A Three-Pronged Approach to Enrollment with the Data you Already Have
Ashley Kern, M.S., Founder and Data Scientist,SightLine
SightLine, Inc., is a growing boutique analytics firm providing targeted, easy to interpret predictive analytics and market intelligence solutions for colleges and universities. We empower higher education institutions to better understand the data they already have without complicated software installations and training.
SightLine is dedicated to personal customer services throughout the engagement. Clients have direct contact with our consultants and data scientists working with their data. We encourage a collaborative relationship with our clients allowing them to request adjustments to the analysis and outputs at any time.Using data science and propriety algorithms combined with customized research-backed interventions, our team provides on average a 15X return on investment for our clients.
We specialize in a blend of macro level market intelligence and benchmarking combined with micro level targeted predictive analyticsfocused on optimizing financial aid, financial aid forecasting and budgeting, increasing enrollment, and increasing retention and graduation rates.
SightLine provides hands-on analytical support to institutions throughout the student lifecycle, and this article highlights our thought process and methodologies for maximizing enrollment in particular.
We view enrollment challenges as multifaceted. Enrollment, financial aid, and admissions teams have their hands full with day-to-day operations during the enrollment period. Therefore, we have developed a three-pronged approach to enrollment management to support university leaders towards meeting their enrollment goals. This approach is composed of top-level market analysis and benchmarking, predictive analytics and segmentation analyses, and disseminating and implementing data driven enrollment and retention strategies to meet university goals; this process is detailed below.
1. Market Analysis
A customized market analysis is usually a first step in helping our clients understand how their published price, discounting, yield, revenue and enrollment trends compare with their direct competitors. Institutions use this information to understand potential options for adjusting pricing to optimizing enrollments and revenue. This leads into more granular questions about how to identify effective marketing and recruiting strategies.
2. Predictive Analytics and Student Segmentation
The typical second prong involves development of rigorous machine learning algorithms for identifying patterns in student enrollment choices and accurately predict behaviors for individual applicants. These highly accurate models based on state of the science machine learning algorithms provide the basis to study many automated what-if scenarios to develop expectations, including uncertainty bounds, for a range of enrollment options.
These models also provide a basis for ongoing revenue and budget forecasting with much greater accuracy than competing methods based on more traditional time series, regression alone, or general demographic-based models.
The combination of phases one and two provide a holistic analysis for;
• optimization of published and net pricing,
• targeted accepted student outreach strategies,
• specification of enrollment factors to drive marketing and recruitment,
• supplemental award decisions, and
• guidance on tailored college and university initiatives.
3. Implementing Enrollment and Retention Strategies
As individual students are identified as likely candidates for enrollment, student level variables are used to segment the applicant pool to provide enrollment staff with strategies for contacting and for setting appropriate aid and scholarship offers. Through this approach we often find groups of applicants with common characteristics which provide an avenue to reach out to similar groups that may not have been previously connected and aware of their likely affinity for a given institution. We have also found that a combination of predictive analytics and segmentation based on student attributes provides an effective way to deploy intervention strategies in a student focused context for both enrollment and retention. We believe retention starts during the enrollment process.
Targeted student outreach strategies, communications, and timelines are thoughtfully developed by our higher-ed consultants. These strategies are developed with a combination of leveraging public research, but also guided by data outputs and insights provided during phase one and two of this process. These consultants work to reduce or even remove the roadblocks that occur when attempting to implement data driven strategies at a given institution. SightLine consultants may even provide hands on training to admissions teams on how to use the predictive analytics output effectively.
What is the Underlying SightLine Process?
SightLine specializes in targeted student-level analyses using our proprietary modeling methods, resulting in the highest accuracy rates in the higher education data analytics space. During our proprietary model development process, we apply a range of predictive model types to your institutions’ data in order to develop and leverage the most accurate model type for your data.These models range from simple logistic regression through complex machine learning and AI models. These models are compared with a variety of metrics including sensitivity, specificity, receiver operating curves and overall accuracy to identify the right model to meet your institutional goals.
In each project it is common for us to formulate hundreds of models based on several algorithms to identify the most accurate and robust predictive methods for each institution. No single method is uniformly best because, unlike traditional statistical models, machine learning algorithm performance is a function of the combination of an algorithm and supporting data. We have found that traditional statistical models such as logistic regression rarely percolate to the top of these rigorous evaluations, although we continue to include them in our tests as a point of reference. These models are generally less accurate than other algorithms because student behaviors are much more complex (i.e. nonlinear functions of data) than can’t be captured by more traditional statistical algorithms.
We combine our unique predictive analytics with econometric principles and SightLine market intelligence reporting. Through this process, we can gain holistic insights into the strength of current enrollment and tuition strategies, simulate ‘what-if’ scenarios, and assess the strength of your current and future market and pricing strategy relative to competitors.