Spend long enough watching students work through extended IB questions and a pattern becomes hard to miss: they start at part (a), move steadily through the low-mark entry steps, and then hit the highest-mark sub-questions with twenty minutes left and no plan for them. The marks were there to read before any working began—they just weren’t consulted. Extended questions in AA HL—Section B problems and Paper 3 investigations—are documents whose parts have been deliberately ordered, weighted, and phrased, and those choices carry information before you touch your calculator or pen. Marks attached to each part, the sequence of command terms, and the mathematical objects named in the stem all signal where the real thinking will happen and where quick, accessible credit sits. Students who notice this stop treating part (a) as the starting gun; they read the architecture first, and only then decide where to invest their time.
The First-Reading Protocol
The decision that governs time allocation on a Section B problem isn’t which technique to reach for first—it’s what you read before writing anything at all. That first read has three steps. Scan the mark distribution and locate where the big clusters sit: a run of 1–2 mark parts followed by a 5–7 mark part signals that early steps are quick entries and the main time-sink lies later. Read the command terms in order next—a chain like “show that … hence … hence” describes one argument built in stages, while repeated “find” prompts usually mean variations on a single technique. Finally, read the mathematical objects defined in the stem—functions, parameters, sequences, vectors—and form a one-line hypothesis about the domain and direction of travel.
Timebox the whole read to about 60–90 seconds and convert it into a small architectural map: highlight the heaviest-mark parts, circle any explicit links such as “hence” or “using your result …”, write the stem object as a headline, and add a brief intent—something like “bank early marks; protect time for (d)/(e).” The map takes sixty seconds to build and stops you from discovering the question’s logic by accident, halfway through the solution.
On a historical AA HL practice Paper 1, Section B, Question 10, the stem defines a spring–height model h(t) and builds a five-part chain around that function. A first-read using the three signals gives you a map immediately. Parts (a)–(c) carry [2], [2], [2] marks; parts (d) and (e) carry [5] and [4]—the complexity lives at the end. The early “Find …” prompts ask for basic features of the model, while a mid-chain “Show that …” marks a checkpoint result the examiner expects to be stated and reused. And because h(t) is the named object in the stem, you can predict before any algebra that the whole chain is a function-driven investigation, not a collection of disconnected mini-tasks.
Paper 3 Investigation: Read Backward
In IB Math AA HL Paper 3, the sub-questions are designed to accumulate toward a terminal result—which makes the final prompt the most information-rich signal in the whole investigation. Read from the end first. A practitioner guide, IB Maths AA HL Paper 3 Explained: How to Score High, describes the paper as a 1-hour, calculator-allowed exam with two extended-response questions built around reasoning and clear communication rather than speed. That format means structural reading isn’t just useful preparation—marks are embedded in the architecture itself. Skimming the final sub-questions before you touch part (a) reveals what the early steps are constructing and which intermediate results the paper expects to reuse.
Reading from the end also clarifies which parts are scaffolds and which stand alone. When a later prompt uses “hence,” “using your result from (a),” or “given that you have shown…,” treat the earlier result as a gate: spend time producing something you can carry forward—a definition, an equation, a rearrangement—even if it’s imperfect. State the assumption explicitly—”Let k = … from (b)” or “Using my value of a (even if approximate)…”—so the examiner can follow the method and award marks on subsequent steps. The practitioner guide flags the opposite failure: students who approach Paper 3 like Paper 2, diving into calculations without explanation and losing marks that written intermediate reasoning would have kept. If dependency language is absent, the stall is local—move to the next part that can be started from the stem alone. This navigation rule works cleanly in Paper 3 because the dependency is made explicit by design. Across Papers 1 and 2, the architecture is real but the signals are subtler, and the calibration looks different.
Calibrating Across Papers
The three-signal read applies across all three papers, but what each paper rewards architecturally isn’t the same—and the calibration shifts accordingly. On Paper 1, the non-calculator condition adds a signal to step 3. As you read the stem and early parts, ask whether the operations stay manageable by hand: do expressions simplify, do denominators and exponents stay clean, does part (a) establish a form that controls later algebra? Skip this and it’s easy to follow an algebraically correct route that collapses into fractions you can’t process at speed without technology. One feasibility check before you start is the difference between a method that works and one that merely works in theory.
On Paper 2, the first read should separate where technology is meant to handle computation from where algebraic reasoning needs to be visible. High-mark parts that produce simple numerical answers often carry most of their marks in the setup—reading command terms and mark weights together pushes you toward writing the algebra first, then using the calculator. For Paper 3, move-forward decisions should track dependency language in the architecture, not how stuck you feel: if the structure doesn’t link the parts, you’re allowed to move. Students rehearsing these paper-specific calibrations on completed past problems—using organized libraries such as Revision Village to group practice by paper type—can ingrain the habit before exam pressure forces the decision.
The Architectural Drill
Build the habit in two modes. The first is prospective: before solving a new extended question, run the three-signal protocol and record your structural prediction—domain, hotspot, dependency. The second is retrospective: on completed problems you already know how to solve, revisit only to analyze what signals the examiner embedded and whether your predictions matched. Most exam practice builds answer-production fluency while leaving signal-reading accuracy invisible. A measurement loop closes that gap by making prediction quality explicit and comparable across sessions. The log below gives signal fluency a concrete shape—what to record, how often to review, and what a positive trend actually looks like.
- After each question, record your pre-read time—aim to trend it toward 60–90 seconds.
- After each question, label each prediction (domain, hotspot, dependency) as Correct, Partly, or Wrong.
- Once a week, review five logs: note one signal you missed and one that worked, and add both to a running signal list.
- When pre-read time is settling near 60–90 seconds and your Correct rate is trending up, your structural reading is genuinely improving.
The log tracks structural reading fluency—accuracy and speed of signal recognition—not overall exam performance, so treat it as a skill check rather than a results predictor.
Architectural Reading: A Habit for Success
Every past extended question you’ve already solved is also an examiner document you never fully read—the mark distribution told you where the complexity sat, the command terms described whether parts depended on each other, and the stem named the object that anchored the chain. Reading that architecture first doesn’t require more content knowledge; it requires a deliberate habit, built across Papers 1, 2, and 3, until the sixty-second first-read becomes automatic. The next time you open a Paper 3 investigation, the question worth asking before part (a) is what the final sub-question is actually reaching toward. That read is available to every student. The ones who take it rarely end the paper wondering where the time went.