The mechanical response of sustainable masonry assemblages depends not only on the absolute strengths of their constituent units and mortars, but also on how coarse- and fine-recycled aggregate substitutions alter the coupled unit–mortar system. This study presents a design-space reanalysis of a ten-mix experimental matrix of cement and geopolymer masonry prisms tested in compression, flexural bond, and shear bond. A stage-wise framework isolates four effects: binder substitution, mortar-class upgrade, RCA-only replacement, and the incremental penalty of adding RFA after full RCA replacement. Three derived indicators are used: performance retention, replacement-stage sensitivity, and mortar amplification. All reported percentages were recalculated directly from the mix-level dataset, and the resulting trends are interpreted as matrix-specific rather than as formal statistical interactions. Within this dataset, prism compressive strength is most sensitive to coarse aggregate replacement, with a 43.3 % loss under RCA-only substitution and a further 18.4 % loss when RFA is added at fixed RCA100. Flexural bond is sensitive to both stages, dropping by 34.3 % under RCA-only replacement and by another 32.8 % when RFA is introduced. Shear bond behaves differently: it increases by about 29.7 % under RCA-only replacement and changes only marginally when RFA is added. Along the equal-replacement path, the most severe flexural-bond loss occurs already at 25% combined replacement, whereas shear bond recovers beyond 50% replacement. Compression failure modes shift from conical break to cone-and-split and then to cone-and-shear as combined replacement increases, while stronger mortars promote face-shell separation. The revised interpretation provides cautious design-oriented guidance for compression-dominated, flexure-sensitive, and shear-sensitive masonry applications within the scope of the studied system.