Lateral erosion of marsh cliffs is a key driver of salt marsh loss but remains poorly understood and underrepresented in numerical models. This study investigates cliff erosion processes using two modeling approaches—subgrid1D and 2D depth-averaged—applied to the Wierum salt marsh in the Dutch Wadden Sea. Both approaches simulated cliff erosion over 9 months and were validated with drone-derived elevation data. Results indicate that monthly-averaged wave conditions yield more accurate lateral retreat predictions than yearly averages, as they better capture short-term variability in wave energy. Both models identify areas without erosion in regions without cliffs, demonstrating good alignment with field data. The subgrid1D model provided closer agreement with observed retreat patterns, while the 2D model, which assumes constant erosion across the cliff face, tended to overestimate retreat. Discrepancies between model results and measurements in areas of high erosion suggest the need to account for additional spatial processes beyond the wave impacts considered here.