Variable stiffness (VS) laminates, characterized by curvilinear fiber paths, offer significant potential for optimizing structural performance in aerospace and automotive applications. Automated Fiber Placement (AFP), an advanced manufacturing technology, has contributed to the development of VS composites through its fiber steering capability. This enables to modify load distribution and improve the response of the structure under prescribed loading conditions. However, AFP inevitably introduces defects like gaps and overlaps, which can significantly alter mechanical properties and failure behavior. Gaps, representing fiber-free, resin-rich regions, lead to local stiffness reduction and act as weak spots for damage initiation [1], whereas overlaps cause thickness variations and weight penalties. A commonly used defect modeling approach, the defect layer method [2], accounts for these imperfections by modifying layer properties based on defect area percentages. This study explores alternative numerical strategies for accurately representing defects in VS laminates to improve predictive capabilities of finite element simulations. Two methods are investigated: a field-variable approach to capture the spatial variation of material properties and thickness, and a domain discretization method, where defective and defect-free regions are meshed separately. A comparative assessment of these methods against the defect layer approach is conducted to evaluate their effectiveness in capturing the impact of gaps and overlaps on stiffness, strength, and failure mechanisms.