Climate sensitivity and feedbacks have traditionally been examined as a zero-dimensional problem, neglecting the crucial patterns of response and forcing. This limitation has long hindered the acquisition of robust, patterned climate information essential for informed decision-making. In this study, we strive to develop an innovative pattern-aware feedback framework by solving an inverse problem in a finite-dimensional space using a data-driven, optimization approach. The resultant reduced-order representation of the original system not only affords a feedback framework to quantify the patterns of the climate response but also reveals the most excitable—and thus most robust—modes of the climate system. The predictive power exhibited by the reduced-order model shows promise for optimizing climate forcing for certain research applications.