University of Phoenix made headlines earlier this month when its parent company, the Apollo Education Group, was bought for $1.1 billion after posting a $45.2 million loss in its most recent fiscal quarter and facing shrinking student enrollment. Despite this high profile news, visitors to the university’s website would be hard-pressed to learn that the institution was struggling. In fact, University of Phoenix is currently running an advertising campaign boasting of its ranking as one of the top 25 large higher education institutions in salary after attending.

This claim relies on data from the Department of Education’s College Scorecard, a comprehensive dataset that includes debt and earnings information for former students from thousands of US universities. According to these data, University of Phoenix’s claim isn’t false: the median income for its federal financial aid recipients 10 years after enrollment is $53,400, placing it in the top 10 percent of universities enrolling more than 15,000 undergraduates, in the company of institutions like Rutgers University, University of Connecticut, and New York University. Of course, there’s a deeper story behind this number, and it reveals the difficulty of generating data that provides students with accurate information about their potential future earnings.

The current “salary after attending” metric used on the College Scorecard is based on the median earnings of federal aid recipients who enrolled in college in the fall of 2001 or 2002 and were working and not enrolled in school 10 years later (i.e. measured in 2011/2012). This lagged metric allows sufficient time for students to finish college and enter the labor market, but the downside is that it functions an indicator of institutional quality for students who enrolled nearly 15 years ago. This is not an issue at institutions where quality is fairly constant over time. But for rapidly changing institutions like University of Phoenix, this measure quickly goes stale.

In the late 1990s, University of Phoenix’s student body was largely made up of adult learners looking to finish their bachelor’s degrees. As a result, the typical University of Phoenix student was distinctly different than a student at other large four-year schools. Independent students who applied for federal financial aid to attend University of Phoenix in 1997/1998 reported an average of $97,626 in household income (adjusted to 2015 dollars), while independent students attending comparison colleges report an average of just $35,437. (My analysis defines comparison colleges as the 50 large four-year institutions with the highest salaries after attending and focuses on students who are independent for financial aid purposes, most of whom are age 24 or older).

Over the ensuing years, however, independent students at University of Phoenix begin to look more and more like students in comparison schools. By 2005/2006, independent students at University of Phoenix reported an average household income of $55,876, compared to $28,254 for comparison schools. My analysis of more recent Scorecard data shows that, starting around 2008, independent students at University of Phoenix have a household income profile that approaches that of independent students at comparison institutions.

Despite the stark difference in entering household income, student outcomes, in terms of earnings 6 and 10 years after enrollment, look similar to comparison schools. As average household income for entering independent students trend downward, so do University of Phoenix student outcomes. Based on mean earnings after 6 and 10 years of enrollment, University of Phoenix students initially rank near the top of large four-year schools for the 1997/1998 cohort, and progressively decline in rank by the 2005/2006 cohort.

University of Phoenix performs much better on older metrics than it does on more recent ones, indicating that its “top 25” bragging rights are unlikely to persist for long. Although the school may relish the College Scorecard’s reliance on old data now, the downside is that any improvements made by the for-profit’s new investors will take time to appear in the earnings data. University of Phoenix may soon view this as a bug rather than a feature.