Did you know? A teacher can write a useful comment on an essay, a pupil can read it carefully, and the next draft can still repeat the same mistake. The missing step is often not effort but translation: what exactly should the pupil change, and what evidence would show that the change worked? “Be more specific” is a beginning, not yet a revision plan.
Turning teacher feedback into a verifiable second draft means converting each important comment into a precise revision action, making that change in the work, and checking the new version against the original learning goal. The fastest route is usually to identify one central problem, revise one representative passage or solution, independently test whether it now meets the criterion, then apply the method to the remainder. More corrections on a page do not automatically mean more understanding.
This practical guide is for students, parents and teachers. We will follow an entirely fictional pupil through a first draft and a better second draft, check a numerical claim by hand, examine feedback on mathematical units, and practise a repeatable method that works across English, Mathematics and Science. None of the examples is an account of actual learner results or a promise of examination marks.
Why the second draft is the real test of feedback
A margin note is information. Learning becomes visible when the student can act on the information and explain the improvement. A comment that says “Explain your evidence” might be accurate, but a child who does not know what counts as evidence may add adjectives instead of a reason. The page becomes longer while the central gap remains.
The Education Endowment Foundation’s Teacher Feedback to Improve Pupil Learning guidance (2021) emphasises laying strong instructional foundations, using appropriately timed feedback to move learning forward, and planning how pupils will receive and use it. The principles matter more than the delivery medium alone. This article takes that emphasis into a concrete pupil-facing process; its sample drafts and arithmetic are original hypothetical teaching examples.
Start by naming the learning goal. An English paragraph may need a supported, carefully qualified claim. A Mathematics solution may need the right quantity and unit. A Science conclusion may need to distinguish observation from explanation. The same vague feedback phrase cannot be applied mechanically to all three. Effective revision is a diagnosis of the particular task, not a ritual of crossing things out.
A full worked example: the first draft overclaims what a small classroom record shows
Imagine a fictional task in which a pupil is asked to explain whether a class’s quiet-start reading routine appears promising. The pupil writes: “Quiet reading helped everyone concentrate. Everyone started work faster because they read first. This proves that our school should use the routine every day.” The teacher writes: “Use the data, distinguish what you observed from what you inferred, and avoid claiming everyone improved.”
For this invented exercise, the data table records the number of pupils who began the reading task within two minutes. In an earlier session it was 9 of 15 pupils; in a later session after a quiet-start routine it was 12 of 15 pupils. The two sessions were not a controlled experiment. Attendance was the same in the exercise, but other conditions such as familiarity with the task were not measured or controlled.
The first sentence is too strong for three reasons. The observations concern starting a task within two minutes, not directly measuring concentration. The counts do not say every pupil improved. And a before-after difference by itself cannot establish that the reading routine caused the difference. A second draft should repair the claim, not just make the paragraph sound more persuasive.
Step one: convert the comment into questions the draft can answer
- What is actually measured? The number of pupils who started the task within two minutes, not the quality of their concentration.
- How many pupils were observed? Fifteen in each session, not all pupils in the school.
- What changed numerically? From 9 of 15 to 12 of 15; three more pupils met the criterion.
- What conclusion is supported? The later session had a higher observed proportion starting within two minutes.
- What remains uncertain? Whether the quiet-start routine was responsible, whether the pattern would recur and how individual pupils responded.
The teacher’s three-part comment is now a set of testable edits. Replace the unsupported universal claim, insert the relevant counts and explicitly mark the limit of the explanation. The pupil should not have to guess what “better evidence” means after this diagnosis.
Step two: hand-check the numbers before writing the new paragraph
For the first session, 9 ÷ 15 = 0.60, so 60% started within two minutes. For the second, 12 ÷ 15 = 0.80, so 80% did. The difference is 20 percentage points; it is also three additional pupils out of the same fifteen. A different comparison, the relative increase in the proportion, is (80 − 60) ÷ 60 = one-third, or approximately 33.3%. Do not write that the percentage-point change was “33.3 points.”
The reverse check is straightforward: 60% of 15 = 9 and 80% of 15 = 12. The figures agree with the invented table. Even perfect arithmetic does not establish cause. The figures describe two sessions, not an experimental estimate of how the routine would affect other children. This is an important separation of calculating accurately from drawing a defensible conclusion.
Step three: write a second draft whose strength matches the evidence
A defensible revised paragraph could be: “In our fictional classroom comparison, 9 of 15 pupils (60%) began their reading task within two minutes in the earlier session, compared with 12 of 15 (80%) in the later session using a quiet-start routine. That is three more pupils, a rise of 20 percentage points. The pattern suggests the routine may be worth testing further, but the two sessions do not prove that the routine caused the difference or that every pupil became more concentrated.”
Notice the actual changes. “Everyone” became an exact, bounded group. “Concentrate” became an observable task-start criterion. The invented measurements replaced an unsupported impression. “Proves” became a conclusion with an explicit limitation. These are improvements in reasoning; the extra words are a consequence, not the goal.
A fair second-draft check asks: Can a reader reconstruct the numbers? Does the claim describe what was measured? Has the pupil distinguished observation from a causal explanation? Does the paragraph state the limit without becoming so vague that it says nothing? The new version meets those checks in this teaching scenario.
The before–after revision ledger
A helpful four-column revision ledger can be drawn on paper: original sentence | teacher’s concern | specific change | proof the change worked. For this fictional paragraph, the entries might be: “everyone concentrated” | overgeneralisation | “12 of 15 started within two minutes” | matches the invented count; “proves the routine works” | unsupported causation | “may be worth testing further” | acknowledges uncontrolled conditions.
The purpose of the ledger is not to produce another long form for the sake of administration. It forces a link between comment and action. A student who cannot name the difference between the original and the new sentence may have revised superficially. A student who can explain why the new evidence constrains the claim is demonstrating a transferable skill.
The parallel Mathematics example: a correct calculation with the wrong label
Feedback is not only for essays. Consider another entirely invented task: a class activity sells ten identical boxes of stationery. The total sales revenue is $120 and the total stated costs are $68. A pupil writes: “Profit is $120 − $68 = $52, so each box earns $52 profit.” The teacher comments: “Check what your number represents; distinguish total profit from profit per box.”
The subtraction is correct: 120 − 68 = 52 dollars total profit under the task’s simplified assumption that all relevant costs are included. But the phrase “each box earns $52” mislabels the unit. A revised solution says: “The combined profit for ten boxes is $52. If we calculate average profit per box over those ten boxes, $52 ÷ 10 = $5.20 per box.” Verify: 10 × $5.20 = $52.
The condition about averaging matters. It does not tell us whether the actual profit on each individual box was equal, because we have only aggregate revenue and cost. If individual selling prices or costs differ, the average remains $5.20 but the per-box profits may vary. Correcting the label exposes what the data allow us to say. This is more valuable than erasing a correct subtraction and repeating it.
A second-draft checklist for that solution would ask: Is the answer total or per item? Are dollars and boxes labelled? Are all relevant costs included in the model? Does the division by ten answer the average question? Can the result be reversed? The teacher’s short note becomes a series of meaningful checks that the learner can eventually apply without prompting.
A science comparison: the difference between an observation and an explanation
An invented school investigation records that a sealed cold metal bottle develops droplets on its outside. A pupil writes, “Water escaped from the bottle because it was cold.” The teacher says, “Check whether your explanation is supported by the setup.” A better draft might say: “Droplets appeared on the outer surface of the cold bottle. Condensation of water vapour from the surrounding air is a plausible explanation when the external surface is sufficiently cool, but further checks would be needed to rule out water spilled onto the surface or a leak.”
The revision is not an excuse to claim that all droplets everywhere are condensation. It separates what was seen from the mechanism proposed and suggests a way to discriminate between causes. A teacher could ask for a dry, sealed container and an appropriate comparison, following normal safety guidance. The pupil need not invent a temperature measurement that was never taken. Accurate language is part of accurate science.
How to turn five common comments into visible work
- “Be specific.” Identify the quantity, person, time, condition or passage detail that would make the claim checkable. If the evidence is unavailable, narrow the claim rather than inventing support.
- “Explain your reasoning.” Insert the missing link between evidence and conclusion. Ask why the stated observation supports the particular conclusion rather than a competing one.
- “Check units.” Label every value and result. Verify whether the number is a total, a rate, an average, a duration or a measurement.
- “Improve organisation.” Give each paragraph a job; use a reverse outline to spot repeated points and missing steps instead of merely adding transition words.
- “Consider another interpretation.” Identify one plausible alternative, test it against the same evidence and explain what remains undecided.
A pupil can test whether these actions worked by hiding the teacher’s comment, rereading the revised work, and asking which observable difference addresses the original problem. “It sounds nicer” is not enough. The explanation should point to the new evidence, changed reasoning or corrected unit.
A learner diagnostic: can the pupil name the revision without adult prompting?
Present this fictional first draft: “The school garden is the best place to learn because it is amazing.” The teacher writes “Give a reason a reader could check.” Ask the student to identify what is missing before writing a replacement. An effective answer notices that “amazing” communicates enthusiasm but not a checkable benefit. A possible revision might name an observable activity, such as counting seedlings or comparing soil moisture, if that activity actually occurred. Without evidence, students should describe it as a proposed activity, not an accomplished outcome.
A student who merely substitutes “excellent” for “amazing” has changed vocabulary without addressing the core feedback. A student who can say, “I need a claim about a particular activity and evidence that it happened,” has made the important diagnostic step. A teacher can then model one short example and withdraw help on the next sentence.
Progressive practice: guided, shared, independent
Level one — repair an overlarge claim
First draft: “All pupils preferred the new game.” The fictional poll shows 14 of 20 respondents chose the new game. What should change? A supported revision is “Fourteen of the twenty respondents (70%) chose the new game in this poll.” The calculation 14 ÷ 20 = 0.70 is exact. Do not assume the six others all disliked it, or that the poll represents the whole school.
Level two — answer the actual question
An invented Mathematics question asks for the time for a journey of 12 km at a constant speed of 4 km/h. The pupil writes “3 kilometres.” The teacher writes “Check the quantity and its unit.” Recalculate: time = distance ÷ speed = 12 km ÷ (4 km per hour) = 3 hours. The number three was right, but “kilometres” answered the wrong kind of question. The revision should explain why the units reduce to hours.
Level three — improve the reasoning, not the decoration
First draft: “The graph proves pupils study harder in April because the blue line is higher.” Suppose the graph’s blue line represents the number of books borrowed, with March 20 and April 28. A revised claim is “Library loans increased by eight books, or 40% relative to March, between these two months in the example. Borrowing more books does not by itself establish how hard pupils studied.” Check: 28 − 20 = 8, and 8 ÷ 20 = 0.40.
Level four — transfer to an unseen task
Ask the learner to take a different piece of work and choose one teacher comment without copying the earlier model. They must write the criterion, show one original passage or step, perform one targeted revision, and explain how the change is independently checkable. If the pupil cannot do this, return to a simpler matched example rather than demanding a wholesale rewrite.
When a second draft should not simply follow the comment literally
Teacher comments are valuable professional judgments, but they can be brief, ambiguous or mismatched to a pupil’s current knowledge. “Use more evidence” may be unhelpful when there is no additional evidence; the right response can be to qualify the claim. “Make it shorter” may conflict with a requirement to show complete mathematical working. If instructions appear inconsistent, ask the teacher which criterion has priority instead of quietly deleting something essential.
Not every learner needs written corrections on every line. Some need a short spoken explanation, a model showing one corrected step, or a different practice task that rebuilds an underlying concept. The EEF guidance stresses planning how pupils use feedback, not simply increasing the number of comments. The objective is a change in learning, not a longer marking record.
Respect the original writer. Save the first draft or photograph it before revising, if permitted. Show the changes honestly rather than silently presenting an adult’s replacement paragraph as the pupil’s own work. For formal graded tasks, follow the school’s rules about allowable assistance and resubmissions. The aim is for the learner to be able to produce the improvement independently.
A practical ten-minute feedback conference
- Minute 1–2: pupil reads one comment aloud and paraphrases it in ordinary language.
- Minute 3: both identify the exact first-draft sentence, step or data label that the comment concerns.
- Minute 4–6: pupil makes one correction, with a model only if necessary.
- Minute 7–8: pupil checks that correction against the task criterion and original evidence.
- Minute 9: adult asks why the new version is stronger and what is still uncertain.
- Minute 10: pupil applies the principle to a fresh, nearby sentence or calculation without a prompt.
Those timings are an illustrative structure, not a studied intervention or a rule for all ages. A child who needs longer should receive longer. The conference can also be done with a written checklist when the learner already knows how to use it. If a concept is missing entirely, teach the concept instead of trying to patch the draft with a formulaic phrase.
How to measure genuine improvement
Use three small indicators: revision accuracy (the new answer addresses the stated criterion), explanation (the pupil can name why the change matters), and transfer (the pupil applies the same idea to a new example). Keep one before-and-after sample and a later independent attempt. Scores can be useful if their rubric and task difficulty are comparable, but a higher score alone does not reveal the mechanism of improvement.
For the fictional quiet-reading paragraph, the observable improvement is the corrected numerical statement and the removal of unsupported causal certainty. For the profit problem, it is the distinction between $52 total and $5.20 average per box. In both cases, the second draft is checkable in a way that the first version was not. The learner can point to the difference without relying on praise alone.
Frequently asked questions
What if I cannot understand the teacher’s handwriting or comment?
Ask for clarification. A useful question is, “Could you show me one place where my claim is stronger than my evidence?” or “Which unit should the final answer represent?” It is better to ask one precise question than guess and rewrite the entire page in the wrong direction.
Should I fix spelling first or reasoning first?
It depends on the task, but when a sentence’s logic is wrong, revising that logic before polishing spelling often avoids editing words that will later be removed. Spelling and punctuation still matter, especially when they affect meaning or are part of the assessed objective.
Is copying a model paragraph the same as revising?
No. A model can make a method visible, but the next task should ask the pupil to apply the method to their own evidence. If the learner can only reproduce the example, the teacher may need to reduce support more gradually.
Can a revised draft be longer but worse?
Yes. Extra examples can distract from the question, introduce unverifiable claims or repeat points. Check whether each new sentence performs a necessary job. The guide on using a reverse outline to find missing reasons can help distinguish useful additions from repetition.
How should a parent help without writing the answer?
Ask the child to name the teacher’s concern, locate the relevant sentence and explain one intended change. After a short discussion, let the child make and check the revision. If adult wording is used in a formal assignment, follow the teacher’s collaboration rules and disclose assistance where required.
The larger educational purpose
The value of feedback is not the red ink, the coloured comment bubble or the number of corrections. It is a learner who can spot the gap, choose a meaningful repair and test the result. This diagnosis-first, explanation-first and independent-checking progression is also at the heart of eduKateSG’s immutable Clementi Mathematics teaching reference. For connected lessons, read revising a topic sentence to match evidence, then explore the Education Hub and Core Aim learning hub.
Takeaway: After a teacher comments, do not begin by adding words. Begin by asking, “What exactly is wrong, what will I change and how will I prove the new version is better?” The strongest second draft is one the student can defend and improve again without someone else holding the pen.