Does attentional control strategy generalize across different visual search tasks? Previous research has failed to observe significant correlations in strategy metrics between different visual search tasks (Clarke et al., 2020), suggesting that strategy is not unitary, or determined by a single trait variable. Here we question just how heterogeneous (non-unitary) strategies are, hypothesizing a similarity gradient account, which holds that strategy does generalize to some degree, specifically across tasks with similar attentional components. To test this account, we employed the Adaptive Choice Visual Search (ACVS; Irons & Leber, 2018a), a visual search paradigm designed to directly measure attentional control strategy. In two studies, we had participants complete the ACVS and a modified, but similar, task with one altered attentional component (specifically, the requirement to use feature-based attention and enumeration, respectively). We found positive correlations in strategy optimality between tasks that do vs. do not involve feature-based attention (r = .38, p = .0068) and across tasks that do vs. do not require enumeration (r = .33, p = .018). Thus, attentional control strategies did generalize across sufficiently similar tasks, although the strength of the correlations was weaker than the within-task test-retest reliability of strategy measure. These results support the similarity gradient account.