GMAT 675 (V82, Q87, DI81) — 94th Percentile — with e-GMAT
The section I came in to fix was not the section that was costing me the
points.I was certain I knew my problem. Quant — around the 85th
percentile, while verbal and Data Insights sat near the 90th. Fix quant,
leave the rest alone. Roughly three months later I finished at 675 (V82, Q87,
DI81), the 94th percentile, with the Q87 at the 93rd. Much of what moved
quant had nothing to do with the topics I thought I was weak in.
The weaknesses a decent score hidesMy accuracy was respectable enough everywhere that I could not
tell where the points were going. Scholaranium,
e-GMAT's practice bank,
splits each subsection into blocks of related topics and reports accuracy on
easy, medium and hard questions separately for each block. Reading that, I
found weaknesses I would never have named on my own.
My instinct had been to work everything in parallel — partly
strategy, partly boredom, because doing one thing only gets dull. What I
settled into instead was one section, then one subsection, then the blocks
inside it, each closed before anything else opened. Working full time you
cannot keep four things half alive, and when you close a subsection properly
its errors do not leak into the sectional phase.
Quant: the rules I wrote for myselfAlgebra was where I bled, and not because of the concepts. I
would fail to write down a constraint, or solve one case and never notice a
second existed — the same leak as in probability and permutations. Hardquestion accuracy in Scholaranium told me which blocks were dragging; my
error log told me why, because I logged two things for every miss — whether I
had misread or failed to note what was asked, and whether my approach had
cost me time even where I got there. One is behavioral, the other is method,
and separating them showed which topics needed concepts and which just needed
me to stop being careless.
So I wrote three rules and followed them on every question with
no exceptions: read every word, translate every sentence into a simple math
equation, and write down what the question is actually asking for before
solving. Then I drilled them with custom quizzes pulled on exactly the tags
that were failing, rather than practising algebra in general. Hard accuracy
in Algebra went from 70 percent to 90, and Word Problems landed at 90.
Lengthy speed, distance and time questions needed volume on top of the rules.
Verbal: pre-thinking on CR, a map for every RC passageMedium CR was fine. Hard CR took me a long time, and strengthen
and weaken were the worst — sometimes I could not tell what an answer choice
was implying. What fixed it was the pre-thinking methodology in the
e-GMATcourse: before touching the five options you work out what would have to be
true for the argument to hold or break, so you arrive already knowing what
you are hunting for. For me that meant slowing down and putting the option's
implication into words before matching. Hard CR accuracy reached 80 percent.
RC broke down on long or unfamiliar passages, so I used the
course's mapping method — for each paragraph, why it is there plus a word on
any concept it introduces. The why notes stack into the main purpose; the
rest sits ready for detail and inference questions. My floor became "I have
the gist and the author's opinion, so I get three of four."
Then I cleared the verbal cementing quizzes — short exam-level
quizzes after each topic — started sectional mocks, and dipped. It had
nothing to do with verbal ability: I had been practising in blocks of five or
ten questions and was not used to writing twenty-three in a row.
e-GMAT's
sectional mocks were the fix as much as the diagnosis. My verbal sectionals
opened in the high seventies and low eighties and closed at 87.
Data Insights: DS earns the attention it getsDI came last and got the least time, which shows in my DI81. Data
Sufficiency is the part I got right. Medium was comfortable, hard was not,
and the fix was the quant rule sheet transferred wholesale: convert each
sentence into simple math, write down the ask. Hard DS accuracy reached 80
percent in Scholaranium, and in the exam I got every DS question right —
which matters more than it looks, because five of my first seven questions
were DS.
The rest took longer. Certain Graphics Interpretation graph types
only came round with volume, so I pulled targeted practice from Neuron, which
serves official-style questions with full step-by-step solutions. Table
Analysis was fine once I closed concepts like correlation, and TPA was a
timing problem the cementing quizzes and DI sectional mocks fixed.
e-GMAT's
DI course is broken out by question type, which is why isolating TPA from MSR
from GI was possible.
Five mocks in two weeksSectionals to full mocks was another stamina cliff, and the
scores bounced: 715, then 675, then 695, then 665, then 735. Read a list like
that expecting a clean line upwards and you will panic. What I did between
them counted more. The leaks were almost all behavioral — a misread, one
missed condition — so I ran custom quizzes on exactly the topics that had
broken in that mock, and went into the next one with the process tightened.
Out of that came a written rule sheet: the quant and DS rules
above plus a line per verbal question type, short enough to recite so you
never fly blind into a question. And a warm-up — one or two questions from
each of the four quant modules in Scholaranium custom quizzes before
starting, since quant was my first section. Test day was the smoothest run of
all of them; I finished five minutes inside the clock.
Having someone else read my dataI had a mentor through the Last Mile Push programme reading my
data alongside me, and the value was not encouragement — it was a second
interpretation. I would look at a mock and reach one conclusion; the mentor
would read the same data and name what I was leaking, sometimes something I
had missed entirely. It also meant that when a mock came in at 665 I was not
sitting alone with it. If you are in the closing phase and you are the only
person reading your own numbers, look into LMP.
If you are stuck on a decent scoreCement every area before you go near a full-length test, then use
sectionals and mocks to build the stamina — it is a separate skill and it
will cost you points first. Then write your rule sheet and follow it. Narrow
diagnosis, and no moving on until a block is closed.