Many AIED systems support self-regulated learning, yet, support for setting and achieving practice goals has received little attention. We examine how middle school students respond to system-recommended practice goals, building on the success of similar data-driven recommendations in other domains. We introduce an adaptive dashboard in an intelligent tutoring system. The dashboard tracks progress (time, mastery), suggests dynamic goal adjustments, and visualizes performance--aiming to increase learning opportunities. In a prototyping study with five middle schoolers (grades 6-7), we used scenario-based tasks to examine reactions to goal recommendations during under- or overachievement. Thematic analysis revealed: (1) students accepted system suggestions as guidance but wanted final control; (2) they valued adaptive recommendations tied to past performance to increase their chance of meeting their goal and pace themselves; and (3) they strongly preferred motivational features (e.g., streaks, celebrations) for engagement. Our study contributes design implications for practice goal-setting in tutoring systems, highlighting personalized, explainable learning analytics and motivational design to support goal-directed behavior.