Good mock percentile, failed Prelims: reading the number behind the score

Mock scores are the most-watched and least-read number in preparation. People check the percentile and stop, when the two numbers underneath it — attempts and accuracy — are the ones that decide the real exam.

Five doubts below. Two of them describe almost the same symptom and require opposite responses, which is the most useful thing in this article.

85–95 percentile in mocks, and I still fail Prelims

Getting 85-95 percentile in mocks but failing actual Prelims like RBI Assistant, SBI PO.

Mocks mein 85-95 percentile aata hai lekin actual Prelims mein fail ho jaata hoon.

This gap is common enough that aspirants have a name for it — the accuracy trap — and it usually comes from mocks being taken with slightly looser risk tolerance than the real exam deserves.

On a mock, a wrong guess costs a small dent in a percentile you will have forgotten by tomorrow. On the real exam, the same guess costs 0.25 marks that can be the exact difference between clearing and not.

The arithmetic is unforgiving:

80 attempts at 80% accuracy beats 95 attempts at 62% accuracy — even though the second attempt count looks far more impressive on a screen.

There is also a real-exam-only factor that mocks cannot fully replicate: exam temperament, the pressure of what may be a one-shot annual attempt, and the fact that IBPS and SBI occasionally shift the Prelims pattern, which mock platforms take time to catch up to.

The fix is accuracy-first mock discipline — treated as seriously as the real thing. Do not let mock percentile become a comfort number if the accuracy behind it would not survive real negative marking. We have written about the mechanics of this separately in accuracy before speed.

My score has dropped and I feel demotivated

Earlier my English score in Pre was 18+, now it has dropped below 15. Feeling demotivated.

Pehle Pre mein English score 18+ tha, ab 15 se bhi neeche aa raha hai. Demotivated feel ho raha hai.

A dip like this — after an earlier stretch of good scores — almost always traces back to accuracy quietly slipping while attempts crept up. Chasing two or three extra questions under pressure, at the cost of getting a few sure ones wrong along the way. Negative marking punishes exactly that trade.

So before adding any new topic or practice set, spend one full mock with a single rule:

Attempt fewer questions than usual, and do not mark anything you are not genuinely confident about.

Then compare that accuracy percentage against your recent mocks. It usually jumps noticeably.

That is the number to rebuild first, because attempts only add real score once accuracy is back above where it used to be.

And on the demotivation: this is a fixable pattern, not evidence that the earlier 18+ was a fluke. The motivation does come back once the score trend turns, which usually takes just a handful of mocks once accuracy is the focus again.

Every mock lands 2–4 marks above cutoff and it never feels like enough

Marks in every mock are just 2-4 above cutoff, no sense of satisfaction.

Har mock mein marks cutoff se sirf 2-4 zyada aate hain, satisfaction level nahi mil raha.

Sitting consistently just above cut-off, mock after mock, says something more specific than not good enough.

It usually means your accuracy is solid but your attempt count has quietly become the ceiling — because the questions you are skipping are being skipped by instinct rather than by a clear read of the question.

That is fixable in a way that has nothing to do with studying harder.

After each mock, look specifically at the questions you left unattempted. Not the wrong ones — the skipped ones. Then check how many of them you could actually have solved if you had looked a little closer.

If that is a meaningful number, the ceiling is not knowledge. It is a slightly too-cautious skip instinct, and a few mocks of consciously pushing two or three extra attempts — while still tracking accuracy — can move it.

The “just above cutoff” pattern is not a plateau. It is the exact edge where a small, specific change moves the number meaningfully.

The two doubts above need opposite fixes

Notice what just happened. One reader was told to attempt fewer questions. The next was told to attempt more.

Both are correct, and the symptom on a scoresheet can look similar: a score that is not where it should be. What separates them is the accuracy behind it, and whether the skipped questions were genuinely undoable.

Which is the entire argument for auditing your own mocks rather than reading advice about mocks. Nobody can tell you which of those two readers you are. Your last five mocks can.

I skip long questions and they turn out doable

What should Mains attempt strategy be? I skip questions based on length that turn out doable later.

Mains attempt strategy kya honi chahiye? Length dekhkar questions skip kar deta hoon jo baad mein doable nikalte hain.

Length is a genuinely unreliable signal at Mains level, which is exactly why this keeps happening. A long question may just have more explanatory text wrapped around a simple calculation. A short one can hide real complexity.

Judge by structure, not length. Give every question a ten-to-fifteen-second scan for whether the pattern is recognisable — not for how much text there is.

In a DI set specifically, there is a common real pattern worth knowing: three of five questions are straightforward calculations and two are deliberately lengthy. Solving the three and skipping the two protects your time far better than judging the whole set by its size and skipping all five.

And build the recalibration into practice deliberately: after each mock, go back to every question you skipped and time how long it would actually have taken. Over a few mocks that drags your gut sense of “this looks long” towards what is true.

Concepts are clear and execution is not — in three subjects at once

Reasoning is strong and needs no extra practice, but in Quant attempts are slow even with concepts clear. In English, patterns are known but accuracy lags. Clerk Pre papers can be fully attempted in time, PO papers can’t. Classes click at first glance, solving alone leads to getting stuck.

Reasoning strong hai, Quant mein concept pata hone ke bawajood attempt slow hai, English mein pattern pata hai par accuracy kam hai. Clerk Pre attempt ho jaata hai time mein, PO nahi hota.

All four of those are the same underlying pattern wearing different clothes: a real, watching-based understanding that has not yet converted into independent, timed execution.

Slow Quant despite clear concepts. Known-but-inaccurate English patterns. Clerk level manageable and PO level not. Classes clicking while solving alone does not. Four symptoms, one gap.

The highest-leverage fix is a hard ratio shift: for every hour of class or video from here on, at least two hours solving independently — without pausing to check anything mid-question, even if several come out wrong.

On the Clerk-to-PO jump specifically: that is not a separate problem. It is the same gap appearing at a harder difficulty, so the fix is identical. Deliberately solve PO-level papers under time pressure even while the percentage right is lower at first. That discomfort is precisely what builds the speed a PO paper demands.

And there is direct evidence this works, in the question itself: Reasoning is already strong. Whatever was done there is what the other three need. It just needs the same independent-solving investment.

The habit all five answers depend on

None of these answers could have been given from a percentile alone. Every one of them needed attempts, accuracy, and the skipped-question list.

That is the argument for spending as long analysing a mock as taking it — not as diligence, but because the score by itself cannot tell you which of the five readers above you are.