Natural-product pharmacology has conventionally inferred activity from averaged biochemical, cellular, tissue, or organism-level measurements. These approaches remain indispensable, but averaging can obscure which cellular states respond, which remain refractory, where pharmacologically relevant species accumulate, and whether apparent pathway normalization reflects direct action or changing cellular composition. Single-cell and spatial technologies create a complementary analytical layer by resolving treatment-associated states, intercellular relationships, tissue neighborhoods, and local molecular distributions. This state-of-the-art review examines how these technologies are entering natural-product research, with particular attention to heterogeneous response, cell-state selectivity, tissue context, and the evidential steps required for stronger mechanistic interpretation. Current applications span defined phytochemicals, complex herbal interventions, perturbational transcriptomics, spatial transcriptomics, and spatial metabolomics. Collectively, the literature shows that cell-resolved measurements can reveal pharmacological responses concealed by bulk assays and can prioritize cellular compartments or interactions for subsequent validation. However, cell-state association does not itself establish causal mechanism, formula-level effects cannot automatically be assigned to individual constituents, and spatial colocalization does not prove target engagement. We therefore propose an evidence-bounded interpretation architecture in which cellular response, local exposure, pathway perturbation, and orthogonal mechanistic validation remain distinct until experimentally connected. The near-term opportunity is not simply to add single-cell or spatial assays to natural-product studies, but to design perturbation experiments in which chemical identity, dose, time, cellular state, tissue location, and validation are jointly interpretable.
