Institutional rules that appear neutral can produce unequal outcomes when applied to students who differ in how they respond to them. We study this in the context of academic probation, a policy used by nearly every U.S. college to warn students whose grades fall below a fixed threshold that continued enrollment is conditional on improvement. To do so, we draw on student records from a large California public university, linked to state administrative earnings data. We identify the causal effects of probation using a regression discontinuity design comparing students who narrowly scored above the cutoff to those who just missed it. We find that probation raises dropout, lowers degree completion, and reduces earnings at ages 24 to 31 among low-income students, while high-income students are largely unaffected. We then contrast the standard policy with an alternative, at the same institution, which incorporates a coaching program. We find that the alternative policy eliminates these effects and improves academic performance. Taken together, these results suggest that the design of institutional policies is critical for closing socioeconomic gaps in degree completion.