Put this into practice: free
The Insight Bay practice portal has graded question sets for every GCSE topic in this article (including ratio, indices, circle theorems and probability) with four difficulty levels per skill.
Open GCSE practice sets →Most students don't fail GCSE Maths questions because the maths is too hard. They lose marks in the same five places, for the same five reasons, year after year. Here is the diagnosis, and a fourteen-day plan that targets the cause, not the symptom.
Grade boundaries on GCSE Maths are closer than most families realise. The gap between two adjacent grades typically comes down to a small handful of marks per paper, which means a single weak topic, met three times across three papers, can quietly decide the grade on the certificate. After years of working through past-paper attempts line by line with students, I see the same pattern: the marks that decide grades are not lost on the exotic final questions. They are lost mid-paper, on five recurring topics.
Two forces make a topic a grade-killer. The first is weighting: exam boards publish how marks are distributed across content areas, and the Higher tier concentrates roughly half of all marks in algebra and in ratio, proportion and rates of change. The second is error type: some topics fail silently. A student who has never met circle theorems knows they have a gap. A student who "knows" ratio but sets up the wrong multiplier feels confident right up until results day.
One more thing before the list. The fix for each topic depends on why it fails. Throughout this article I'll classify errors into three kinds, because they need three different treatments:
Tick every statement that sounds familiar. Be honest: this is a diagnostic, not a test. The result tells you which of the five topics to prioritise and what kind of error you're dealing with.
Ratio is the single most under-respected topic on the Higher tier. Students file it under "primary school maths", sharing sweets in a ratio, while the exam treats it as a fifth of the paper and hides it inside currency, density, speed, recipes, similar shapes, and reverse percentages.
The deep issue is a missing mental model. Students who struggle see ratio as a procedure ("add the parts, divide, multiply") rather than as a multiplicative relationship. The procedure survives one question type; the model survives all of them.
"Ratio questions are about dividing an amount into parts." So when the question gives a difference ("Asha gets £24 more") or a partial amount ("Ben's share is £36"), the learned procedure has nowhere to start.
Always ask: what is one part worth? A 3:5 split with a £24 difference means 2 parts = £24, so one part = £12. Every ratio question (total given, difference given, one share given) collapses to finding one part.
Algebra is the biggest single slice of the Higher tier, but the marks aren't lost on "solve 3x + 5 = 20". They're lost where two or three manipulation skills stack: algebraic fractions, rearranging formulae where the subject appears twice, and index laws with negative or fractional powers.
What makes this topic dangerous is cognitive load. Each individual move (factorise, cancel, cross-multiply) may be secure in isolation. Under exam pressure, holding four moves in working memory at once is what fails. The remedy is not more notes; it is making the component moves so automatic that they stop consuming working memory at all.
Why drilling components works: cognitive load theory (Sweller) shows that fluent, automatised sub-skills free working memory for the novel parts of a problem. Ten minutes a day of mixed index-law and factorising drills does more for multi-step algebra than an hour of watching worked videos, because watching never builds automaticity.
Circle theorems and angle chains are unusual: students frequently get the number right and still lose half the marks. That's because these questions assess written mathematical reasoning, and the mark scheme reserves marks specifically for naming the theorem or property used, in acceptable language.
"Angles in a triangle" is not enough when the mark scheme wants "base angles of an isosceles triangle are equal". The skill to practise is not finding angles: it's writing one line of justification per step, using the exam board's preferred phrasing. This is a learnable, almost mechanical habit, and it routinely recovers two to four marks per paper for students who adopt it.
Train this layout on every angle problem for two weeks. Left column: the statement. Right column: the reason, in board language.
∠OAB = 90°: angle between tangent and radius is 90°
∠ACB = 58°: angle at centre is twice the angle at the circumference
∠ABC = 32°: angles in a triangle sum to 180°
The structure forces the reasoning mark to exist before the next step is taken, so the mark can never be "forgotten" under time pressure.
Probability trees are taught everywhere; what fails is the updating. The phrase "without replacement" requires the second set of branches to change, and a large share of students draw identical branches on both stages, because their mental model of probability is "count the favourable outcomes" rather than "track how the situation changes".
The deeper version of the same gap appears in "given that" questions, where students answer with the unconditional probability. Both errors have the same root: probability is being treated as a static label instead of a quantity that responds to information.
The final grade-killer isn't a topic at all: it's the skill of converting a written situation into an equation or calculation plan. Boards have steadily increased the weight of multi-step, contextual problems, and these questions are where otherwise strong students stall, not because they can't do the maths, but because nobody ever taught them a procedure for starting.
Give students a translation protocol and the paralysis largely disappears:
This plan assumes 45-60 minutes per day. It deliberately covers only two of the five topics: the top two from your diagnostic. Depth beats breadth: two topics made reliable are worth more than five topics made familiar.
| Days | Focus | What actually happens |
|---|---|---|
| 1-2 | Diagnose | One timed past paper (or the diagnostic above plus a topic test). Build the error log: for every dropped mark record the topic and the error type. Gap, retrieval failure, or slip. |
| 3-6 | Topic block A | Daily: 10 min retrieval warm-up (yesterday's questions, from memory) → one worked example studied line by line → 20 minutes of graded questions → log every error. |
| 7 | Interleave | Mixed set: topic A questions shuffled with old, secure topics. Interleaving feels harder and less fluent, that difficulty is the point. |
| 8-11 | Topic block B | Same daily structure as block A. Keep topic A alive with five retrieval questions per day: spacing is what stops the fix from decaying. |
| 12 | Mark-scheme study | Take five past-paper questions on topics A and B, write full solutions, then mark them against the official scheme. Note where method marks live and what phrasing earns reasoning marks. |
| 13 | Timed paper | Full paper under exam conditions. Compare the error log with day 1: the win condition is fewer errors of the same type, not a perfect score. |
| 14 | Review & plan | Update the error log, celebrate the categories that shrank, and pick the next topic pair. The cycle repeats. |
Retrieval practice (testing yourself from memory) outperforms re-reading by a wide margin in controlled studies: it's the most replicated result in learning science. Spacing the follow-up keeps the gain. Interleaving mixed topics improves the skill exams actually test: deciding which method a question needs. And keeping an error log exploits the hypercorrection effect: errors corrected with feedback are remembered unusually well.
The Insight Bay practice portal has graded question sets for every GCSE topic in this article (including ratio, indices, circle theorems and probability) with four difficulty levels per skill.
Open GCSE practice sets →Two, at most three. The limiting factor is practice depth, not study time. A topic is "fixed" when your child can answer unprompted questions on it a week later, that takes repeated, spaced sessions, which is hard to sustain across five topics at once.
Two weeks won't rebuild absent foundations, that is a longer project. But for a student near a boundary, converting one or two leaky topics into reliable marks is frequently the difference between adjacent grades. The plan above is exactly what I run with new students in their first fortnight.
One timed past paper, marked honestly, beats any amount of asking. Classify every dropped mark into never-learned / forgot / slipped / time. The clusters tell you everything. If you'd like a professional version of this, a free assessment session does precisely that diagnosis.
Ratio, probability basics and problem translation: yes, heavily, on both. Circle theorems and algebraic fractions are Higher-tier content (and appear on IGCSE Extended); Foundation students should reinvest that time in number fluency and ratio. IGCSE 0580 students will recognise all five topics: the Extended paper leans even harder on multi-step translation.
The free assessment session is a structured diagnostic: one hour, a marked mini-paper, and a written priority plan like the one in this article. No obligation afterwards.
Book the free assessment