Functional matter is often designed and reported as disconnected advances in synthesis, structure, and performance, which limits transferability across materials classes and length scales. This review introduces a cross-scale design grammar that unifies how researchers specify, test, and validate causal links from processing and reactions to architecture, interfaces, defects, and device-level outcomes. We formalize grammar units as controllable operators, interaction rules, constraints, and measurable metrics, and show how uncertainty and failure modes propagate along the synthesis to structure to function chain. The framework integrates mechanistic control across solution, solid-state, vapor, electrochemical, and mechanochemical routes; mesoscale assembly and hierarchical architectures; interfaces and interphases as transport gatekeepers; and defect chemistry as both a performance lever and a degradation driver. We then treat multimodal characterization as an inference and evidence-fusion problem and map it onto a modeling ladder that spans mechanistic, continuum, statistical, and hybrid approaches with uncertainty quantification and validation. Finally, we provide benchmarking and reporting templates, truth-table criteria for what counts as improvement, and case-study scorecards that identify the dominant bottlenecks in representative applications. By converting fragmented knowledge into a reusable grammar, this review offers a practical, end-to-end playbook for designing functional matter with measurable causality, reproducibility, and observation-ready.