How to Give Feedback That Students Actually Use to Improve

You've spent two evenings writing comments on 30 essays. Students get them back, look at the grade, and put them in their bag. The feedback never influences revision. The next essay has the same problems.

This is the most common outcome of teacher feedback, and it's not because students are lazy. It's because most feedback — however detailed and well-intentioned — isn't designed in a way that produces improvement. The gap between effort-to-write and impact-on-learning is one of the most frustrating in teaching.

The research on feedback is actually quite specific about what works and what doesn't. Understanding it doesn't mean writing less feedback — it means writing different feedback, attached to different structures.

What Doesn't Work

Grades without comments. A grade tells students where they ended up. It tells them nothing about what to do differently. Students who care mostly about grades treat the grade as the final word. Students who want to improve have nothing specific to improve toward.

Comments without grades. Research by Psychologist Ruth Butler found that when feedback consisted of comments only, students showed more learning than when comments were accompanied by grades. When grades and comments are together, students look at the grade and largely ignore the comments. This finding is robust enough that some schools have experimented with comment-only feedback on formative work.

Praise-criticism-praise sandwiches. The "compliment sandwich" structure reads as formulaic to students who receive it repeatedly, and the actual critical feedback gets buried. Students remember the bread and ignore the meat.

Comments on final products students can't revise. If the essay is submitted, graded, and moved on from, the feedback has no actionable purpose. Students can't improve the specific piece. Unless the course explicitly returns to the same skills on future pieces — and makes that connection explicit — final product feedback is largely wasted.

Vague or evaluative language. "Good analysis" doesn't tell a student what made it good. "Your argument lacks evidence" doesn't tell a student what evidence to add or where. Feedback that describes quality without describing the specific mechanism of that quality doesn't transfer.

What Works

Feedback on drafts before the final product. When students receive feedback at a point where they can still act on it — midway through a draft, after a first attempt, before the final submission — the feedback has an obvious purpose. The research consistently shows that revision-cycle feedback produces more learning than end-point feedback on the same piece.

Specific, actionable language. "Your argument in paragraph 3 makes a claim but doesn't include any textual evidence. The sentence starting 'This shows that' needs a quote from the text before it." This is actionable. A student can open the document and do something specific.

Feedforward rather than feedback. Instead of commenting on what was wrong in this piece, tell students what to attend to in the next piece. "Your next argument should open with the textual evidence before stating your claim — try reversing the order you've been using." Feedforward connects feedback to future work rather than anchoring it to the past piece.

Selected feedback, not exhaustive feedback. Students can't process 20 marginal comments simultaneously, and exhaustive feedback signals that the whole piece is broken. Select 2-3 issues to address. Write about those specifically. This is also more sustainable for teachers — and sustainable feedback systems are the ones that actually get done.

Peer Feedback That Actually Works

Student-generated feedback has more effect on learning than teacher feedback in some studies — partly because giving feedback requires articulating what makes something work, which is itself cognitively valuable, and partly because receiving feedback from a peer feels less evaluative and therefore less threatening.

Peer feedback only works with structure. "Tell your partner what you think of their essay" produces "It's good." A structured protocol — identify one place where the argument is strong and explain why; identify one place where evidence is missing and explain what's needed — produces feedback that can actually be used.

Teaching students what good feedback looks like, modeling it explicitly, and practicing it in class before requiring it for real work are prerequisites. Peer feedback as an afterthought produces peer praise. Peer feedback as a deliberately taught skill produces peer insight.

The Conversation Instead of the Comment

The most efficient and effective feedback tool available to teachers is the one-to-one conversation. A 2-minute conference during independent work time — "What are you trying to do in this paragraph? What's making it hard?" — often accomplishes more than two paragraphs of written comment.

Conferences are possible during writing time, reading time, or any extended independent work period. They require a structure: know what you're looking for before you circulate, prioritize which students need conversation, and have a consistent format (what's working, what's next, what are you going to try). They don't require an appointment schedule.

The conference has one advantage written feedback cannot have: it's responsive. A student can answer back. The feedback adjusts to what they don't understand. That dialogue is where a significant portion of learning actually happens.

Making Feedback Visible

Whatever form feedback takes, students need to do something with it — not just receive it. Exit strategies for making feedback visible include:

• Students write a one-sentence summary of the most important feedback they received
• Students write a goal statement for the next draft based on feedback
• Students code feedback (circle the one comment they'll address first; star the one they disagree with)
• Students self-assess against a rubric and then compare their self-assessment to teacher feedback

Doing something with feedback — any action, any processing — increases the probability that the feedback influences future work. Receiving feedback passively rarely does.

The feedback you write isn't the feedback students use. The feedback that gets used is the feedback that gets acted on.