The University of Arizona had launched an institution-wide effort to revamp their data environment and achieve their "Recruit to Retain" mission, increasing first-year retention from 82–90% by 2020.
They had been using an in-house enrollment projection model to predict first-year GPA for incoming students, but this model did not reflect the rich data that becomes available after a student arrives. They knew that getting to 90% retention would mean finding that extra student in 10 for whom they could improve outcomes.
Using Illume, they uncovered a group of 660 non-resident female students with GPAs between 2.4–3.0 who were not initially on their radar as being an at-risk student population. This group accounted for 1% of the attrition rate. The institution has difficulty retaining non-residents, but this work allows them to narrow the field to find which of them are at risk and why.