Background: Given the high prevalence of low back pain and its financial weight on the healthcare system, practicing physiotherapy exercises is crucial for sustainable therapy success. Effective rehabilitation requires high-quality movement execution, demanding technology that provides optimal feedback. Objective: Our interdisciplinary approach, combining neuroscientific insights on body representation, sports science findings on motor learning, and user experience research on feedback perception aims to give a multifaceted insight into the different effects that varied feedback modalities (auditory, haptic, and combined), have on the performance of a physiotherapy exercise, and on the cognitive workload and body image representation of the patients during the exercise itself, in people with and without nonspecific back pain. Methods: This study employs a mixed-methods design to investigate the impact of different feedback modalities (auditory, haptic, and combined) on physiotherapy exercises. In a quantitative Wizard-of-Oz experiment (n=57), participants performed bent knee side planks while receiving feedback—secretly provided by physiotherapists via a smart shirt. Outcome measures included cognitive workload (NASA TLX), body image representation (Body Map Task), and exercise improvement (physiotherapist evaluation). Control variables such as trust in technology (Surgical Robot Trust & Trust in Automation Questionnaires) and pain level (Chronic Pain Grade Scale) were also assessed. A semi-structured interview gathered qualitative insights into participants' feedback perception and usability. Results: Results indicate no significant differences in cognitive workload or body image representation across modalities, though qualitative data suggest a preference for haptic and combined feedback over auditory alone. Performance outcomes did not significantly vary across conditions, but qualitative insights highlight the benefits of multimodal feedback in enhancing movement perception and engagement. Conclusion: Findings suggest no single optimal feedback modality, but combining haptic and auditory cues enhances usability and motor learning. Participants favor this approach, initially relying on auditory feedback and then switching to haptic feedback in the long-term. Despite not clearly emerging from quantitative statistical analysis, these results support the development of a multisensory feedback strategy. Our interdisciplinary approach demonstrates that multimodal feedback is not only beneficial but necessary for designing adaptive, accessible, and effective rehabilitation technologies.