Over the past few years, machine learning has transformed everything from movie watching habits to personal transportation. In student recruitment, advanced enrollment management teams are now benefiting from this technology as well.
Elizabeth Kee, a senior analyst with EAB Enrollment Services, sat down with Dana Strait, principal for EAB New Product Development, to learn about how analytics are transforming the future of enrollment management.
Elizabeth Kee: What are enrollment managers most concerned about right now?
Dana Strait: Managing the recruitment funnel is becoming harder and harder. Many of the metrics that enrollment managers have historically used to predict and manage yield are either no longer available or are becoming less reliable. When FAFSA rankings were discontinued, many schools saw their ability to predict yield diminish substantially.
In addition, campus visits, historically one of the top indicators of student interest, are less common among first-generation students. In order to predict the behaviors of fast-growing, less-traditional segments, enrollment managers need to be monitoring a wider range of engagement variables. On top of that, yield rates have been consistently declining across all segments as students have started applying to more schools.
The lack of predictability and declining yield rates have made it considerably harder for enrollment managers to meet enrollment targets in recent years. About one in three institutions missed both enrollment and net tuition revenue targets in 2016. A few years ago, enrollment managers used to be able to effectively manage yield outcomes with a handful of standard yield metrics. Now, if you’re just relying those same metrics, you may feel like you’re flying blind.
EK: What changes when you bring predictive analytics into the picture?
DS: Advanced predictive analytics can help enrollment managers gain insight into student behavior. With large amounts of data and advanced analytic techniques, enrollment managers can ask and answer critical questions. For example, EAB recently worked with a large private university in the Midwest that had significantly increased their admit volume in order to meet aggressive enrollment growth goals. Because of the dramatic change in their admit pool, they couldn’t accurately project enrollment based on historic yield data.
More on predictive analytics
We partnered with them to build an affinity model based on student interaction and behavioral data in order to understand if the larger admit pool was as engaged as in the past. We used the previous year’s engagement and enrollment data to train the model and then scored the current year’s admit pool. We found that, compared to the previous year, the current admit pool was significantly more engaged, meaning that they were more likely to yield.
Based on that insight, the university took two key actions: They stopped vigorously admitting students and they started triaging counselor outreach to the most engaged students. The school ended up bringing in their largest class ever, while maintaining student ACT score and increasing net tuition revenue by almost seven million dollars.
EK: So what’s stopping this type of advanced analytics from being used broadly?
DS: It’s not easy for most schools to pull off, for three main reasons.
First, schools’ data tend to be isolated, both because data from different divisions is stored in inconsistent formats and because of a lack of automated connections and query ability. Few schools have existing infrastructures for connecting financial aid, CRM, and SIS data in an automated way.
Second, turning data into insight is no easy task. Advanced analyses require specialized knowledge, and with the current, expensive market for data science expertise, many schools don’t have the resources to staff those positions in-house.
Third, even when you have insights from the data, it can be extremely difficult to know how to act on them in order to produce your desired outcomes—and to push those “prescriptive” recommendations out to the entire enrollment team.
EK: If a school is able to overcome those obstacles, what’s the biggest benefit they can expect to gain from advanced analytics?
DS: Advanced analytics can help you flag potential problems earlier, allowing you to course-correct before it’s too late. For example, let’s say you’re an enrollment manager trying to project May 1 deposit numbers a few months in advance. Just looking at basic enrollment metrics compared to last year, like admits and year-to-date deposits, you might see that your volumes have increased.
While that looks good, it doesn’t necessarily mean that you’ll meet your target deposit volume come May 1. There are a lot of reasons that could be the case—maybe the makeup of your admit pool is subtly but significantly different from last year and you’re drawing more heavily from ZIP codes that historically have had lower yield rates. Advanced analytics tools can help you identify nuanced trends in the data that you wouldn’t have otherwise been able to see. They can tell you that, while the basic overall metrics look good compared to last year, the behaviors and makeup of this particular admit pool indicate that you’re unlikely to hit your final target.
Once you add class shaping goals into the mix, advanced analytics become even more critical—they can help enrollment managers simultaneously monitor numerous dimensions of the incoming class, including academic quality, specific program enrollment, revenue, and diversity.
EK: Once enrollment leaders know that there’s a problem on the horizon, how should they determine the best way to correct it, given limited resources and so many possible plans of attack?
DS: Advanced analytics can help enrollment managers identify the underlying reasons for an issue they see. Those insights can then be used to generate effective solutions that directly address the problem.
For example, an enrollment manager could isolate the segments in question and drill down into specific behaviors, such as web activity and most-visited web pages. If you did that, you might see that a certain target segment was spending a lot of time on the net price calculator and aid website, which would let you know that affordability was a major concern for them. In response, you could launch a multi-channel affordability campaign. To ensure that your campaign was most impactful, you could prioritize your outreach according to the likelihood to enroll scores of each student within the audience, as identified by the algorithm. You could limit your most expensive outreach, like mailers, to the students who were identified as somewhat or fairly likely to enroll.
Advanced analytics can enable enrollment managers to identify the problem, isolate the root cause, and then quickly pivot their strategy using a data-driven, efficient approach. While there’s an upfront cost associated with these analytics capabilities, the increase in efficiency pays off the initial investment in spades—they’re a truly powerful tool in enrollment management.