The term “summer melt” is used to describe students who apply, get admitted, and then commit to a college but never actually show up to their classes. This is not a new phenomenon; higher education institutions are keenly aware of this issue and have actively sought to reduce melts by keeping students on track through scholarships, offering free classes, implementing chatbots on their websites, and increasing communication efforts across the board.
The annual melt is projected to increase dramatically this year as a result of the COVID-19 pandemic. According to a recent survey from this past spring, one in six high school seniors who, before the pandemic, expected to attend a four-year college on a full–time basis said that they will choose a different path this fall. These students will likely take a gap year, enroll part-time in a bachelor’s program, pursue work opportunities, or attend a community college. The repercussions of this could reverberate broadly throughout the higher education space, affecting the students, their families, and the economy at large.
What if schools could better predict which students are most susceptible to summer melts?
In a previous post, I illustrated how to use Einstein Prediction Builder to predict which one–time donors could become sustainers, and what next steps your institution can take to ensure the long-term health of your sustainer program. That same functionality could also work to predict the “churn” of summer melts.
In order to replicate the variables that might create a summer melt, I loaded 6,000 contact records (the minimum is 400) into the Education Data Architecture (EDA) and asked the question, “Is this student likely to churn after their application is approved?” The answer I am looking to receive from Einstein is a simple “Yes” or “No.”
Next, Einstein needs to know if there is a field that can answer my prediction question. I don’t have a field for “summer melt,” so instead I set up filters to address the question head-on.
I chose prediction variables such as the date of their last communication date with the school, whether or not they’ve paid their most recent financial obligation to the institution, etc. Have they started their financial aid application process? Are they the first of their family to attend college? So on and so forth. The goal here is to feed the system with enough pertinent information to ensure an actionable prediction.
Einstein will take the reins and conduct its magic from there, allowing me to identify, based on the (fake) data that I loaded, which one of my indicators have the highest probability to predict the melt. Next, your school should consider an action plan based on these indicators and the persistent risk of summer melt.
Marketing Cloud can then use the predictors to automate, prioritize, and personalize messages based on the summer melt risk. From there, you can create a custom journey that will combine email, mobile, and social interactions to pre-empt the student churn or summer melt. Doing so can ensure your institution is better prepared to combat any coming summer melts, leading the way for a more equitable and mutually beneficial semester for institutions and students alike.
