCLA Anderson, and it’s been a priority since my first month or so to help my institution make better decisions with data. It sounds pretty easy in theory, given the premise above and the amount of data we already have. In fact, the data itself is the easier part of the equation, as it turns out. That’s not to suggest that the technology and approach are seamless. The technology is just the easier of the two components, which I’ll explain shortly.
Reducing technical debt and taking a cloud-first approach were also two of my primary initiatives from the start along with data analytics. We engaged with both AWS and Microsoft from the start. We knew they were well-equipped to help us explore some relatively easy and straight-forward use cases to get acclimated to build, store and analyze data in a cloud-native environment. We would have been fine with either of these two companies and their platforms regardless of the outcome of our pilot use cases. We ended up selecting Microsoft Azure since we were already primarily a Microsoft shop.
The next challenge after vendor selection was picking a dataset for which data was readily available and begin building out a new architecture and our first data warehouse. It’s been an eighteen-month labor of love with plenty of twists and turns, including changes in both our internal project team as well as managing personnel changes with our business partner. As it turns out, we now have a certified data set for our MBA admissions and a series of dashboards and visualizations built on top of Azure. We have a data lake, a data warehouse, and a primary database that all sits on top of Azure Synapse. Our visualizations are built with Power BI. We did not at the time, nor do we now, have the internal expertise or skills that we need to perform a majority of this work on our own. We eventually will. In the meantime, we rely on other business partners to help us move this initiative forward.
As we look toward what’s next and how to use these tools to better understand trends and just one sliver of the data we already have, we instead need to take a step back. The technology and the tools themselves are wonderful. As previously mentioned, the data and the technology itself, especially where we currently stand, were the easier part.
We tried to figure out what data to go after next and, in fact, received some guidance that we should just pump any dataset we could find into our ecosystem so we could potentially query it. All of the advice we were receiving became quite confusing, in fact, as to what the best approach for us to proceed might actually be. At this juncture, it really dawned on me (with some help from some very valuable strategic consultants) that we really needed to put some additional focus on our key business outcomes to move ahead. In a Jeopardy-like way, we didn’t clearly focus on what business questions we were trying to answer to figure out what data (if we didn’t already have it) could be loaded next to help us make better data-driven decisions.
We haven’t loaded any new data as a result. Instead, we are on a good path now to develop 20 or so use cases with a solid hypothesis, risks, and attributes that will help us with data science and identifying the right data to make the best decisions. As for the technology, this will be transformative for our business. Some of our peer schools are ahead of where we are and even have Chief Data Officers who only focus on dashboards, trends, etc. With data as the new oil, making keen business decisions with data is tantamount to our future success. Ecosystems such as Microsoft Azure and Power BI are the keys to that kingdom.
As mentioned, the road to this point certainly has been a bit bumpy. We still aren’t sure exactly what these new technologies will ultimately cost us on a daily basis (operating expense) once we’re fully operational. The cost of doing business in the cloud actually scares me a bit, I’ve heard too many stories about runaway queries and rogue, unbudgeted costs. As previously mentioned, we also don’t have staffed reskilled (yet) in some of the underlying technology (e.g., Power BI) or the actual data science itself to be self-sufficient. With that said, the engagement of our business stakeholders on this journey, in terms of their time, interest and participation in multiple multi-hour sessions to help us prioritize the most important business performance indicators, has been truly outstanding.