Consumer-grade video capture has made detailed performance evidence cheap and abundant in higher education settings, but the availability of evidence is not the same as its use. The open question is not whether video can support learning in principle, since controlled studies report gains from self-regulated video feedback, but how learners interpret recorded evidence, select a next action and reduce dependence on instructor prompts. This paper proposes a dual-loop model for mixed-ability university classes. The first loop, Capture–Compare–Commit–Check, structures the immediate performance problem as an information cycle: record the event, compare it against an intended referent, commit to a single change, and verify the result. The second loop tracks the development of feedback literacy across lessons through judgment accuracy, action specificity and independent rechecking. The model draws on higher-education feedback theory, self-regulated learning, motor-learning research and domain-specific evidence. It supplies a low-equipment lesson routine, an eight-week integration schedule, a fidelity log, a draft uptake rubric, privacy safeguards and a matched-condition study design that separates the contribution of video from the quality of instructional dialogue. The contribution is the integration of feedback-use assessment into performance assessment, not a claim that smartphone recording is a novel technology. No recordings, survey data or intervention results are reported; the model is intended for piloting and validation before conclusions about effectiveness or fairness are drawn.