Research on face matching has traditionally focused on inner facial features under controlled conditions, such as the eyes, nose, and mouth. While this has deepened our understanding of feature-based processing, it lacks ecological validity, as real-world face matching often includes outer features like hair and face color. These features may provide critical context for identifying individuals in more naturalistic settings, yet their role remains unclear. Moreover, previous studies have rarely examined how reliance on different features may vary across individuals with differing face recognition abilities. This study addresses these gaps by incorporating both inner and outer facial features into a face-matching task, creating a more ecologically valid paradigm. Using eye-tracking, we explored how attention to inner and outer features affects matching accuracy and investigated individual differences across participants, including those at the extremes of face recognition ability. 98 participants (including 8 super-recognizers and 3 prosopagnosics) completed the Kent Face Matching Test (KFMT) and the Cambridge Face Memory Test (CFMT+). In trials where participants provided correct answers, there were fewer but longer fixations towards the inner-face features. In contrast, incorrect responses were associated with more and prolonged fixations on the mouth and hair regions. Individual analyses showed that 2 super-recognizers focused more on eyes and nose, while all prosopagnosia participants focused on the mouth and hair. Sustained attention to the mouth and hair worsened performance, underscoring the role of single-case analysis in understanding individual differences in face matching.