Changes in ocean-driven basal melting have a key influence on the stability of ice shelves, the mass loss from the ice sheet, ocean circulation, and global sea level rise. Coupled ice sheet–ocean models play a critical role in understanding future ice sheet evolution and examining the processes governing ice sheet responses to basal melting. However, as a new approach, coupled ice sheet–ocean systems come with new challenges, and the impacts of solutions implemented to date have not been investigated. An emergent feature in several contributing coupled models to the 1st Marine Ice Sheet–Ocean Model Intercomparison Project (MISOMIP1) was a time-varying oscillation in basal melt rates. Here, we use a recently developed coupling framework, FISOC (v1.1), to connect the modified ocean model ROMSIceShelf (v1.0) and ice sheet model Elmer/Ice (v9.0), to investigate the origin and implications of the feature and, more generally, the impact of coupled modeling strategies on the simulated basal melt in an idealized ice shelf cavity based on the MISOMIP setup. We found the spatial-averaged basal melt rates (3.56 m yr−1) oscillated with an amplitude ∼0.7 m yr−1 and approximate period of ∼6 years between year 30 and 100 depending on the experimental design. The melt oscillations emerged in the coupled system and the standalone ocean model using a prescribed change of cavity geometry. We found that the oscillation feature is closely related to the discretized ungrounding of the ice sheet, exposing new ocean, and is likely strengthened by a combination of positive buoyancy–melt feedback and/or melt–geometry feedback near the grounding line, and the frequent coupling of ice geometry and ocean evolution. Sensitivity tests demonstrate that the oscillation feature is always present, regardless of the choice of coupling interval, vertical resolution in the ocean model, tracer properties of cells ungrounded by the retreating ice sheet, or the dependency of friction velocities to the vertical resolution. However, the amplitude, phase, and sub-cycle variability of the oscillation varied significantly across the different configurations. We were unable to ultimately determine whether the feature arises purely due to numerical issues (related to discretization) or a compounding of multiple physical processes amplifying a numerical artifact. We suggest a pathway and choices of physical parameters to help other efforts understand the coupled ice sheet–ocean system using numerical models.