It remains unclear how galactic environment affects star formation and stellar cluster properties. This is difficult to address in Milky Way-mass galaxy simulations because of limited resolution and less accurate feedback compared to cloud-scale models. We carry out zoom-in simulations to re-simulate 100–300pc regions of a Milky Way-like galaxy using smoothed particle hydrodynamics, including finer resolution (0.4M⊙ per particle), cluster-sink particles, ray-traced photoionization from O stars, H2/CO chemistry, and ISM heating/cooling. We select ∼106M⊙ cloud complexes from a galactic bar, inner spiral arm, outer arm, and inter-arm region (in order of galactocentric radius), retaining the original galactic potentials. The surface densities of star formation rate and neutral gas follow $\Sigma _\mathrm{SFR}\propto \Sigma _\mathrm{gas}^{1.3}$, with the bar lying higher up the relation than the other regions. However, the inter-arm region forms stars 2–3x less efficiently than the arm models at the same Σgas. The bar produces the most massive cluster, the inner arm the second, and the inter-arm the third. Almost all clusters in the bar and inner arm are small (radii <5pc), while 30-50 per cent of clusters in the outer arm and inter-arm have larger radii more like associations. Bar and inner arm clusters rotate at least twice as fast, on average, than clusters in the outer arm and inter-arm regions. The degree of spatial clustering also decreases from bar to inter-arm. Our results indicate that young massive clusters, potentially progenitors of globular clusters, may preferentially form near the bar/inner arm compared to outer arm/inter-arm regions.