22 September 2026 · Hugo Cheyne
Your child gets the homework right and drops marks on the exam. It's often not nerves: it's choosing the method. Here's what examiners see and how to fix it.
The homework comes back ticked. The mock comes back a grade lower than you expected. You ask what happened and get the same answer most parents get: "I knew how to do it in the lesson, but the exam was so much harder."
Often both halves of that are true, but "harder" hides what actually changed. The maths on the paper is frequently no harder than the maths in the lesson. What's harder is that nothing on the paper says which maths it is. Working that out is a specific skill, separate from knowing the methods, that lessons and homework rarely practise and that the exam leans on heavily.
This post explains what that skill is and how much of the GCSE Maths exam depends on it. It shows what examiners saw go wrong in 2025, gives you a quick way to check whether it's your child's gap, and sets out what closes it.
Key Takeaways
- A drill helps your child learn how to follow a method correctly. An exam question makes them choose which method to use. Those are different skills, and most homework only practises the first.
- The exam leans heavily on the second. Half the marks on Foundation tier and 60% on Higher are for reasoning and problem solving, where the method usually isn't handed over.
- Examiners see this directly. On one 2025 Foundation question, "very few candidates were able to recognise that Pythagoras' theorem was required", and scored nothing as a result.
- More practice of the same kind won't close the gap. Mixed questions, worked examples, and asking "what is this question about?" before solving will.
- You can check for it at home in about ten minutes, without knowing any of the maths yourself.
A drill hands your child the method. A question hands them a situation and leaves the method to them. Here's an illustrative example showing the same piece of maths written both ways.
A child who can do the left-hand version every time may still freeze on the right-hand one. They haven't forgotten Pythagoras. They've never had to notice, unprompted, that the corner of a photo is a right angle.
The Education Endowment Foundation makes the same distinction in its guidance for teachers. Many "word problems" are just routine questions dressed in a story, and "if students are only required to carry out a given procedure or algorithm to arrive at the solution, it is not really problem solving; rather, it is just practising the procedure" (EEF, Improving Mathematics in Key Stages 2 and 3, Recommendation 3, retrieved 22 September 2026).
Most homework is the left-hand panel. At least half of the exam is not. Neither is real life. A "50-inch" TV is measured corner to corner, and working out whether it fits in the living room is a Pythagoras question that no one labels for you.
A lot. Half the marks on Foundation tier and 60% on Higher are for reasoning and problem solving, and those are the marks where the method is least likely to be handed over.
Every GCSE Maths paper is built from three assessment objectives. They're set nationally, so they're the same for AQA, Edexcel, OCR and Eduqas:
The weightings are fixed by tier (Department for Education, GCSE mathematics: subject content and assessment objectives, retrieved 22 September 2026). Each board reproduces them in its specification, AQA's for example:
One caution before reading too much into this. The split isn't neat. A single question can carry marks from more than one objective, and AO1 includes "set tasks requiring multi-step solutions": long calculations where every step is still routine. So a long question isn't automatically a problem-solving one. What matters is whether the question tells your child the method. When it doesn't, they're being tested on the choosing as much as the calculating.
The practical consequence is that a child who is flawless at drills still has a large share of the marks at risk. And the share grows as they go up: Higher tier gives more of its marks to reasoning and problem solving, not fewer.
The failure examiners describe usually isn't "didn't know Pythagoras". It's "didn't see that this was Pythagoras".
The Eduqas examiners' report for Summer 2025 describes individual questions in unusual detail. Three patterns run through it.
Not recognising the method. On one Foundation question, "very few candidates were able to recognise that Pythagoras' theorem was required to answer the question and so did not gain any marks". These are students who have almost certainly practised Pythagoras. The likeliest reading is that nothing in the question said "Pythagoras", so they never reached for it.
Starting well, then losing the thread. On a Foundation speed, distance and time problem, "attempts at the question rarely used all aspects of the information required to obtain the final answer. It was common to see a correct first step". The same thing happened on Higher tier. A density question needed three stages, and "only a minority of candidates successfully applied all three stages". Many picked up marks for getting one or two stages right.
Stopping one idea short. In a ratio problem about a jar of 10p and 20p coins, most Higher candidates started correctly, then worked out the total as if every coin were a 20p. The examiners' verdict: candidates "failed to appreciate that the 10p coins have half the value of the 20p coins".
In each case the student had the tools. What they lacked was the plan: which tool, in which order, and noticing they hadn't finished yet.
The report's executive summary is blunt about how permanent this is: "Multi-step problem solving questions will always be asked." Its advice is to organise working "in a manner that can be followed, ideally adding a label or sub-heading to their work".
That advice is about marks, not neatness. Mark schemes award method marks for a correct approach, and follow-through marks for "correct working following a mistake in an earlier step" (AQA GCSE Mathematics mark scheme, Paper 1 Higher, June 2022, retrieved 29 September 2026). So if your child works out the cost of one item wrongly in the first step, the total they build from it in the next step can still score marks. But the examiner can only give marks for working they can follow.
Examiner quotations: Eduqas, GCSE Mathematics Examiners' Report, Summer 2025, retrieved 22 September 2026. AQA, Edexcel and OCR publish their own reports too, some only through schools. We've quoted Eduqas because its report is public and unusually detailed.
For the wider picture of where marks go, including misreading the question and skipping checks, see where examiners say marks are lost in our parents' guide.
Because most homework is organised by topic, and the topic heading has already done the choosing.
When a worksheet says "Pythagoras" at the top, your child never has to decide that it's Pythagoras. Twenty questions later they're faster at the calculation and no better at spotting when to use it. The work looks good, which is why the gap stays hidden until a mixed paper exposes it.
Ofsted's review of maths research calls the missing piece conditional knowledge: knowing when and why to use a method, not just how. It warns that "if a problem-solver does not have conditional knowledge, they are more likely to be distracted by the surface features of problems" (Ofsted, Research review series: mathematics, 25 May 2021, retrieved 22 September 2026).
Surface features are what a question looks like: a photo, a ladder, a jar of coins. The structure is the maths underneath. A classic study of physics students found that experts sort problems by the principle needed to solve them, while novices sort them by what the problems are about on the surface (Chi, Feltovich and Glaser, Cognitive Science, 1981, retrieved 22 September 2026). Topic-by-topic homework never forces a student to make that shift, because the structure is printed at the top of the page.
Changing the order of practice does make a measurable difference. In a large randomised trial across 54 Year 8-equivalent maths classes, a follow-up to the studies in our interleaving guide, students who practised with mixed question types outscored those who practised one type at a time by 61% to 38% on a surprise test a month later (Rohrer, Dedrick, Hartwig and Cheung, Journal of Educational Psychology, 2020, retrieved 22 September 2026). Our post on mixing topics in revision explains why that works and why it feels worse while it's happening.
"Can do it at home, can't do it in the exam" covers four different problems. They look the same from the outside and need different fixes.
| What you see | What's actually missing | What examiners reported in 2025 | What fixes it |
|---|---|---|---|
| Can't do even the labelled homework question | The method itself | Topic-level gaps, such as estimating a mean from a grouped table, where "very few" completed the method | Re-teach the method, then practise it |
| Fine on homework, blank on the exam question | Recognising which method to use | "Very few candidates were able to recognise that Pythagoras' theorem was required" | Mixed questions, and naming the method before solving |
| Starts well, stops after the first step | Holding a plan across several steps | "It was common to see a correct first step" | Labelling each stage; studying worked examples |
| Right method, wrong answer | Accurate execution | "The correct method is shown but the execution of the method is incorrect" | Arithmetic fluency, especially without a calculator; re-reading the question |
Examiner quotations: Eduqas, GCSE Mathematics Examiners' Report, Summer 2025.
The second row is the one this post is about, and it's the easiest to miss, because the homework evidence says everything is fine. You can check for it with a short exercise at home.
This is a suggested exercise, not a formal assessment, and you don't need to know the maths. It's a parent's version of the self-check in our interleaving guide, with one difference: it scores naming the method separately from getting the answer.
Naming and solving tell you different things:
It's worth saying something about nerves. They're real, and some children do underperform under pressure. But the test above runs at the kitchen table, not in an exam hall. If the recognition gap shows up there, nerves aren't the whole explanation.
Practising the choice itself: meeting problems whose method isn't labelled, and comparing problems that look alike but aren't.
The EEF's recommendation on problem solving is specific about what helps. Drawing on a US What Works Clearinghouse review, it judges the evidence "strongest in support of the use of visual representations and worked examples, and encouraging pupils to monitor and reflect on the problem-solving process" (EEF, Recommendation 3). Worked examples come straight from that list, and naming the method first and labelling each stage are both ways of monitoring the problem-solving process. Mixing and comparing problems come from elsewhere in the same recommendation. Together they make five habits:
None of this means drills are the problem. Ofsted is clear that "pupils need to be fluent with the relevant facts and methods before being expected to learn how to apply them to problem-solving" (Ofsted, 2021). Drills build the tools. What goes wrong is when drills are the only practice a student gets.
Ask the questions a good tutor asks. You don't need to know the answer to any of them.
The EEF suggests pupils ask themselves a short set of questions while they work. Asked by a parent, they do the same job. Here they're adapted slightly, with the examiners' plausibility check added at the end:
The first two target recognition. The third targets the plan. The fourth prompts your child to draw on problems they've already solved. The last is the check that, in the examiners' words, "very few candidates employed".
Beyond that, one change to the week is worth trying before adding hours. Swap one topic-by-topic homework session for a set of mixed past-paper questions, done without notes. It'll feel harder and the scores will probably dip at first, which is expected. The same logic applies to testing rather than re-reading in every subject.
For more on supporting revision when the content is unfamiliar, see helping when you don't know the curriculum. And if the cover-the-heading test shows a gap that isn't closing, the signs a child needs a GCSE tutor will help you judge whether to bring in outside help.
Our sessions are built on the Tugo Method, and mixing topics is one of its three pillars. In practice, that means a session doesn't work through one topic under one heading. Questions from different topics arrive in an order the student can't predict, which is the condition the exam sets.
That gives the choosing step regular practice with a tutor watching, which makes it easier to spot which of the four gaps is getting in the way. You can see how we teach GCSE and A-Level Maths.
Often because homework tells them the method and the exam doesn't. Homework is usually set by topic, so the heading has already chosen the method. Exam questions arrive mixed and unlabelled. Half the marks on Foundation tier and 60% on Higher are for reasoning and problem solving, which is where recognising the method counts most.
Try the cover-the-heading test above in a calm setting. Strip the topic headings from eight homework questions, shuffle them, and ask your child to name the method for each one before solving anything. If they struggle to name methods at the kitchen table, nerves aren't the main cause. They may still add to the problem, but the recognition gap is the part you can train.
In practice, it's a question that doesn't tell the student which method to use. The official definition (assessment objective AO3) asks students to "translate problems in mathematical or non-mathematical contexts into a process or a series of mathematical processes", and to connect different areas of maths. A long question with a named method is still a routine one.
No. Ofsted's review of the research says pupils need to be fluent with facts and methods before they can apply them to problems. Drills build that fluency. The fix is to add mixed, unlabelled questions alongside them, so your child practises choosing the method as well as using it.
Sit in on a lesson led by one of our expert PGCE qualified teachers.
Book your free lesson
