At 10:47 p.m. on Sunday night, a stack of 82 Year 8 science books sits beside a cold mug of tea. Each page appears to ask for the same impossible service: read closely, score fairly, explain the error, praise the effort, and suggest a next step. By book 31, “next step” can become a sentence written mainly to reassure the marker.
That pile is a teacher workload problem, but the answer is not a universal quota of comments or a heroic weekend. The more useful choice comes earlier: decide what the assignment is for, then choose a marking strategy that serves that purpose.
Two approaches deserve an honest comparison. Approach A marks the complete set in comparable depth. Approach B triages feedback: it identifies the evidence that will change teaching or a student’s next attempt, then uses individual, group, and whole-class responses accordingly. Both can be careful. Both can fail.
Start with the decision, not the pile
Before opening the first book, write one sentence: “After seeing this work, what will I or the student do differently?”
That sentence prevents a common form of marking drift. A teacher starts by checking spelling, notices a weak explanation, adds a note about evidence, spots a missing unit, and ends up responding to every visible feature. The page is covered. The original learning purpose has disappeared.
Use the assignment’s purpose to make the first choice:
- A summative task that must support a report, moderation decision, or portfolio record needs dependable evidence across the relevant criteria.
- A formative task meant to shape tomorrow’s lesson needs a fast read of patterns and misconceptions.
- A draft matters only if students will have time and a reason to revise it.
- A short practice exercise may need a checked answer, a model, or a brief correction rather than a paragraph addressed to every learner.
Tasks can carry more than one purpose. Name the primary one. If a “practice” task is also being used to produce a formal grade, the marking burden is not a personal failure; the task has been asked to do two jobs.
Dylan Wiliam’s useful distinction is between feedback that moves learning forward and information that merely describes performance. Hattie and Timperley’s 2007 framework gives the marker three prompts: Where am I going? How am I going? Where to next? Most Sunday piles grow because teachers try to answer all three for every student, on every criterion, whether or not the student can use the answer.
One counterintuitive finding is worth keeping in view. In Ruth Butler’s 1988 study of sixth-grade pupils, comments without grades produced stronger effects on performance and interest than grades alone; adding grades to comments did not preserve the same advantage. It was one study, not a rule that makes grades irrelevant. The practical point is narrower: more written feedback does not automatically mean more feedback taken up. If students look only for the number, a long annotation can become expensive wallpaper.
Approach A vs Approach B: two marking strategies
Approach A is complete-set marking: read each submission, apply the agreed criteria, and provide individual notes wherever the rubric calls for them. Approach B is feedback triage: scan work against a defined learning goal, sort the response needed, and reserve detailed comments for the places where they can do useful work.
The differences are easier to see side by side.
| Decision point | Approach A: complete-set marking | Approach B: feedback triage |
|---|---|---|
| Core question | What does this individual submission show against each required criterion? | Which evidence should change the next teaching move or the student’s next attempt? |
| Strongest fit | Final assessments, coursework records, moderation, and tasks where every criterion must be evidenced | Low-stakes practice, drafts, retrieval checks, and work followed by revision or reteaching |
| Time profile | The cost is high but visible; a full set can be scheduled and audited | The per-script cost can be lower, but the teacher spends time judging patterns and planning the response |
| What students receive | A personal record with broad coverage and, usually, a grade or rubric profile | A focused individual action, a class response, a model, or a self-check route tied to the same goal |
| What the teacher learns | More complete information about each learner, including unusual strengths and gaps | A quicker picture of common errors and the exceptions that need individual attention |
| Fairness strength | Everyone is processed through the same visible procedure | Students may receive different responses because their next needs differ |
| Main risk | Feedback arrives late, becomes repetitive, or is too dense to use | A quiet misconception is missed, or students read selective feedback as selective care |
| Follow-through required | Accurate records, moderation, and time to explain the result | A planned moment for students to act on the feedback; without it, the saving is mostly clerical |
Approach A has virtues that workload conversations sometimes brush aside. It creates an audit trail. It is easier to defend when a parent, colleague, or student asks how a judgment was reached. It can reveal a student-specific problem that no class pattern predicts. For a final Year 11 essay, a full criterion check may be the responsible choice, even if it takes longer.
Approach B earns its place when timing matters more than coverage. If the next lesson can address a misconception while it is still active, or if students will rewrite a paragraph tomorrow, a targeted response may teach more than a detailed report returned after the unit has moved on. That is the case for efficient feedback, but only when the saved time buys a learning event.
The surprising part is who to prioritise. Triage should not mean starting with the weakest work or the students who are quickest to identify as “needing help.” A page with five unrelated gaps may not offer a useful single next step. A mostly successful response with one widespread reasoning error can tell the teacher what to reteach to everyone. Sort by the decision the evidence can support, not by a ranking of children.
Build a triage routine that students can use
Try Approach B first on a low-stakes task with a clear next move. Do not begin with coursework that determines a qualification.
Write the learning target at the top of a spare sheet. Then read the same question, criterion, or feature across the set rather than giving the first script a full literary analysis. For a short-answer science check, that might mean looking only at the explanation of cause and effect. For an extended essay, the target may be the quality of evidence in the introduction and one body paragraph; the whole essay still needs a professional read if it is being assessed formally.
Give each response one of three routes:
Teach next: the same error appears often enough to shape a mini-lesson, worked example, or retrieval prompt.Prompt individually: the student has a misconception, access need, unusual strength, or partial solution that a class response will not address.Confirm and extend: the work meets the target; return a clear success signal and offer a self-check, challenge, or model for the next step.
The third route is not “no feedback.” Every student should be able to tell what to keep and what to do next. A whole-class feedback sheet can include one anonymous example that meets the target, one common error, and a two- or three-minute correction task. Students can compare their own work with the examples before revising. Avoid naming or displaying identifiable work without permission.
Keep the coding modest. Two or three codes linked to the current task are enough. A code such as E for “evidence needs explaining” is useful when students have seen the key and have a chance to act on it. Seventeen codes create a private marking language that students have to decode before they can learn anything.
Written comments work best when they point to evidence and a move. “Good analysis” is a verdict. “Your claim about taxation is supported by the source detail; now explain how that detail changed public support” gives the student somewhere to go. In mathematics, “Redo line three and test the result by substitution” is more usable than “careless sign error.” In each case, the comment is short because the thinking belongs to the student.
A practical 30-script workflow can fit into one sitting without pretending that all tasks take the same time. Spend the first pass locating the target evidence and recording the class pattern. Select a small number of anonymous examples that show the pattern clearly. Write the whole-class response before individual notes. Then add personal prompts only where the route requires them. If the class response cannot be delivered in the next lesson, pause: triage has lost its main advantage.
Put guardrails around the shortcut
Triage fails when the work is the only evidence for a high-stakes judgment, when every criterion must be documented, when students have no chance to use the response, or when a learner’s need cannot be represented by the class pattern. It also fails when a school quietly treats “different feedback” as “lower expectations.” The same learning target and standard should remain visible even when the response route changes.
In those cases, use a hybrid rather than forcing a choice. Mark the required criteria for everyone. Add a brief class response for repeated issues. Replace long comments on every line with one precise individual action. Ask students to use that action in a revision, a correction, or a short conference. This preserves the record that Approach A provides while avoiding a second essay written by the teacher.
School leaders can make this easier by agreeing which tasks are summative, which are diagnostic, and what students will do with feedback afterward. A policy that demands full written comments on every exercise creates teacher workload without necessarily creating more learning. A shared expectation such as “one assessed task per unit gets full criterion feedback; practice tasks receive a class response and selected prompts” is not a magic formula, but it gives teachers permission to match effort to purpose.
Audit the first trial by asking four plain questions:
- Could every student identify one accurate success?
- Did every student have a specific next action or extension?
- Did the class response address the error that actually appeared, rather than the one the teacher expected?
- Did students use the feedback in a later attempt?
Do not judge the method by the number of ink marks or the neatness of the spreadsheet. Judge it by the next piece of work. If the same misconception returns untouched, the comments were not efficient, however quickly they were written. If students revise more successfully and the teacher gets a clearer signal for planning, a thinner set of annotations may have been the more rigorous choice.
For the next low-stakes assignment, write the primary decision at the top of the first page, set a 20-minute scan before you begin detailed marking, and schedule the revision or class response before the pile arrives. Let that small trial decide which approach deserves your Sunday night—and which part of the pile can stay out of it.