A student who finishes a three-hour question-bank session unable to name the two topics most likely to cost them marks hasn’t been studying—they’ve been managing anxiety with a timer. The score reflects it: same errors in month three as in month one, just more of them. IB Math Questionbank is built for the opposite approach—custom sets with consistent filters and markschemes—but most students use it as an unfiltered problem feed and skip the features that would actually tell them something. Without a deliberate workflow that turns each session into diagnostic evidence, more practice doesn’t close the gap to a 6 or 7. It just makes the gap more familiar.
Week 1 — Create an Honest Skills Map
In week 1, your only job is to map where you actually stand. Use the custom test tools in IB Math Questionbank—filter by syllabus topic and mid-range difficulty—to build focused sets for every major topic cluster, not just the comfortable ones. For each cluster, work a consistent number of questions and record two numbers: raw accuracy and a simple consistency rating under light time pressure. Sort clusters into three tiers: frequent errors, inconsistent under time pressure, and solid ground you can safely deprioritize. If more than six clusters land in frequent errors, select the four with the biggest mark share for focused work in weeks 2–3 and park the rest for weeks 4–5.
The problem with a single untimed session is that it measures your best-case performance on a calm day, not your reliability when conditions tighten—which means it can easily misrank the topics that actually matter. To get a map you can trust, run two mini-sets per topic cluster on different days: first untimed for accuracy, then lightly timed with the same question count, so comparisons across clusters are valid. Define light time pressure as a per-question pace you can sustain without guessing: fast enough to stop endless self-correction, not so fast that method breaks down. A topic earns a place in your top remediation tier when it’s repeatedly error-prone across both sets, or when it holds untimed but reliably deteriorates under that same mild pace rule. If it only felt harder, that’s not enough.
Weeks 2–3 — Target Weaknesses with an Error Log
Weeks 2–3 turn that skills map into targeted repair. Each study day, aim for roughly 40–60 questions spread across two or three high-priority weak clusters rather than extended runs on a single topic. As you mark, classify each error—conceptual gap, procedural slip, method-selection failure, reading error, or notation—before deciding what to do next. A large systematic review of 50 classroom experiments on retrieval practice found that students learn more when testing is paired with feedback and when they process why errors happened, not just when they answer more questions. What that finding implies is that the quality of the feedback is the operative variable—and error-type classification, not mere logging, is what gives feedback the specificity to change what you practice next. Knowing you got a question wrong tells you nothing actionable. Knowing whether it came from a concept gap, a method-selection failure, or a notation slip does, because each one requires a different response. Those same entries also make your weekly review a factual check—either the dominant error type shifted or it didn’t—rather than the familiar loop of trying to gauge progress by feel.
- During marking, write one fast log line for every missed or shaky question: topic cluster, prompt type (for example, calculation, proof, interpretation), what you actually did, why it failed in plain language, what trigger you should have noticed, and a brief next action.
- For each logged item, classify exactly one main error type: concept gap, procedure slip, recognition or method selection, interpretation or reading, or communication and notation where you lost marks despite mostly correct math.
- Map each error type to a specific next-day action before you start the next set: concept gaps mean stop drilling, study a short explanation or worked markscheme, then redo two to three near-variant questions; procedure slips get a timed micro-drill of six to ten similar problems aiming for identical setup and clean algebra; for recognition or method-selection errors and for interpretation or reading errors, use the markscheme to identify the exact trigger or condition you missed and base your next action on that insight; communication errors get a rewrite of your final solution in clear markscheme style followed by two questions done with extra attention to justification and notation.
- Keep each log entry under about 60 seconds; if you cannot describe why you missed the question that quickly, use the markscheme to identify the real failure mode before you write.
- In the last five minutes of every session, choose tomorrow’s first ten questions based on the dominant error type in today’s log so your next set automatically targets the pattern, not whatever feels most comfortable in the moment.
Weeks 4–5 — Train Exam Thinking Without Labels
Once your worst topics have been patched in isolation, the next challenge is one most students skip: training the identification step that determines whether you can even begin a solution under exam pressure. Real IB papers don’t label their questions by topic. A recent classroom study in a calculus course comparing blocked, topic-by-topic homework with mixed, interleaved sets found that mixed practice better supported performance on exam-style assessments. The reason isn’t mysterious—topic-blocked drills teach you to solve problems that have already been classified for you. Mixed sets force the prior step: figuring out what kind of problem you’re actually looking at.
Build your sets from the question bank by combining several topic clusters and hiding the labels even if you used them to filter initially. Front-load questions from formerly weak areas—those topics need the reps. Solid areas still appear as occasional stability checks, and include a smaller share of questions from stretch areas that are not yet fully comfortable so they get practice without dominating the set. After a few sessions, check specifically how you’re performing on those formerly weak topics when they’re unlabelled. If the same error patterns from weeks 2–3 still show up, pull the topic out, run it through the error-to-action loop in isolation, and reintegrate it once it holds. Getting a question right on a mixed set is genuine progress—but it still isn’t the same thing as scoring marks under full exam conditions, against the clock, on a complete paper.
Weeks 6–8 — Timed Papers as the Main Event
In the final three weeks, full timed papers become your main training format and question-bank sets shift to a supporting role. Sit fully timed papers at regular intervals across the three weeks, then mine each script for targeted corrections rather than simply scheduling another paper. A multi-course study of spaced retrieval in introductory STEM classes found that spacing practice delivers its strongest gains when each round is followed by feedback and focused follow-up study—not just additional attempts. Timed papers give you the spacing and full-format load; short, question-bank patch sessions in between convert those paper errors into durable changes. After marking a paper, pull a focused patch set from IB Math Questionbank using the specific topic clusters where you lost marks, then use the weeks 2–3 error log categories to decide whether each issue needs re-explanation, a micro-drill, better recognition triggers, tighter reading of conditions, or more precise communication. Pay close attention to prompt type—whether a question asks you to calculate, interpret, or justify—since misreading that distinction often drives errors logged under interpretation or communication. When two consecutive papers show no new content weaknesses, that’s the signal to stop remediation and spend the final week on speed and mark-earning precision. Without a measurement cadence tying all of this together, the paper phase risks becoming its own form of unstructured practice—more attempts, same errors, less time remaining to fix them.
- After every session, spend about 30 seconds logging four items: the set type you just did (blocked, mixed, timed paper, or patch), your accuracy percentage, the dominant error type from your log, and one sentence on what you learned.
- Once a week, scan the last five to seven entries for each priority topic cluster so you are reacting to trends rather than single highs or lows.
- If accuracy is rising and the dominant error type for a topic shifts from concept or recognition problems toward procedure or communication issues, keep that topic in rotation but raise difficulty slightly or move a larger share of its questions into mixed, unlabelled sets.
- If the same error type appears in two or three consecutive sessions for a topic, stop adding volume and change the intervention instead—for example, a deeper concept review, more precise recognition triggers, or a stricter reading checklist before you start new questions.
- After each timed paper, patch only the top two or three topic clusters that cost you the most marks; lower-impact issues wait until a later paper confirms they’re persistent.
- When two recent papers show that most remaining marks are being lost to timing or communication rather than new content mistakes, stop introducing fresh topics and spend the final week on speed, concise solutions, and markscheme-style presentation.
Eight-Week Workflow as an Evidence-Driven System
What makes this plan work isn’t the eight weeks or the question volume—it’s that the plan is self-correcting. Each phase’s output adjusts the next phase’s input, so the revision stays calibrated to actual weaknesses rather than drifting toward comfort. Question banks such as IB Math Questionbank are the raw material; the sequence of diagnostics, error classification, and deliberate adjustments is the system that gives that material traction. Start week 1 with an honest skills map, and from that point you’re not just practicing—you’re practicing with information. That distinction is what separates a grade that moves from one that stubbornly doesn’t.