Information held in working memory (WM) is not always maintained with equal priority; instead, observers strategically prioritize subsets of information that are particularly relevant for upcoming behavior. Historically, strategic prioritization in WM has been induced through distinct experimental manipulations—including test-relevance retro-cuing, differential reward incentives, and instructed refreshing—raising the question of whether these manipulations engage a single unified prioritization mechanism or distinct operational modes. By systematically comparing test-relevance cuing, rewarding, and refreshing within a unified visual working memory paradigm, we evaluate the behavioral consequences of strategic prioritization for both prioritized and unprioritized memoranda, including their susceptibility to perceptual interference. Our findings support a dual-mode account of strategic prioritization in working memory, distinguishing between a focused attentional mode driven by predictive test-relevance cues and a more diffuse prioritization mode engaged by reward and refreshing manipulations.