Drought propagates from precipitation deficits into soil-moisture deficits before affecting water supplies, yet whether the timescale of this meteorological-to-surface-soil-moisture transition can be predicted from basin properties has not been established for Iran. Using monthly CHIRPS precipitation and ERA5-Land surface (0–7 cm) soil moisture for 2003–2024 across 66 HydroBASINS level-5 sub-basins, we compute standardized indices (SPI, SSI) and estimate, for each basin, the precipitation accumulation timescale that maximizes the SPI–SSI correlation, with significance corrected for serial autocorrelation. Coupling is strong and ubiquitous: every basin shows a significant relationship (p < 0.01), the median correlation is 0.58 (range 0.43–0.74), and 85% of basins exceed |r| = 0.5. The surface soil-moisture response is fast, with a median accumulation timescale of two months and 94% of basins responding within two months. The timescale is spatially organized (Moran’s I = 0.17, p = 0.002) and is associated in-sample with cropland fraction, vegetation cover, aridity, and snow fraction. Critically, however, these static attributes fail to predict the timescale under spatial cross-validation (gradient-boosted-tree R² = −0.18, i.e. worse than predicting the mean; 95% CI [−0.19, −0.13]), while neither a location-only model nor a combined model exceeds the no-skill level. A positive-control test confirms the same procedure recovers a genuine attribute signal at this sample size, so the null result reflects an absence of transferable attribute control rather than insufficient data. We conclude that meteorological-to-surface-soil-moisture propagation speed in Iran appears to be set by transient hydroclimatic dynamics and regional spatial structure rather than to be predictable from the fixed basin attributes examined here; the per-basin timescale moreover shows limited temporal reproducibility. These results are a reproducible caution for attribution studies that report in-sample attribute importance without spatial validation. Conclusions are insensitive to the standardization method.