Inger Bergom, Senior Data Analyst at Tufts University, spoke with Rana Glasgal, Associate Vice Provost for Institutional Research and Decision Support, to discuss Northeastern University's fast-growing Tableau environment for supporting data visualization and analysis throughout the university.
How has the Tableau environment at Northeastern evolved over time?
I can start out by saying that there are lots of universities with very good use of Tableau, and certainly Northeastern is not unique in that way. But I think that we have come a long way in a short time. When I got to Northeastern in 2016, people were using Tableau mostly as a desktop tool. To share something, you would create a packaged workbook or send it to somebody who had Tableau Reader. I found it was difficult to get people to download and install Tableau Reader because nobody likes to install more stuff on their computers.
Our enrollment management team had a small Tableau Server installation that was based on named users, so you can only have a certain number of users, and they had to be specific people. I knew that there was a lot of demand for sharing, and you can really make the case that it's much more secure if people go to the server to get their Tableau document because if you just attach a Tableau workbook to an email somebody could forward it, and you can lose track of it and the data gets stale or out-of-date.
In 2017, I managed to convince our senior leaders that we'd be much better served by a university-wide Tableau server license. Once we did that, the floodgates were opened, and a lot more people started using Tableau as their data visualization tool of choice and publishing things to the server. Use has grown so much that we are actually expanding from one eight-core server to three eight-core servers. I think it's been a really big success, and a lot of that is just grassroots analysts learning how to use it.
What is a recent accomplishment at Northeastern related to the implementation of Tableau?
There are so many. Certainly, the factbook was a big accomplishment, to have that online and interactive. When you move your factbook into Tableau then it's interactive, and you can have breakdowns by college or gender, whatever it is you need. Everybody can find what they want. That was my main use case, which I think is a very compelling use case for people. For the factbook, the whole active directory can see it, so all of our faculty and staff have access to it. Normally, you have more restricted security for dashboards, but in that particular case everybody can use it. Since we're not limited by the number of users, that opened up a lot for us.
We've done some diversity work in Tableau, so that's also very interactive. Dashboards include data on faculty, staff, students, graduation rates, and retention rates by gender, race, and ethnicity. We've also done some diversity benchmarking in Tableau and more analytical work as well. Also, we built dashboards for the deans so they can see their colleges’ data and certain metrics that they want to see. Being able to maintain that in Tableau instead of doing it in Excel is much faster. There were some upfront costs to creating the deans' dashboards, but now it's less maintenance than it was when we were trying to do things in Excel.
What are the most pressing challenges related to data governance within the Tableau environment at Northeastern?
We have a really well established and thorough data governance operation here that predates me. Because we have a very well-developed data governance organization, we can ask our data stewards and data custodians and subgroups of our data stewards and data custodians to work on specific issues related to data governance. How does Tableau fit into that? Data governance provides the people who are creating the content with appropriate access to data. The same restrictions apply if you're creating content now with Tableau.
We've had growing pains. If you create really cool stuff like a factbook or a dashboard for deans, which is easier to do in Tableau, somebody still has to be responsible for maintaining it. You're going to be updating it once a quarter or once a year, like our factbook. There's still maintenance involved, and if you multiply that by 1,500 or however many dashboards you have, that's a lot of maintenance! Some of these dashboards are quite complex and they have data coming in from all over the place, so making that transition is difficult. But, if you keep your IR analysts always responsible for all of that upkeep, then they don't have time to do the new analyses. So that is a challenge for us: How do we free analysts from maintaining all the great dashboards so they can create new ones? We have had challenges with being victims of our own success.
Another challenge is making sure that best practices are being applied across 39 sites and 85 administrators. So that's a challenge when you grow: you have to still ensure quality without losing flexibility. That's why we're going to instill a little more accountability for our site administrators, for instance, to do a yearly audit to check-in on what's been published, making sure it's still working and making sure the right group has access to view it. Keeping permissions tight and up to date is critical.
What advice do you have for someone who wants to build a similar Tableau environment?
You definitely don't want to go it alone. Setting up a university-wide Tableau environment might require having a conversation to say, "Let me tell you why we need Tableau and why we need a good, solid, reliable infrastructure." You want to bring in a partner. It may be your BI team or your CIO, depending on where your relationships are, to talk about it and how it actually might take some burden off of their team. It can be a real time saver for the IT team if they don't have to create a report for every little thing that people need. People in IR tend to realize the value of data storytelling and data visualization, but you can't blindside your IT folks by making this decision in the IR department alone.
I think having a culture where data matters is incredibly important. It helps to have a culture of using data and having data literacy among your decision makers. I think a good relationship with your IT organization is also important. They're really good partners, so we're very fortunate. I wouldn't have gotten our first Tableau server back in 2017 without the support of our IT folks and the CIO at the time. I think that was a good start, and it set the tone for continued partnership between the two organizations.