How to give feedback on a pile of essay drafts

Essays are the heaviest marking there is, because the feedback is the product, not a by-product. Craft that protects the weekend without shortchanging the student.

Written by the Markpilot team, practising secondary teachers · July 2026

A pile of tests and a pile of essays are different problems. A test wants a mark; the feedback mostly writes itself from what was right and wrong. An essay wants a reader: every script is a bespoke read, the mark is the least useful thing you produce, and what the student actually needs (what to do differently in the next draft) has to be composed, one student at a time. That is why a class set of drafts can absorb a whole weekend, and why the habits below are all about spending your reading once and making every written line earn its place.

Read once, decide at the end

The expensive habit is the double pass: read to get a feel, then read again to mark and comment. Discipline yourself to one pass: annotate in the margin as you read, then make the banding judgment and write the summary comment immediately at the end, while the essay is still loaded in your head. If you use a rubric, have it beside you during the read, not after; you are collecting evidence against criteria as you go, not reconstructing it from memory later.

Comment on the pattern, not every instance

Marking the same comma splice fourteen times teaches the student nothing the second mark didn't, and it costs you thirteen annotations. Mark the first one or two instances, then one note: "this happens throughout · find the rest." Same for unsupported claims, drifting paragraphs, quote-dropping. The instance belongs in the margin; the pattern belongs in the summary comment; and a pattern shared by half the class belongs in a ten-minute whole-class session, not in twelve private paragraphs you wrote near midnight.

Put the effort into the draft, not the final

Feedback on a final is an autopsy: thorough, fair, and largely unread, because nothing the student does with it changes the grade it sits under. Feedback on a draft is a plan, and it is the highest-leverage writing you do all term. If your assessment design allows any redraft at all, shift your effort forward: rich feedback on the draft, brief feedback on the final that names what improved. Students read draft feedback because acting on it pays; honouring that in the final ("you tightened the analysis in body two, exactly as planned") is what teaches them feedback is worth acting on.

Quote their writing back

The most-read sentence in any feedback is the student's own, quoted. "Your opening line, 'Macbeth is a play about ambition', promises a general essay; your actual argument about guilt is more interesting · start there" lands in a way that "strengthen your introduction" never will, because it is proof you read them. One quoted phrase per piece of feedback is a good discipline: it anchors the praise or the push to something real, and it is the single strongest antidote to feedback that reads like it could sit under anyone's essay. That is the test, incidentally: feedback that would be true of any essay in the pile is worth nothing to the one it's under.

One main step

An essay draft can have nine things wrong with it. The student can act on one, maybe two. Choose: what single change most improves the next draft? Lead with it, be concrete about what doing it looks like, and let the minor issues live in the margins or wait for the next cycle. Feedback lists longer than two items convert to zero action; every experienced marker has watched it happen.

Where AI honestly fits

We build a tool in this space, so weigh this accordingly. Essays are also where teachers are rightly most sceptical of AI, so the honest version matters:

• Reading handwriting and organising evidence is real: transcribing handwritten scripts and lining the writing up against your rubric's criteria, with the relevant lines quoted, is work a machine now does well, and it is exactly the mechanical part of the read.
• Drafting per-criterion feedback works when it is grounded in the actual script (quoting the student's own sentences, banded against your rubric) and you edit it. Ungrounded, it produces fluent generalities, the "could sit under anyone's essay" feedback that students detect instantly.
• The judgment stays yours. Where a script sits on a band boundary, whether the argument actually holds, what this particular student is ready to hear: a tool that quotes its evidence and flags its line-ball calls for you is helping; one that just hands down bands is marking your class without you.