Treating IB Math AI SL revision as one continuous stream of practice is the most reliable way to work hard and make no visible progress. The IA, Paper 1, and Paper 2 each reward a different kind of readiness—reliable procedure built under familiar conditions, technique held together under timed pressure, and contextual judgment applied to modeling and statistics problems that don’t name their topic upfront. Practice that blurs those demands doesn’t sharpen any of them.
A three-phase sequence—topic consolidation, integration drilling, then full-paper simulation—works because each phase targets a different bottleneck. Moving through them in order means addressing the right limitation at the right time; jumping ahead means rehearsing skills you haven’t yet built, which tends to be expensive in time and unrewarding in feedback. The gates below give concrete, testable criteria for when you’re ready to move.
- Gate 1: Topic consolidation → Integration drilling. Pick one topic cluster. Complete 12–20 topic-labeled questions untimed, without mid-solution prompts—markscheme only after finishing. You can move this cluster into integration sets when you can consistently choose the correct method without hints, complete the full method with clean working, and explain what any calculator output means in one sentence. If you miss the gate, stay topic-labeled and fix the specific failure type: method-choice confusion, algebra or hand-calculation slowness, calculator setup mistakes, or interpretation gaps.
- Non-negotiable cross-check (all phases): If early Paper 1 questions solvable by hand are repeatedly costing you time or accuracy, schedule non-calculator fluency blocks inside phase one. Do not compensate with more calculator-heavy Paper 2 practice.
Phase One — Topic Consolidation and the Calculator-Fluency Divide
Phase-one consolidation means building reliable accuracy across AI SL’s skill clusters—statistical hypothesis testing and financial mathematics are representative anchors, and the same standard extends to regression modeling, probability distributions, and measurement applications—using topic-labeled practice without the time pressure or contextual framing that full papers layer in at the same time. The consolidation threshold is specific: you can execute the correct procedure when the topic is labeled and no clock is running. Surface familiarity falls short of that. A student who recognizes that a question resembles a t-test but cannot carry the procedure through cleanly without mid-step prompts has not consolidated the skill.
Practitioner examiner analysis of recent AI SL exam scripts identifies exactly what consolidation must address. Four failure modes recur: using regression tools without justifying model choice or commenting on fit; using equation solvers without checking for contextually invalid solutions; treating calculator differentiation or integration outputs as exact without acknowledging their approximation limits; and defaulting to standard graph windows that miss key features such as intercepts or asymptotes. Paper 1’s early questions—designed to be solvable by hand—consistently expose students who cannot perform basic algebra without a calculator. Consolidation must therefore explicitly separate calculator-dependent and by-hand practice, building the habit of recording and interpreting key outputs from the first session onward. Closing those four gaps is what moves a cluster from surface familiarity to exam-ready execution.
Phase Two — Integration Drilling and the Context-Recognition Gap
Integration drilling is the phase most students skip, and it’s the one most responsible for preparation plateaus. Its target is a specific cognitive step: reading a worded scenario, identifying which mathematical domain and tool the scenario activates, and setting up the approach correctly before any calculation begins. Topic drilling doesn’t build this recognition skill automatically. A student who can run a chi-square independence test flawlessly on a labeled question can still stall when a Paper 2 scenario presents the same statistical relationship in real-world language, without naming the test type.
Each session works through a mixed set of unlabeled questions, with any topic-revealing headings covered. For each question, spend a brief, fixed amount of time writing the domain and tool in words—”chi-square independence,” “exponential regression with parameter interpretation,” “binomial distribution and expected value”—and circling the command term; no calculator or algebra until this identification is written. Then move to the full setup and solution. After checking the markscheme, log one line per miss: whether it was a recognition or execution error, and what cue misled you. Open the next session by redoing that quick identification step only on the prior session’s missed items before any new full solutions. The mark logic is direct: correctly identifying the method earns method marks even if the final answer is wrong; choosing the wrong procedure loses both. Recognition in a quiet session with one question at a time is one thing—holding it across a full mixed paper with a clock running is what simulation tests.
Phase Three — Simulation, Archive Sequencing, and Diagnostic Review
Full-paper simulation means sitting complete, timed IB Mathematics AI SL Practice Exams under authentic conditions—Graphic Display Calculator (GDC) available throughout, strict 90-minute limits per paper, no mid-paper consultation. Use older examination sessions and verified past-paper sets for earlier attempts, and reserve the most recent sessions for the final simulation weeks when their feedback is most valuable. After each paper, every lost mark is classified into error categories—contextual misreading, procedural error, GDC setup mistake, or time management failure—so that the dominant category, rather than habit, decides what you practice next.
- For every lost mark, log: Question ID (paper and question number or part) → Error category: misread / procedure / GDC setup / time → One-sentence cause → One matching fix activity.
- 20-minute cap: Stop once you have classified all losses and written one concrete fix per repeat error type. Re-solve a question only if needed to identify the category—not to get it right this time.
- Cadence: After two full papers (or at week’s end), tally which category leads and set next week’s primary focus accordingly.
- Decision rules: Misread leads → start next week with integration drilling (unlabeled, context-first identification before calculation). Procedure leads → return to topic-labeled practice for that cluster until steps are reliable without time pressure. GDC setup leads → targeted calculator-output practice (record key outputs and parameters used, not just the final answer), then re-enter simulation. Time leads → one timed section drill plus an explicit skip/return strategy in the next full paper.
- Proof it worked: The next timed paper should show fewer repeats of the leading category. If it does not, change the fix activity—do not default to adding more papers.
IA Timing and the Week-by-Week Revision Calendar
The IA contributes 20% of the final grade and, if left to the same period as peak simulation, creates a workload collision that compromises both. Target a complete IA draft before integration drilling begins—unfinished IA work competes directly for the time and mental bandwidth that late-stage exam preparation needs, and it rarely stays contained to the hours a student plans for it.
Both the ten-week and the compressed six-week track follow the same three-phase structure—consolidation, then integration drilling, then simulation with diagnostic review. The difference is compression, not sequence. In a ten-week track, consolidation with embedded IA milestones fills weeks one through four, integration drilling with early simulation runs through weeks five to seven, and full simulation with diagnostic review covers weeks eight to ten. In the compressed track, consolidation targets only identified weak clusters in the first ten days, integration drilling and early simulation run concurrently through the middle period, and the final ten days are reserved for recent-session papers with full diagnostic review—IA should already be submitted before this track begins. In both cases, minimum effective simulation volume is measured by the number of completed cycles: a full timed paper, followed by a diagnostic log of every lost mark, followed by one targeted fix activity per repeat error type—not by how many papers a student has taken.
Integrating the Three-Phase IB Math AI SL Revision Framework
Each phase removes a different bottleneck: consolidation builds reliable procedure, integration drilling builds scenario recognition, and simulation with diagnostic review builds paper-management judgment. The practical starting point is identifying which phase your current preparation is actually in—not which phase feels comfortable—and applying its transition criterion before moving forward. The diagnostic loop is what keeps the whole sequence compounding: each session’s output determines the next session’s focus, and a student who runs it consistently finds that later practice sessions are materially sharper than earlier ones, not just more of the same.
