In this paper we consider the implications of the distributions of dust and metals in the disc of M31. We derive mean radial dust distributions using a dust map created from Herschel images of M31 sampling the entire far-infrared (FIR) peak. Modified blackbodies are fit to approximately 4000 pixels with a varying, as well as a fixed, dust emissivity index (beta). An overall metal distribution is also derived using data collected from the literature. We use a simple analytical model of the evolution of the dust in a galaxy with dust contributed by stellar sources and interstellar grain growth, and fit this model to the radial dust-to-metals distribution across the galaxy. Our analysis shows that the dust-to-gas gradient in M31 is steeper than the metallicity gradient, suggesting interstellar dust growth is (or has been) important in M31. We argue that M31 helps build a case for cosmic dust in galaxies being the result of substantial interstellar grain growth, while the net dust production from stars may be limited. We note, however, that the efficiency of dust production in stars, e.g., in supernovae (SNe) ejecta and/or stellar atmospheres, and grain destruction in the interstellar medium (ISM) may be degenerate in our simple model. We can conclude that interstellar grain growth by accretion is likely at least as important as stellar dust production channels in building the cosmic dust component in M31. ; Comment: 12 pages, 7 figures. Published in MNRAS 444, 797. This version is updated to match the published version