In order for ceramic compositional analysis to yield more than tentative attribution of ceramics to broad source areas, models of ceramic resource procurement and production will have to incorporate some of the acknowledged complexities of real-world ceramic production. In this paper, we discuss some experiments using artificially-mixed data sets derived from raw materials collected in contemporary potter-making communities. These experiments suggest that we can distinguish compositional effects due to tempering from effects due to clay source differences. Next, we show how patterns observed in the artificial data recapitulate patterns observed in a real data set of Late Preclassic Guatemalan fine red ceramics. Finally, we offer a reconstruction of fine red production based on our analysis of the compositional data, comparison with the artificial data sets, and ethnographic observations. Our results demonstrate the utility of an approach that combines analysis of ancient pottery with ethnographic and compositional study of the relevant ceramic environment.