Giving Homework Feedback That Students Actually Use

Research on feedback delivers an uncomfortable finding: much of it has no effect, and some actively harms learning. Meanwhile, marking remains one of the largest consumers of teacher time. The gap between effort and effect is not inevitable — it comes from a handful of fixable patterns.

Grades eclipse comments

When students receive a score and a comment together, most read the score, compare it with their neighbour's, and skip the comment entirely. Studies going back to Ruth Butler's work in the 1980s show comment-only feedback produces more improvement than grades or grades-plus-comments. You cannot always withhold scores — many systems require them — but you can separate them in time: comments first, with a task attached; scores released afterwards. If your platform supports delayed mark release, this is the pedagogical reason to use it.

Feedback must cause thinking, not just record judgement

'Good work', 'vague', and 'see me' record a judgement but demand nothing. Effective feedback gives the student a specific next action that they — not you — must perform:

• Instead of correcting the error, locate it: 'There is a sign error between lines 2 and 3. Find it and redo the question.'
• Instead of 'explain more', specify the gap: 'You have described what happens but not why. Add one sentence using the word density.'
• Instead of re-teaching in the margin, set a micro-task: 'Attempt question 4 again using the method from question 2.'

The test of any comment is simple: does it create work for the student? If the teacher does the thinking, the teacher does the learning.

Build in time to respond

Feedback that never gets acted on is expensive decoration. The single highest-leverage change most teachers can make is reserving ten minutes of lesson time after homework is returned — sometimes called DIRT, dedicated improvement and reflection time — in which the only permitted activity is responding to the feedback. No response time, no reason to expect improvement.

Mark less, but mark smarter

• Whole-class feedback: skim the pile, list the three most common errors, and address them once for everyone instead of thirty times in margins.
• Selective deep-marking: mark two questions thoroughly rather than ten superficially, and tell students in advance which skills you will be looking at.
• Self-checking layers: answers for routine questions can be provided for students to self-mark, reserving your attention for the questions where judgement is required.
• Error logs: have students maintain their own record of mistakes and corrections. It moves the archival work to the person who benefits from it.

Automation changes the economics, not the principles

Auto-marked homework and AI-assisted grading can return results in minutes instead of days, which matters because feedback decays fast — a correction received while the attempt is still fresh in memory is worth far more than the same correction a week later. But automation only accelerates the loop; it does not close it. The irreplaceable teacher move remains the same: make the student act on what came back. A platform can tell a student what went wrong within seconds of submission; only classroom culture makes them care.