You finished the mock, opened the analysis, and the number that hurt wasn't your score. It was the negative marking. You attempted 24 in QA, got 14 right, and the eight wrong ones quietly ate 8 marks off the top. So next mock you told yourself you'd be careful, and you still attempted 23, because in the moment every question felt doable. A good CAT attempt strategy is the one thing nobody actually teaches you how to build — they just hand you a table and say "be accurate." This blog is about fixing exactly that.
If you have searched this in the last month, you have read the same advice ten times. Attempt 18 to 20 in Quant. Attempt 12 to 14 in DILR. Accuracy over attempts. Skip the traps. All true. All useless at 4:47 PM on exam day when your hand is hovering over question 21 and your brain is screaming "you can solve this one." A real CAT attempt strategy has to survive that exact moment, not just sound sensible on paper.
Why a copied CAT attempt strategy fails on exam day
The generic table you keep seeing — QA 18–20, VARC 22–24, DILR 12–14 — comes from averaging the "good attempts" of people who hit 99 percentile in a specific year. It is a description of what happened, not an instruction for what you should do. A copied CAT attempt strategy inherits someone else's year and someone else's accuracy, and two problems make that dangerous.
First, those numbers move with paper difficulty. CAT 2025 DILR was brutal — puzzle and Venn sets were so time-consuming that a 99 percentile in that section needed only 11 to 12 correct, not the 14 a calmer year demands. If you walk in married to "I must attempt 14 sets' worth," a hard slot will wreck you, because you will force attempts on sets that have no clean entry point. A real CAT attempt strategy bends with the paper instead of fighting it.
Second, and this is the part nobody says out loud: the table assumes you already have the accuracy of a 99-percentiler. You probably don't yet. If your QA accuracy in mocks is sitting at 70 percent, copying a 90-percent-accuracy person's attempt count means you attempt their number of questions with your error rate — and the negatives bury you. The attempt count and the accuracy are a pair. You cannot borrow one without the other.
The number you actually need is yours, not theirs
Here is the thing your coaching class won't sit you down and explain, because it requires your data, not their slogan. Your personal attempt ceiling is the point where attempting one more question stops adding net marks and starts subtracting them. Past that line, every extra attempt you make is, on average, a wrong answer pulling your score down. Below that line, you are leaving safe marks on the table. Finding that exact line for each section is the whole game, and a working CAT attempt strategy is just the habit of finding it and then obeying it.
How to find your personal attempt cap from mock data
You need at least four or five analysed mocks for this. Not attempted — analysed. The whole CAT attempt strategy rests on this raw data, so it is worth doing properly. For each section in each mock, write down three things: how many you attempted, how many you got right, and your accuracy on those attempts. Then look for the band where your accuracy holds up.
Say across five mocks your QA looks like this. When you attempted 14 to 16, your accuracy stayed around 85 percent. When you pushed to 19 or 20, accuracy dropped to 65 percent. That drop is the signal. The questions you added between 16 and 20 were the ones you weren't sure about — and they came back as negatives. Your QA cap, for now, is roughly 16. Build your CAT attempt strategy around the number where accuracy stays high, not the number where it collapses.
Run the same read for VARC and DILR separately, because they behave nothing alike. VARC accuracy usually degrades slowly — one bad RC passage costs you, but you can often attempt 18 to 20 and hold a decent strike rate. DILR is the opposite: it is binary. A set either cracks or it doesn't, so your "attempts" there are really "sets entered," and your cap is two or three clean sets, not a question count. Treating all three sections with one rule is the most common reason a CAT attempt strategy falls apart.
The marking math that makes the cap real
The reason this works is the +3 and −1 structure. Each correct MCQ is worth three marks; each wrong one costs one. So one correct answer is worth three wrong answers in raw terms. That sounds forgiving until you do the percentile math. The +3/−1 structure is the engine the entire CAT attempt strategy runs on, so it pays to feel it in your bones. In a tough slot, four or five wrong answers can drop you 5 to 10 percentile points, because everyone's raw scores are bunched and a 6-mark swing moves you past a few thousand people. Accuracy isn't a virtue here — it is arithmetic, and your CAT attempt strategy is how you keep the arithmetic on your side.
One more lever most aspirants underuse: TITA questions. The type-in-the-answer questions carry no negative marking. That means they sit outside your cap entirely. Your attempt ceiling is really a ceiling on risky MCQs — you can and should take every TITA question you have a genuine shot at, because the worst case is zero, never minus one. Folding that into your CAT attempt strategy often adds three or four safe marks a section that you were skipping out of blanket caution. A good CAT attempt strategy treats TITA as free territory.
Obeying the cap when adrenaline says attempt more
Knowing your number is half of it. The harder half is stopping at it when every instinct says keep going. This is where most people who have read all the same advice still fail, because nobody trains the stopping — they only preach the number. The discipline half of a CAT attempt strategy is what actually separates the aspirants who hold their score from the ones who watch it leak away.
One technique that solves this is to talk to someone who has sat in that exact chair and felt that exact pull. The challenge is usually that your coaching mentor teaches the syllabus, not the live psychology of holding your hand back at minute 38. Platforms like eSalahKaar let you book a per-minute call with a verified student from IIM-A, IIM-C, or XLRI who actually converted — so you pay only for the actual conversation time, asking exactly how they enforced their own cap on the day. Worth bookmarking if you are deep in the mock phase and watching your negatives undo your prep.
The mechanical version, the one you can drill yourself, is to set the cap inside your mocks and treat hitting it as a hard stop. Decide before the section starts: 16 in QA, then I switch to pure review of what I've marked. When you hit 16, you do not attempt a 17th even if it looks easy — you go back and recheck your existing answers instead. You are training your hand to stop, the same way you trained it to skip. Do this for ten mocks and the stop becomes reflex. A CAT attempt strategy you have rehearsed under pressure survives exam day; one you only read about does not.
Other ways to build attempt discipline
The mentor call is one route. Here are other legitimate ways to build the discipline a CAT attempt strategy needs, with honest trade-offs:
First, keep an attempt-and-accuracy log yourself. A simple sheet with three columns per section per mock. It costs you nothing and is the single most reliable way to see your real cap emerge over time. The downside is it only works if you are honest in your analysis and disciplined enough to maintain it across every mock — most people start one and abandon it by mock four.
Second, use community threads where aspirants post their own attempt-versus-score breakdowns. Forums like PaGaLGuY have long discussions where people share the exact attempts and accuracy that got them a given percentile in a specific slot. Reading several of these calibrates your expectations against real numbers rather than a polished table. The trade-off is noise — you have to sift genuine breakdowns from bravado, and a stranger's cap is still not yours.
Third, lean on your mock provider's analytics if they offer them. Some test series flag your accuracy decay by attempt order, which is exactly the curve you want. It is convenient and automatic. The catch is it shows you the data but not the judgment — it will not tell you where to draw your line, only what your line currently looks like.
Each has a cost. The log takes discipline, the forums take time and a filter, the analytics take a paid series and still leave the decision to you. The fastest shortcut is asking someone who has already drawn their line, but none of these replace your own mock data as the foundation. Whatever route you pick, the work to build a real CAT attempt strategy is the same: gather your numbers, find your band, rehearse the stop. If you still have doubts about how the whole approach fits your profile, the questions other aspirants asked are a reasonable place to start.
What this looks like for a real aspirant
Take Ananya, a final-year B.Com student in Indore prepping for CAT 2026 around a part-time articleship. For her first six mocks she attempted everything she could in QA — 21, 22 questions — and her percentile sat stuck in the low 80s. When she finally logged her data, the pattern was obvious: above 15 attempts her QA accuracy fell off a cliff. She capped herself at 15, redirected the last seven minutes to rechecking marked answers, and her net QA score went up despite attempting seven fewer questions. Same brain, same syllabus. Different CAT attempt strategy. That is the entire difference between her two score bands, and it cost her nothing but the discipline to stop. Her whole CAT attempt strategy was one number and the nerve to honour it.
The aspirants who clear the sectional cutoffs cleanly are almost never the ones attempting the most. They are the ones who know their number cold and refuse to cross it when the timer is loud. If you are in the mock phase right now — what's your real attempt cap in your weakest section? Most people have never actually checked. Pull your last five mocks and find it before your next one. It takes twenty minutes and usually explains the percentile you've been stuck at.