Vaccination programs are the mainstay of control for many infectious diseases. Heterogeneous coverage is hypothesised to reduce vaccination effectiveness, but this impact has not been quantified in real systems. We address this gap using fine-scale data from two decades of rabies contact tracing and dog vaccination campaigns in Serengeti district, Tanzania. Using generalised linear mixed models, we find that current local (village-level) dog rabies incidence decreases with increasing recent local vaccination coverage. However, current local incidence is most dependent on recent incidence, both locally and in the wider district, consistent with high population connectivity. Removing the masking effects of prior non-local incidence shows that, for the same average prior vaccination coverage beyond the focal village, more spatial variation increases local incidence. These effects led to outbreaks following years when vaccination campaigns missed many villages, whereas when heterogeneity in coverage was reduced, incidence declined to low levels (<0.4 cases/1,000 dogs annually and no human deaths), such that short vaccination lapses thereafter did not lead to resurgence. We inferred ongoing rabies incursions into the district, suggesting regional connectivity as an important source of residual transmission. Overall, we provide an empirical demonstration of how the same average vaccination coverage can lead to differing outcomes based on its spatial distribution, highlighting the importance of fine-scale monitoring in managing vaccination programs.