Researchers have long expressed their discontent with the existing measures of skill mismatch. This paper argues that traditional measures cannot fully capture the essence of an inherently multidimensional and job-specific concept such as skill mismatch. An empirical job-based methodology is proposed that classifies the types of skill (mis)matches based on performance of core skills and supplementary skills. The proposed methodology is tested on a sample of 600 Portuguese retail bankers. The results support the job-specific nature of skills. The differentiated distribution of skill (mis)match by job is a significant determinant of earnings among the sampled retail bankers.