What to Do in the 90 Seconds After a Question Goes Bad
A blank where you expected an answer, a question eating your clock, a mistake you just spotted — here is the concrete recovery routine for each, minute by minute.
Every serious exam eventually hands you a question that goes wrong. Not a hard question — a question that breaks, mid-paper: the blank where you expected recall, the numerical that has quietly eaten nine minutes, the sick realisation that you misread option (b) three questions ago. What separates a good score from a ruined one is rarely the first bad question. It is what you do in the ninety seconds after it.
Name the failure, then apply its rule
Recovery starts with classification, because each failure has a different correct response. A blank — you know the chapter, the answer just isn't there — gets a hard cap: roughly one minute of genuine retrieval effort (write what you do know, work backwards from the options, try to reconstruct the formula from units or a limiting case). If nothing surfaces, flag it and leave. A time sink — a question running past its budget — gets cut the moment you notice, not five minutes later. A spotted error — you realise an earlier answer is wrong — gets a two-second note in the margin and nothing more until you return.
The rule that governs all three: never let a question cost you more than its marks are worth. A two-mark question that consumes ten minutes has already taken marks from three other questions you never reached.
Cut losses on a timer, not on a feeling
"I'll just try one more approach" is how ten minutes disappear. Replace the feeling with a number. Before the exam, decide your per-question budget from the paper's structure — total time divided by number of questions, with a small buffer — and treat it as a rule, not a suggestion. In the exam, when a question crosses that budget, you leave. Not after the next step, not once you've simplified it. Now.
Leaving well matters too. Write down whatever partial working you have — a substituted formula, a labelled diagram, the first two steps of a derivation — because board evaluations award step marks, and a flagged question with working is worth more on return than a blank one. Circle the question number, mark it with a single symbol (say, a star), and move on within seconds. Do not re-read the question as you leave; that is how you carry it with you.
Flag-and-return, done properly
A flag is only useful if returning to it is systematic. Use one symbol for "attempted, incomplete" and a different one for "untouched, skipped" — when you come back in the final stretch, you want to know instantly which questions need fresh thinking and which need finishing. Return in a fixed order: first the incomplete ones with working (they are usually quick to close), then the untouched ones you're most confident about, then the rest.
Reserve the last ten to fifteen minutes of the paper for flagged questions and for a scan of your answer sheet — not for starting anything new. If a flagged question still resists on the second attempt, give it one more bounded try and then let it go. Two failed attempts is your signal, not a reason for a third.
Stop the spillover before the next question
One bad question contaminates the next through two channels: leftover emotion and leftover time. The time part you've handled by cutting losses. The emotional part needs its own deliberate reset, because "calm down" is not an instruction your brain can follow. What works is physical and procedural: put the pen down for five seconds, roll your shoulders, take one slow breath out — longer than the breath in — then read the next question's last line first (what is it actually asking for?), then its data, then its first line. Reading in that order forces fresh attention onto the new question instead of letting the old one keep running in the background.
If you spot a mistake on an earlier answer, resist the urge to fix it immediately. Correcting under stress, mid-paper, is how students turn one wrong answer into two. Note it, keep going, and fix it in the reserved final minutes when you can check the correction properly.
Rehearse the recovery, not just the syllabus
Recovery is a skill, and like any skill it improves with practice under realistic conditions. In your next mock, deliberately practise the routine: enforce the per-question budget with a visible clock, use your two flag symbols, and run the final-minutes return in order. Afterwards, review not just which questions you got wrong but how long you stayed on them after you knew they had gone bad. That number — your recovery lag — is the one to shrink.
The takeaway worth carrying into the hall: a bad question is not a verdict on the paper, it is a decision point. Decide in seconds — cap the blank, cut the time sink, note the error — and the question costs you its marks and nothing more.