Employee self-ratings are often a required and important part of performancemanagement programs. We asked self-raters to provide in-depth descriptions of their thought processes when rating their own work performance using either a relative (i.e., comparative, percentile-based) or a conventional absolute rating process. Thematic analysis indicated that relative self-ratings typically engendered a two-step evaluation process: initial consideration of one's performance, then comparison to a reference group. Conventional absolute self-ratings tended to rely on a single-step, less comprehensive process. Relative self-rating mean scores were significantly lower than were absolute self-ratings. Our results provide further insights into the mechanism that may underlie relative ratings' psychometric advantages as previously found in a variety of domains.