Over the past several decades, digital technologies have evolved from supporting business processes and decision-making to becoming an integral part of business strategies. Although the IS discipline has extensive experience with digitalization and designing sociotechnical artifacts, the underlying design knowledge is seldom systematically accumulated across different settings and projects, and thus cannot be transferred and reused in new contexts. Motivated by this gap in the research, we turn to the data management field, where reference models have become important sources of descriptive and prescriptive domain knowledge. To study knowledge accumulation in reference models, we analyze the revelatory and extreme case of a longitudinal DSR process involving more than 30 European companies and 15 researchers from three universities over 12 years. The insights into reference model development allow us to theorize about knowledge accumulation mechanisms from both a process perspective and an artifact perspective: First, we observe that knowledge accumulation occurs in stages in response to technology’s evolving roles in business (problem space) and as a result of maturing design knowledge (solution space). Second, we find that reference models act as design boundary objects; they explicate and integrate knowledge from different disciplines and allow for the development of design knowledge over time—from descriptive (conceptual) models to prescriptive (capability or maturity) ones. To cope with fundamental changes in the problem space, these models require continuous updating as well as transfer/exaptation to new problem spaces. Our findings inform the IS community about the fundamental logic of knowledge accumulation in longitudinal DSR processes.