We conduct a randomized controlled trial with households of secondary school students in Bangladesh to investigate how parents adjust their investments in response to three educational interventions: an informational campaign about an educational phone application, an internet data subsidy, and one-on-one phone learning support. We find that offering an educational service in a context where other barriers to take-up exist can still trigger parental educational investments by acting as a signal or nudge. These behavioral changes result in lasting learning gains concentrated among richer households, reflecting that the relevant behavior change--increased tutoring investment--is easier for them to implement. In contrast, when interventions do increase take-up, they have the potential to narrow the socioeconomic achievement gap. We observe that increased usage of the targeted educational service limits parental behavioral responses. This implies that learning gains in these cases are directly caused by the potential effectiveness of the services adopted. In our setting, remote one-to-one teacher support improves learning among students from poorer households, whereas receiving the free data package jointly with the app information has no impact on learning.