This paper builds on a range of traditions in educational research and design to argue, with empirical evidence, that constructing powerful instructional materials and approaches that work at scale requires a grounding in theory and a commitment to engineering practice, including rapid prototyping and multiple development cycles. Specifically, we claim that improving practice within a reasonable timescale requires replicable materials that integrate: (1) grounding in robust aspects of theory from prior research; (2) design tactics that combine these core ideas with a design team's creativity; along with (3) flexibility in the draft materials that affords adaptation across contexts; (4) rapid prototyping, followed by iterative refinement cycles in increasingly realistic circumstances; with (5) feedback from each round of trials that is rich and detailed enough to inform revision; and (6) continued refinement on the basis of post-implementation feedback 'from the field'. Examples of successful implementation are analysed and related to the various roles that research-based theory and programmatic research-based methods of development can and should play in the complex process of turning insights from research into improvements in practice. In contrast, we shall argue that materials which are written and published without the development processes (4) to (6)--still the great majority--lack research validity for use at scale.