CAT Attempt Strategy by Percentile: 90, 95 and 99

Two aspirants walk into the same CAT slot with the same attempt sheet in their heads: attempt as many questions as you can, stay accurate. One is aiming for 90 percentile. The other wants 99. They are running the wrong plan, because that single instruction cannot be right for both of them. The number of questions you should attempt, and the accuracy you must hold, depend entirely on the score you are chasing. A CAT attempt strategy built for a 90 percentile target will quietly cap a 99 aspirant, and the reverse leaves a 90 aspirant frozen and under-attempting.

This guide gives you a percentile-target matrix: how many questions to attempt in VARC, DILR and QA for 90, 95, 97, 99 and 99.5 percentile, the accuracy each one demands, and the maximum wrong answers each can survive. More useful than the numbers is the idea behind them, which is that the lever that moves your score changes as you climb.

Why one attempt plan cannot serve every target

Recent CAT papers run 68 questions across three sections: 24 in VARC, 22 in DILR, and 22 in QA, with 40 minutes per section and a fixed order you cannot change. Every correct answer is worth three marks, and a wrong MCQ costs one. That marking rule is the reason a single attempt plan breaks down. The penalty for a wrong answer is fixed, but the value of an extra mark is not. Near 90 percentile, the score-to-percentile curve is gentle, so a few extra marks barely move you. Near 99, the same curve is steep, and three marks can be worth a full percentile point.

So the question is never just how many to attempt. It is how many to attempt given the accuracy you can hold at that count, and given how much each mark is worth at your target. A 90 aspirant usually loses marks by attempting too few questions, often out of caution. A 99 aspirant usually loses marks by attempting one or two too many, taking low-confidence swings that draw negative marking on the part of the curve where mistakes are most expensive. Same exam, two different leaks. This is the selection logic that runs through every serious CAT preparation guide, applied to the one decision you repeat sixty-eight times.

The percentile-target attempt matrix

The table below is a working model, not an official cutoff sheet. Real attempt numbers shift with slot difficulty, normalisation, and the mix of TITA and MCQ in your paper. Treat it as a starting line you then calibrate against your own mocks. The columns that matter most are the last two: as the target rises, the accuracy floor climbs and the room for wrong answers shrinks fast.

Read down any single column and the attempt counts rise slowly. Read across the bottom two columns and the picture changes sharply. Moving from 90 to 99 adds only a handful of attempts, but it nearly cuts the allowed wrong answers in half and lifts the accuracy floor by almost twenty points. That gap is the whole message of the matrix. You do not climb from 90 to 99 by attempting many more questions. You climb by getting far more of the ones you attempt right.

Volume game versus precision game

Think of the bottom of the percentile range and the top as two different games with two different scoring rules.

At a 90 target you are playing a volume game. The accuracy bar sits around 70 percent, which is forgiving, so the binding constraint is how many questions you reach and engage with. Most aspirants stuck near this band are not inaccurate. They are slow or over-cautious, leaving five or six attemptable questions untouched in each section. The fix is to push the attempt count up and accept that a moderate error rate is fine, because at this part of the curve the marks from extra attempts outweigh the marks lost to the odd wrong answer.

At a 99 target you are playing a precision game. The accuracy bar jumps near 88 percent, and the allowed wrong answers fall to about five across the whole paper. Here the binding constraint flips. Reaching questions is rarely the problem for a strong aspirant; the problem is the temptation to attempt the two or three you are not sure of. Each of those carries negative marking on the steepest part of the curve, where one wrong answer can undo the gain from a question you nailed. The fix is selection: attempt only what you can defend, and let the unsure ones go.

Reading the matrix section by section

The three sections do not behave alike, so a smart plan sets a separate target for each rather than splitting attempts evenly. The differences come from how risk is distributed inside each section.

VARC: steady selection, low variance

VARC rewards consistency more than the other two. Reading comprehension questions arrive in sets tied to a passage, and once you have read the passage well, three or four questions follow at a similar difficulty. The accuracy you hold here is the most stable of the three sections, which is why the VARC attempt range is the highest in the matrix. For a 99 target, leaving more than two or three RC questions is usually a sign you read too slowly, not that the questions were unfair.

DILR: the section that sets your ceiling

DILR carries the lowest safe attempt count and the most danger. One badly chosen set can swallow fifteen minutes and return nothing. This is why the DILR numbers in the matrix look conservative even at the top: a 99 aspirant attempting 15 to 17 of 22 is usually choosing three or four sets well and solving them fully, not spreading thin across all of them. Set selection in DILR matters more than raw speed, a point worth pairing with a deliberate plan for which section you open the paper with, covered in our guide to CAT section order strategy.

QA: where accuracy discipline pays most

QA is the section where negative marking bites hardest, because the temptation to attempt a half-solved problem is strongest. The matrix asks for high attempts here, but only if accuracy holds. A 95 aspirant attempting 17 of 22 at 80 percent is in good shape; the same aspirant attempting 21 at 65 percent has lowered their net score. The accuracy floor is doing the real work in this column. If you want the deeper math behind that tradeoff, it is laid out in accuracy versus attempts in CAT.

The sectional cutoff trap

One reason to set section-wise targets rather than a single overall number is that the IIMs apply sectional percentile cutoffs. A strong overall score still gets rejected when one section dips below its required line. So the matrix is not only about maximising total marks; it is about clearing every section's floor first, then pushing the total. A balanced 95 across all three beats a lopsided profile with one brilliant and one weak section, and it is far safer for shortlists.

How to build your own target line

The matrix is a map, not your route. Your route comes from three numbers you already have after a few mocks: your current accuracy in each section, your current attempt count, and the gap to your target. Build the line in this order.

1. Fix accuracy before adding attempts. If your QA accuracy is below the floor for your target, do not attempt more questions yet. Raising a leaky attempt count just adds wrong answers. Lift accuracy first, then climb the attempt count.
2. Find your weakest section against its cutoff. The section closest to falling below a sectional line gets your next block of practice, because clearing the floor protects the whole score.
3. Match the matrix to your real data. If your mocks show you hitting the 95 row in two sections but the 90 row in DILR, your target line is set by DILR. Move that one first.
4. Re-test the line in the next mock. Treat each mock as a check on whether the attempt and accuracy combination is actually producing the percentile you modelled, and adjust.

This is where a generic table runs out of usefulness and your own pattern takes over. Optima Learn's performance-based guidance reads your mock-by-mock attempt and accuracy data and rebuilds the target line for each section, so the plan tracks the aspirant you are this month rather than a textbook average. You can sanity-check any version of your line against the CAT score predictor and keep drilling weak areas on the Optima Learn question bank.

Common questions on CAT attempt strategy

Stop running one plan for every score. Decide your target, read the right row, and notice which lever it asks you to pull: more attempts at 90, sharper selection at 99. Then build your own line from your mock accuracy, fix the weakest section against its cutoff first, and re-test the combination every mock. The aspirants who hit their target percentile are rarely the ones who attempted the most questions. They are the ones who attempted the right number for the score they were actually chasing.