How did the demand for digital and non-digital learning products change during periods of elevated school closures due to COVID-19? In this paper, we model the relationship between public school closure levels and the revenue from digital and non-digital learning products for a large test prep service. We leverage a difference-in-difference approach, focusing on COVID-19 school closures to estimate our models. The results show that as school closures increase, purchases for digital learning products decrease but are offset by a corresponding increase in demand for non-digital learning products, such as flashcards or books. We find that these results are moderated by demographic variables, namely that most gains were amplified from those with greater broadband access, areas with greater non-White racial composition, urban areas, and areas with relatively high median income.