Embodied learning technologies have shown efficacy in laboratories with ideal supportive conditions, but their effectiveness in classroom with “real-world” constraints is yet understudied. Inspired by the innovation implementation framework, we compare the classroom-situated engagements of two student pairs and their teachers with the action-based embodied design for proportions with earlier laboratory and classroom study findings and conjecture on influential factors. Much of these classroom students’ sensorimotor learning resembled laboratory findings, but they had more opportunities to be overtly engaged with their hands and self-directed in including artifacts, likely influenced by (unintended) technological changes and setting-specific environmental affordances. Their teachers’ engagements resembled laboratory findings to some extent, but showed less perceptiveness to students’ qualitative multimodal expressions and more directedness in introducing new quantitative forms of engagements, likely influenced by setting-specific fragmented access and novelty of the embodied pedagogy. We discuss the importance of focusing on teachers and conducting semi-natural efficacy research.@en