Amendment studies in saline–alkali cropland often emphasise treatment–control comparisons, whereas rate-dependent responses within the same field system remain less well resolved. We evaluated water-soluble organic fertiliser and Bacillus subtilis rates in a saline–alkali cotton (Gossypium hirsutum L.) field under magnetised–ionised water irrigation. A single-season (2025) randomised complete block experiment with three replicates included CK1 (magnetised–ionised water without amendments), CK2 (conventional water without amendments), and a 3 × 3 factorial of organic fertiliser (300, 600, and 900 kg ha⁻¹) and inoculant (30, 52.5, and 75 kg ha⁻¹) rates. Gravimetric soil water content (GSWC) and electrical conductivity (EC) were measured across five growth stages and seven soil layers; post-harvest bulk density, boll-stage microbial composition, leaf gas exchange, and seed cotton yield were also assessed. GSWC and EC showed stage- and depth-dependent responses. Post-harvest bulk density showed a non-monotonic pattern across organic fertiliser rates. Y2W2 produced the highest observed net photosynthetic rate and seed cotton yield, 14.04% and 10.51% above CK1, respectively, and the lowest observed bulk density at both depths, 2.84–4.00% below CK1. Mantel analysis indicated positive covariation of bacterial assemblage composition with boll-stage GSWC and seed cotton yield, although statistical support was limited after multiple-testing correction. Y2W2 showed favourable responses across several indicators but requires multi-season and multi-site validation.