This study focuses on the process by which multi-unit organizations develop and replicate high-quality codified routines. Previous studies have primarily focused on the exploitation stage of routines, emphasizing the debate between strict adherence and local adaptation. However, as cases have emerged where large-scale replication of immature routines fails to deliver the expected results, the importance of enhancing the quality of the replicated routines themselves has been increasingly recognized. A critical factor in successful routine replication is whether the "Arrow Core," which reflects the key success factors of a business model, is effectively embedded in the routine. Building on Pentland et al (2012)’s routine emergence model, this study simulates a "bottom-up routine formation process," where problem-solving action patterns developed by operational units are collected and standardized by a central unit. Specifically, the analysis explores how factors such as the richness of experience at the operational unit level, the duration of routine formation, and the central unit's selection and degree of reference to specific operational units affect the quality of the codified routine. The significance of this study lies in its focus on the exploration phase of improving the quality of codified routines, an area that has received limited attention in routine replication research. It also visualizes the causal ambiguity surrounding Arrow Core identification and the knowledge creation processes of operational units through a computational model. By doing so, this study offers a new perspective for reevaluating the routine formation and replication processes, contributing to the sustained performance improvement of multi-unit organizations.
