Design science research (DSR) is increasingly conducted in contexts in which neither the problem space nor the effects of artifacts can be fully specified in advance. Wicked socio-technical challenges, participatory settings with heterogeneous stakeholders, and technologies with emergent capabilities (e.g., generative AI) often yield outcomes that are unexpected yet epistemically significant. Nevertheless, prevailing DSR method guidance tends to treat such outcomes as anomalies to be reduced through iterative refinement rather than as resources for knowledge production. This paper advances Toward a Science of the Unexpected by conceptualizing serendipity as an epistemic mechanism that can be systematically incorporated into DSR. Building on research on serendipity, pragmatist epistemology, abductive reasoning, and the concept of epistemic objects, we develop a four-phase iterative model comprising (i) an unexpected encounter, (ii) recognize and reflect, (iii) abductive leap, (iv) materialization. We further derive three guiding principles (treating artifacts as epistemic objects, cultivating a prepared mind, and conducting reflexive evaluation) to complement refinement-oriented iteration with discovery-oriented inquiry. The paper contributes a conceptual foundation and methodological guidance for DSR scholars seeking to engage systematically with unexpected outcomes under conditions of emergence and uncertainty.