enrollment

How predictive modeling benefits enrollment managers

Charles Ramos

April 9, 2013

Co-written with Brian Jansen of Noel-Levitz

Predictive modeling can pinpoint which prospective students are more likely to enroll.
Predictive modeling can pinpoint which prospective students are more likely to enroll.

As the National Candidates’ Reply Date of May 1 quickly approaches, you may find yourself looking back on the past recruitment cycle and weighing how well your team was able to cultivate your prospective student population for the incoming class. As you do so, consider the following questions in your evaluation:

  • Did you use a shotgun approach when purchasing names in your student search efforts? Did you experience less than stellar response and conversion rates? What percentage of your enrolled population came from your search name purchases, and are you satisfied with the outcome?
  • Were your counselors struggling to build early and solid relationships with prospective students due to a large inquiry or applicant pool? Do you feel that your counselors are unable to truly make impactful and influential phone contacts with students/families due to the size of your pool of students at each stage of the funnel?
  • Was there a lack of prioritization or absence of segmentation when communicating with prospective students? Would you deem your communication flow/plan as a “one-size-fits-all” approach?
  • Did budget cuts limit your ability to work the prospective student funnel efficiently and effectively?
  • Did the size of your inquiry and/or applicant pool shrink?
  • Did your counselors travel extensively in secondary and tertiary markets?

Many campuses experience challenges in these areas, and with increased “secret shopping” by students and tighter recruitment budgets, these challenges are likely to increase. That’s why more and more campuses are using an advanced statistical method to guide their recruitment: predictive modeling.

What is predictive modeling?

Predictive modeling uses statistical analysis of past behavior to simulate future results. For campuses, a predictive model assesses the likelihood that a student will enroll at your institution by the degree to which the student shares the characteristics of the current student body.

The history behind who enrolled at your school tells a powerful story about who you can expect to enroll in future terms. You can find characteristics that influence enrollment and also weigh the amount of their influence on enrollment, then apply that model against each student in your pool to see how much they fit the profile of a student who did enroll. (This can be refined and strengthened even further by appending additional socioeconomic and demographic data to the model, as Noel-Levitz does.) Your prospective students receive a qualifying score that helps you quickly gauge their enrollment likelihood.

A predictive model not only can enhance your primary market recruitment, but also identify how your institution can effectively and deliberately target students and populations in your secondary and tertiary markets—leading to increased enrollment numbers via a more efficient recruitment process. These models inform and affect not only communication strategies, but also territory management and travel, providing enrollment managers and counselors with valuable model scoring information by region and high school that assists in understanding where your institution has the greatest potential for future and/or continued growth. Such a tool, therefore, takes travel—notoriously a low return on investment venture—and improves this return significantly by placing statistical data and modeling at the center of territory management.

Four ways predictive modeling makes your funnel more manageable

Using predictive modeling can be highly beneficial in developing specific stages of the enrollment funnel. For example, if you have traditionally cast a wide net and purchased hundreds upon thousands of test takers as part of your student search, a predictive model can help you pinpoint which of these students would most benefit from an extended mail flow. Those that are highly likely to enroll based on your predictive model could receive multiple communications through various channels whereas those students deemed less likely to enroll may only receive an inexpensive postcard mailing or e-mail communications, asking them to respond. You still reach out to your entire pool, but you reallocate more of your postage and print dollars toward those prospective students who have the best chance of enrolling. Or better yet, you can take the students from your purchase with the highest propensity to enroll and communicate with them as if they were inquiries. If your historical data has already told us through a predictive model that they are a good fit for your school, why waste valuable time searching them in the traditional sense? Go ahead, send them that fancy viewbook!

Here are four ways that predictive modeling helps enrollment managers be more efficient and effective.

Benefit 1: Qualifying lists before purchasing student names

Have you ever felt that name purchasing for search sometimes feels like throwing a dart at a dartboard—blindfolded?

You are not alone. Presented with a huge pool of potential prospects, many campuses are faced with the daunting task of somehow choosing which names to buy, and hope that effective communications with this group will positively impact future enrollment numbers. The problem is, if you choose a majority of students who are less than likely to enroll at your institution, you then fight an uphill battle all year. You could continue to buy more names, but this approach is akin to “throwing spaghetti on the wall”—even if it sticks, it’s messy and inefficient.

Predictive modeling can help you qualify names before you purchase them. Instead of guessing, you have a statistical model to guide you and add more students to your funnel who are already more likely to enroll.

Benefit 2: Prioritizing your inquiry pool

Working with an inquiry pool can be as equally daunting and inefficient. If you treat each inquiry the exact same way in your communication plan (mail, e-mail, phone communications), you inevitably wind up spending valuable resources on many students who are unlikely to enroll, or risk not having enough contact with the most valuable inquiries in your pool.

With predictive modeling, you can statistically identify those students most likely to enroll and stratify your communication plan to ensure that these students receive the most impactful printed materials, phone calls, and e-mail communications. With ever-looming budget cuts, the ability to target specific students provides you the opportunity to judiciously expend your limited resources in a manner where you receive the highest return for your investment.

Benefit 3: Distinguish “soft applicants” from those you should pursue in your applicant pool

Outreach to your applicant pool is vital in your institution’s ability to complete as many of these applications as possible. How can you identify which of your applicants are just “soft applicants” versus those who, with direct contact and influence, will complete the application process and potentially move through the funnel?

Predictive modeling can strategically guide your decisions during a time of year when you are trying to increase inquiry-to-application conversion while having limited time and resources. You can quickly target applicants based on model score so that you can maximize the enrollment yield potential of the entire applicant pool.

Benefit 4: Optimize communications with students in your admit pool

Once students are admitted, you obviously should connect with all students, and the expectation should be that each admit is contacted regularly. This does not necessarily erase the need for predictive modeling, though.

Earlier we discussed the ability to stratify communications based on student predictive modeling scores, but at the admit stage you now have an opportunity to prioritize your communications. Instead of choosing who you will communicate proactively with, you can now reach out to all students, but in order of highest to lowest probability to enroll based on the predictive model. This strategy ensures that your time and efforts are maximized. Or perhaps you have a spring yield event scheduled, but registrations are low and your postage dollars are nearly exhausted. While e-mail follow-up and RSVP reminder calls are key, could an additional invitation mailed to the high probability admits make a difference in having a good turnout versus an outstanding turnout? Another thought: how about enlisting faculty to contact the highest probability admits? Their time is extremely valuable, if not limited, so put the science of predictive modeling behind your request for their assistance.

Results: Deploying resources more strategically to reap better enrollment yields

With campuses using Noel-Levitz predictive modeling, 85 percent of their enrollees came from students scoring in the top 44 percent of the models. Only 6 percent came from the bottom 36 percent. This illustrates how predictive modeling can help campuses target their efforts and resources to more effectively communicate with a much smaller portion of their overall prospective student population and still gain significant results toward their enrollment goals.

My colleague Brian and I are happy to continue this discussion with you or answer any questions you may have. Please e-mail us and we will respond as quickly as we can. No matter what, we encourage you to learn more about predictive modeling. It brings more science to enrollment management and provides a way to prioritize, quantify, and manage a search process that seems to become more amorphous and anonymous with every passing year.


Read More In:
Enrollment