Hi all, first time posting here. This is going to be long, sorry in advance.
Quick background. I'm an Indian engineering grad, male, which if you know anything about MBA admissions is basically the most crowded bucket that exists. I spent last year reading M7 admit profiles and honestly it was depressing. People who ran programs at NGOs, people out of serious quant firms, founders doing real revenue, YC folks. And then me. Decent engineering degree, a job that looks exactly like a hundred thousand other jobs.
So the GMAT was pretty much it. Everyone says you need 715+ for a profile like mine. I don't know if that's officially true anywhere but that's the number floating around, and it was the one thing on my application I could still change.
I figured quant would be the easy half. I cleared JEE, I've written more entrance exams in this country than I want to think about. Quant was supposed to be free marks.
Yeah, no.
Funny part is verbal actually went fine, which I really did not see coming since English isn't my first language. But verbal responds to just doing more of it. GMAT Ninja on YouTube basically carried me there, that channel is genuinely excellent and it's free. Manhattan books for volume after that, mostly CR and RC. You watch enough and read enough and it moves. Quant did not work like that at all.
So here's what my prep actually was, and I'm being specific because this is the whole point of the post. Official guides, the 2025-26 Focus material, did a decent chunk of questions out of there. GMAT Ninja for verbal. Manhattan for question volume. A bit of extra work on DI. And the four official mocks, 1 and 2 are free, I paid for 3 and 4.
Mock totals came out 675, 685, 705, 715. Quant was between 87 and 90, so zero to two questions wrong. Best one was 715 and I looked at that and went okay, that's me, I'm ready.
Actual score: 665.
Below every single mock I sat, including the worst one. I'd put real months into this and I walked out with a number I could probably have got without doing much at all. Was not a good day.
Took me a while to work out what happened. Two things.
One, four mocks isn't practice. It's four times I checked where I was standing. A mock tells you your position, it doesn't move you anywhere. I spent the entire prep measuring the temperature and basically never turned the heat on. Sounds obvious written down. Was not obvious at the time.
Two, and this one annoys me more, the variance was sitting right there and I read it as good news. 675 to 715 across four sittings isn't a score, it's a 40 point range. Exam day just picks one number out of that range. I looked at 715 and decided that was my level, when the honest read was "somewhere between 675 and 715, and probably worse when it's real." 665 is roughly what the bottom of my range looks like once you add a test centre and nerves.
And the reason the range was that wide had nothing to do with concepts. Q90 to Q85 is like two or three careless mistakes. On this scale that's 30-40 points of total, since one section point is worth about 6.7. Also worth knowing, Q85 is the 88th percentile. Which sounds fine until you remember I'm an Indian engineer and quant is supposed to be my thing. 88th percentile there isn't a flex, it's a question mark.
Anyway. It was the same mistake every single time. Some long wordy problem, I'd miss one small thing in the phrasing or drop a calculation somewhere. Then I'd look at the options, spot something close to what I got, and go with it feeling completely fine.
This is the bit I actually want to warn people about, because it took me months to notice. My wrong answer was almost always sitting right there in the options. Those distractors aren't random. Each one is built to be exactly what you get if you make one specific predictable mistake. So when I slipped, the answer choices agreed with me. I wasn't catching anything, because the test is designed to hand your own mistake back to you looking like a real answer.
Side note, GMAT Ninja has a line in one of his CR videos about how you lunge at an answer that looks right because you jumped over one word, and about not confusing speed with efficiency. That's a verbal video and it described my quant problem exactly. Took me way too long to realise it applied.
Three versions of this that kept getting me:
1. Solved for the wrong thing. Long setup and then the last line quietly changes what they're asking for. Two cars, second one sells for 3/5 of the first, 48,000 together. You get x = 30,000 and you're done. Except they asked for the second car, which is 18,000. And 30,000 is right there in the options. Your algebra was perfect, your answer is wrong, nothing feels off at all.
2. Averaged something that needed weighting. 120 km at 40 kmph, back 120 km at 60 kmph, average speed. Brain says 50. It's 48, because you spent longer at the slow speed. And 50 is in the options, and it takes you four seconds to get there.
3. Percent change off the wrong base. 80 down to 60. "60 is what percent less than 80" is 25. "80 is what percent more than 60" is 33.3. Wordy questions hide which one they want and both numbers are always in the choices. That's literally the only reason the question exists.
Then the second half of it, which honestly is worse. Once I'd put three minutes into a question I could not leave it. Classic sunk cost, I'd already spent the time so quitting felt like throwing it away, so I'd keep digging. Six minutes on one question. Then four questions at the end with no time left, where I'd make exactly the mistakes above.
So that six minutes never cost me one question. Cost me five.
DI had the same disease except there it was purely the clock, that was the entire problem.
What actually helped was accepting that walking away from a question can't be a judgement call. At minute three you're the worst possible person to decide whether you're close, because you're the one who's already paid for it. It has to be a clock rule you follow without asking yourself.
So what I actually needed was volume. Timed volume. My rough guess is 60-70 hours of it, not four mocks and some books.
And that's where I got stuck. The questions exist. They're in the OG, they're in Manhattan, they're everywhere. Just not in a form where you're sitting them against a clock in something that looks like the actual exam screen. Doing questions untimed out of a book is a completely different activity. It trains the maths and trains zero of the decision making, and the decision making was my actual problem. There's no point practising without the timer, because the timer is the thing that breaks you.
Also there are only six official mocks in existence. Once you've burned them there's nothing left that behaves like test day.
So I ended up building myself a practice setup. Not as a product, just locally on my laptop, because I wanted somewhere to sit a few thousand timed PS questions in something that felt like the real screen. Around 1,500 questions in it now. I've done all of them.
I haven't retaken yet. Still sitting at 665 and I'm not going to pretend otherwise. Doing the hours now and I'll come back and post whatever I get. Especially if it doesn't move, honestly, that's useful information too.
One thing I actually want to ask you guys. I want to build the same for DI, because I couldn't find anything decent there either and my DI problem was purely timing. Is DI where people are genuinely stuck, or was that just me being bad at it?
Quick background. I'm an Indian engineering grad, male, which if you know anything about MBA admissions is basically the most crowded bucket that exists. I spent last year reading M7 admit profiles and honestly it was depressing. People who ran programs at NGOs, people out of serious quant firms, founders doing real revenue, YC folks. And then me. Decent engineering degree, a job that looks exactly like a hundred thousand other jobs.
So the GMAT was pretty much it. Everyone says you need 715+ for a profile like mine. I don't know if that's officially true anywhere but that's the number floating around, and it was the one thing on my application I could still change.
I figured quant would be the easy half. I cleared JEE, I've written more entrance exams in this country than I want to think about. Quant was supposed to be free marks.
Yeah, no.
Funny part is verbal actually went fine, which I really did not see coming since English isn't my first language. But verbal responds to just doing more of it. GMAT Ninja on YouTube basically carried me there, that channel is genuinely excellent and it's free. Manhattan books for volume after that, mostly CR and RC. You watch enough and read enough and it moves. Quant did not work like that at all.
So here's what my prep actually was, and I'm being specific because this is the whole point of the post. Official guides, the 2025-26 Focus material, did a decent chunk of questions out of there. GMAT Ninja for verbal. Manhattan for question volume. A bit of extra work on DI. And the four official mocks, 1 and 2 are free, I paid for 3 and 4.
Mock totals came out 675, 685, 705, 715. Quant was between 87 and 90, so zero to two questions wrong. Best one was 715 and I looked at that and went okay, that's me, I'm ready.
Actual score: 665.
Below every single mock I sat, including the worst one. I'd put real months into this and I walked out with a number I could probably have got without doing much at all. Was not a good day.
Took me a while to work out what happened. Two things.
One, four mocks isn't practice. It's four times I checked where I was standing. A mock tells you your position, it doesn't move you anywhere. I spent the entire prep measuring the temperature and basically never turned the heat on. Sounds obvious written down. Was not obvious at the time.
Two, and this one annoys me more, the variance was sitting right there and I read it as good news. 675 to 715 across four sittings isn't a score, it's a 40 point range. Exam day just picks one number out of that range. I looked at 715 and decided that was my level, when the honest read was "somewhere between 675 and 715, and probably worse when it's real." 665 is roughly what the bottom of my range looks like once you add a test centre and nerves.
And the reason the range was that wide had nothing to do with concepts. Q90 to Q85 is like two or three careless mistakes. On this scale that's 30-40 points of total, since one section point is worth about 6.7. Also worth knowing, Q85 is the 88th percentile. Which sounds fine until you remember I'm an Indian engineer and quant is supposed to be my thing. 88th percentile there isn't a flex, it's a question mark.
Anyway. It was the same mistake every single time. Some long wordy problem, I'd miss one small thing in the phrasing or drop a calculation somewhere. Then I'd look at the options, spot something close to what I got, and go with it feeling completely fine.
This is the bit I actually want to warn people about, because it took me months to notice. My wrong answer was almost always sitting right there in the options. Those distractors aren't random. Each one is built to be exactly what you get if you make one specific predictable mistake. So when I slipped, the answer choices agreed with me. I wasn't catching anything, because the test is designed to hand your own mistake back to you looking like a real answer.
Side note, GMAT Ninja has a line in one of his CR videos about how you lunge at an answer that looks right because you jumped over one word, and about not confusing speed with efficiency. That's a verbal video and it described my quant problem exactly. Took me way too long to realise it applied.
Three versions of this that kept getting me:
1. Solved for the wrong thing. Long setup and then the last line quietly changes what they're asking for. Two cars, second one sells for 3/5 of the first, 48,000 together. You get x = 30,000 and you're done. Except they asked for the second car, which is 18,000. And 30,000 is right there in the options. Your algebra was perfect, your answer is wrong, nothing feels off at all.
2. Averaged something that needed weighting. 120 km at 40 kmph, back 120 km at 60 kmph, average speed. Brain says 50. It's 48, because you spent longer at the slow speed. And 50 is in the options, and it takes you four seconds to get there.
3. Percent change off the wrong base. 80 down to 60. "60 is what percent less than 80" is 25. "80 is what percent more than 60" is 33.3. Wordy questions hide which one they want and both numbers are always in the choices. That's literally the only reason the question exists.
Then the second half of it, which honestly is worse. Once I'd put three minutes into a question I could not leave it. Classic sunk cost, I'd already spent the time so quitting felt like throwing it away, so I'd keep digging. Six minutes on one question. Then four questions at the end with no time left, where I'd make exactly the mistakes above.
So that six minutes never cost me one question. Cost me five.
DI had the same disease except there it was purely the clock, that was the entire problem.
What actually helped was accepting that walking away from a question can't be a judgement call. At minute three you're the worst possible person to decide whether you're close, because you're the one who's already paid for it. It has to be a clock rule you follow without asking yourself.
So what I actually needed was volume. Timed volume. My rough guess is 60-70 hours of it, not four mocks and some books.
And that's where I got stuck. The questions exist. They're in the OG, they're in Manhattan, they're everywhere. Just not in a form where you're sitting them against a clock in something that looks like the actual exam screen. Doing questions untimed out of a book is a completely different activity. It trains the maths and trains zero of the decision making, and the decision making was my actual problem. There's no point practising without the timer, because the timer is the thing that breaks you.
Also there are only six official mocks in existence. Once you've burned them there's nothing left that behaves like test day.
So I ended up building myself a practice setup. Not as a product, just locally on my laptop, because I wanted somewhere to sit a few thousand timed PS questions in something that felt like the real screen. Around 1,500 questions in it now. I've done all of them.
I haven't retaken yet. Still sitting at 665 and I'm not going to pretend otherwise. Doing the hours now and I'll come back and post whatever I get. Especially if it doesn't move, honestly, that's useful information too.
One thing I actually want to ask you guys. I want to build the same for DI, because I couldn't find anything decent there either and my DI problem was purely timing. Is DI where people are genuinely stuck, or was that just me being bad at it?
