We present a crowdsourced study (N=120) that investigated the role of goal-setting on learning during search. To study the role of goal-setting, we developed a tool called the Subgoal Manager (SM). The SM was designed as a simple tool to help searchers deconstruct a learning-oriented search task into smaller subgoals, take notes with respect to specific subgoals, and mark subgoals as completed. During the study, participants completed one learning-oriented search task that focused on learning about the concepts of diffusion and osmosis. After the search task, participants completed a multiple-choice test that measured deep understanding of diffusion and osmosis. The study used a between-subjects design with three subgoal conditions. In the SelfSetSubgoals condition, participants had access to the SM and set their own subgoals. Participants were instructed to set high-quality subgoals that included a specific action, content, standard, and time. In the AssignedSubgoal condition, participants had access to the SM with pre-written, high-quality subgoals. Participants could take notes with respect to specific subgoals and mark subgoals as completed, but could not edit, delete, nor add new subgoals during the search session. Finally, in the NoSubgoals condition, participants were not instructed to generate subgoals and were simply provided with a text editor tool to take notes. We investigate the effects of the subgoal condition on learning outcomes and perceptions of: the task, support for specific self-regulated learning processes, the quality of subgoals pursued during the task, and engagement. Our results found that participants had the best outcomes in the SelfSetSubgoals condition. Additionally, within the SelfSetSubgoals condition, outcomes were even better when participants set high-quality subgoals.