Abstract: The modern generation of telescopes are sharp enough to observe chemical inhomogeneities in the interstellar medium of galaxies on a range of spatial scales, allowing us to directly see the small-scale processes that play a critical role in shaping how galaxies grow and evolve. In this talk, I present a selection of novel statistical techniques that can be used to model galaxy data, characterise the spatial chemical distribution of galaxies, and make meaningful comparisons to both analytical and simulated models. For well-resolved data (finer than ~200 pc per pixel), I introduce the geostatistical modelling framework that allows the multiscale metallicity structure of galaxies to be modelled and key physical parameters to be extracted. For data that is more coarsely resolved, I demonstrate the power of forward-modelling to fit more accurate metallicity gradients, as well as asymmetrical models that capture more details on a galaxy’s evolutionary history. Finally, I discuss a way to infer the presence of metallicity variations in unresolved sources by combining absorption and emission-based spectroscopy in host galaxies of gamma-ray bursts. These results shed light on the evolution of the interstellar medium of galaxies throughout cosmic time, and hold great promise for use in the analysis of future data sets.