It’s late September and the new academic year has settled upon us. Admission offices across the land are already busy recruiting the next set of classes, while at the same time conducting postmortems on the good and the bad of the most recent campaign. Some institutions have found the 2009 recruitment cycle to be quite unpredictable.
Recently, the Providence Journal ran a story about the difficulty some Rhode Island institutions had enrolling and managing the inflow of new students for the fall semester. This, despite the fact that nationally the number of graduating high school students remains at nearly record highs (although down slightly from the previous 2 or 3 record years). What was driving the instability? The economy, of course.
Private institutions in Rhode Island, according to the article, seem to have been hit the hardest and many were short of their goals. What struck me, however, were the tactics that were used by admission officers to stave off the potential declines. Seemingly, institutions broadly applied a combination of two tactics; either, 1) increase the amount of institutional gift aid to effectively discount the tuition, or, 2) lower admission standards to increase the eligible pool of applicants. Sometimes they did both.
While I may be over simplifying the true remedies attempted by these institutions, I can tell you from experience that the typical responses to shifts in the marketplace for many admission officers tend to be linear and broadly applied:
- When the demographic pool shrinks – buy more search names.
- If times are tough economically – increase the discount.
- When the numbers don’t come through as expected and the cause is not clear – lower admission standards.
The problem with these approaches are that they often conflict with other strategic goals of the institution (e.g. quality, diversity, etc.).
Enter predictive analytics. Although this may seem contradictory to the premise that 2009 was an unpredictable year, statistical modeling techniques are not only warranted in times like these, but even more necessary than during ‘normal’ recruitment cycles. You see, predictive modeling does more than forecast aggregate numbers. The idea is really to understand the key drivers of enrollment using mathematical models. Once you truly understand who and why students enroll at your institution, you can better segment or target the audience and more effectively recruit the students that are more likely to enroll.
As a result, you will be able to address the problems I’ve listed above with more surgical precision. So instead of giving a blanket discount to all accepted students, you will be able to identify the students that will benefit the most from the award through the analysis of a combination of driving factors.
“What-if scenario” decision support applications make it possible to interact with predictive models so that you can “war-game” various scenarios. So, for example, you could make the output of your model more conservative in an uncertain recruitment cycle by applying limits or weighing certain variables (like household income) more heavily in your enrollment propensity models.
The lesson is that there is almost always a way to use data to your advantage, even in turbulent times. Broadly applying tactics across all of the prospective student body should be a last resort.