Feedback has a shelf life

Late feedback is a verdict rather than an aid.

Feedback that arrives while the learner still remembers their reasoning can change what they do next. The same feedback three weeks later is received as a grade with commentary, and the learner has moved on.

In self-paced courses this is harder than it looks, because submissions arrive continuously rather than in a marking period, and a small team can be permanently behind. That is a resourcing question that should be settled at design time rather than discovered in delivery.

Automated feedback covers more than people expect if it is written per option rather than per item. Telling a learner why the answer they chose is wrong, specifically, is far more useful than telling them the correct answer, and it is written once.

For work that genuinely needs human response, being explicit about turnaround and meeting it matters more than the depth of the response. Learners tolerate a modest response delivered when promised considerably better than a thorough one that arrives unpredictably.

It is worth stating a turnaround commitment publicly and building the resourcing to meet it rather than leaving it implicit. Learners tolerate a stated ten working days far better than an unstated three that occasionally becomes fifteen, because the uncertainty is what makes them chase, and chasing consumes more staff time than the marking.