The present study conducted a randomized control trial to assess the efficacy of two spatial intervention programs aimed to improve Grade 4 (N = 287) students’ spatial visualization skills and math performance. The first treatment (N = 98) focused on isolated spatial training that included 40 min of digital spatial training across fourteen weeks. The second treatment (N = 92) embedded spatial visualization skill development into math lessons, along with the digital spatial training that provided practice of the newly acquired skills. A business-as-usual group acted as a control (N = 97). Engagement with the embedded intervention program (i.e., both lessons and digital training) showed large additive effects, highlighting the role of spatial reasoning tools to support the transfer of spatial reasoning to math. The isolated intervention program with the digital spatial training had a transfer effect on math, compared to a business-as-usual control, while spatial reasoning improvements for this group were mixed. The spatial skills targeted in the digital training had a mediation effect on math performance, despite not increasing in the pre–post-test design. The effects of the digital training cohort were moderated by initial spatial skill, with students with lower spatial reasoning making the least gains in math.