As of mid-2016, national nonprofit Teach For America (TFA) had struggled with three consecutive years of declining application totals, and senior management was re-examining the organization's strategy, including recruitment and selection. A few months earlier, former TFA corps member Michael Metzger had spearheaded the launch of a predictive analytics team responsible for integrating strategy across TFA's recruitment and admissions continuum. Metzger believed that predictive analytics could help address TFA's recruitment and admissions challenges. Now, TFA's senior management wanted Metzger and his team to put their data models to the test. From a data modeling standpoint, what variables and models should TFA consider as it began a new recruitment season, and more generally, how heavily should TFA rely on predictive models to reverse the negative trends?