This note shares early implementation science insights from scaling Differentiated Learning (DL) to 16,000+ schools. We identify several frictions to sustaining high implementation fidelity at scale. In Ghana, teachers reported strong adherence to DL, but classroom observations revealed gaps in the critical but high-effort practice of grouping students by ability. To address this gap, the team developed a low-cost coaching checklist and ran an A/B test on the new tool. We find real-time improvements during the school term, with the checklist increasing the frequency of student grouping within DL lessons by 15 percentage points. This cost-effective implementation tweak, identified and deployed via rapid measurement and iterative adaptation within government systems, offers a clear example of using implementation science to achieve higher fidelity and effective scaling.