Bulge–disc decomposition is a valuable tool for understanding galaxies. However, achieving robust measurements of component properties is difficult, even with high-quality imaging, and it becomes even more so with the imaging typical of large surveys. In this paper, we consider the advantages of a new, multiband approach to galaxy fitting. We perform automated bulge–disc decompositions for 163 nearby galaxies, by simultaneously fitting multiple images taken in five photometric filters. We show that we are able to recover structural measurements that agree well with various other works, and confirm a number of key results. We additionally use our results to illustrate the link between total Sérsic index and bulge–disc structure, and demonstrate that the visually classification of lenticular galaxies is strongly dependent on the inclination of their disc component. By simulating the same set of galaxies as they would appear if observed at a range of redshifts, we are able to study the behaviour of bulge–disc decompositions as data quality diminishes. We examine how our multiband fits perform, and compare to the results of more conventional, single-band methods. Multiband fitting improves the measurement of all parameters, but particularly the bulge-to-total flux ratio and component colours. We therefore encourage the use of this approach with future surveys.