Further characterization to properly assess the fate of organic matter quality during anaerobic digestion and organic carbon mineralization in soils is required. Organic matter quality based on its accessibility and complexity was employed to successfully classify 28 substrate/digestate pairs through principal components and hierarchical clustering analysis. The two first components explained 58.02% of the variability and four main groups were separated according to the feedstock type. A decrease in the accessibility (16–66%) and an increase in the complexity (34–98%) of the most accessible fractions was noticed. Besides, an increase of non-biodegradable compounds (17–66%) was globally observed after anaerobic digestion. The observed trends in the conversion of organic matter during anaerobic digestion have allowed to fill the gap in the modeling of the anaerobic digestion process chain. Indeed, partial least squares regressions have accurately predicted the organic matter quality of digestates from their inputs (R2 = 0.831, Q2 = 0.593) although the digester operational conditions (temperature and hydraulic retention time) were non-explicative enough. As a novel approach, the predicted digestate quality was used to feed a partial least squares regression model previously developed to predict organic carbon mineralization in soil. The combined models have predicted experimental organic carbon mineralization in soil (R2 = 0.697) with a model quality similar to the model for organic carbon mineralization in soil (R2 = 0.894). This is the first study that has successfully conceived an additional step in the prediction of organic matter fate from raw substrate before anaerobic digestion to soil carbon mineralization.