The Guwahati graduate and the SSC CGL Tier-II

He has been preparing for the Staff Selection Commission Combined Graduate Level (CGL) Examination for sixteen months. He failed the last attempt. He cleared Tier-I — the objective CBT, 100 questions in 60 minutes — with 149 out of 200. He then sat for Tier-II Paper-I in October 2025 and scored 364 out of 600. The NE-region general category cut-off that year was approximately 381. He was 17 marks short.

The sectional breakdown tells the story plainly. English: 146 out of 200. General Awareness: 122 out of 200. Quantitative Aptitude: 96 out of 200, against a Quant cut-off of around 113. He had, in a sense, failed a very specific and narrow test, disguised as a general one.

He is trying again. The 2026 CGL notification dropped on March 14. Tier-I is scheduled August 25 to September 11. Tier-II runs October 21 to 26. Five months. He has done eleven mock tests in the past three months on a paid test series costing ₹800 a month — a real sum in this household — and bought ₹2,400 worth of books: R.S. Aggarwal, Kiran's chapter-wise compilation, a Word Power Made Easy he already knew he didn't need. On the current eleven mocks, his Quant score hovers between 102 and 108. The projected 2026 cut-off is expected around 110 to 115. He is not there yet.

His coaching teacher — a well-meaning man running a small batch of twenty-two aspirants near Dispur — tells him, when Mridul raises this in class: "bas mehnat karte raho, padhai chal rahi hai." The diagnosis is not incorrect. It is also not actionable.

🗓️ The shape of the problem

Mridul's Quant gap is not a knowledge gap. He has read the Aggarwal. His Arithmetic is competent — Time and Work, Speed and Distance, Profit and Loss — 65 to 70 percent, which is fine. The problem is structural, and it sits in two specific corners of the syllabus.

The first is Mensuration — specifically three-dimensional Mensuration: cones, cylinders, spheres, frustums, composite solids. His accuracy on 3D Mensuration across eleven mocks is around 25 percent. One in four. On a paper where each wrong answer costs a mark, 25 percent accuracy is worse than leaving them blank. The failure is not knowledge; he can recite the surface area of a frustum without looking it up. The failure is time. He spends approximately 70 seconds per 3D Mensuration question. The target, to finish Quant with enough buffer for Data Interpretation, is 45 to 55 seconds. Those extra seconds, compounded across five or six Mensuration questions per mock, are exactly the seconds he loses on DI.

The second corner is Data Interpretation. The CGL Tier-II DI questions come in sets of four, built around a table or bar chart. Mridul gets the first two right and the last two wrong — not because the last two are harder, but because he spent too long on Mensuration and is now rushing. Consistently: 50 percent DI accuracy. Consistently: the back half of every set, dropped.

This is a time-budget problem, not a content problem. He is studying as though it were a content problem — buying more books, doing more chapters — because the alternative, which is to sit with the mock reports and map where the seconds are going, is something he has not done.

The mock reports are generated automatically: section-wise score, chapter-wise accuracy, time-per-question averages, comparison against the top 10 percent of test-takers. Eleven PDFs, sitting in his phone's download folder, unopened past the first two minutes of each Thursday when he checks the total score and closes the app.

This is not laziness. This is what anxiety does to data. When you know, in some unexamined way, that the data contains a specific bad news, the mind acquires a remarkable capacity to be busy with something else.

1. 📋

   October 2025 — Tier-II: Quant 96/200

   17 marks below the NE-region cut-off. 3D Mensuration accuracy: 25%. DI accuracy: 49%. Eleven mock PDFs generated — none fully read past the front page.

2. 📱

   March 2026 — Payel installs the agent

   All eleven prior reports read in the first week. Two priority topics identified: 3D Mensuration (71 seconds/question) and DI (back-half dropout). Thursday-evening review rhythm begins.

3. 📈

   Weeks 4–12 — Quant mock score: 105 → 113

   Forty hours of timed 3D Mensuration drills, 50-second ceiling. Strict 3-minute cap on DI sets. Agent reports each Wednesday which sub-type is still slow.

4. ✅

   October 2026 — Tier-II actual: Quant 118

   Total 394, NE-region cut-off 378. All India Rank 1,832. Allocated Auditor post, CAG Northeast Regional office, Guwahati — same city, 20 minutes from Beltola.

⚠️ What Payel did

Payel had read a blog post on gabforge.in — about a Patna coaching student named Rakesh who had generated forty-five test-report PDFs for ten months without reading any of them, and whose aunt had read three of them in one evening. She read it on the Cotton University common room wifi, between two lectures. She found it unremarkable in the sense that she recognised it immediately. "Dada r kotha likheche," she told her roommate. He has written about my brother.

She installed gabforge.in's agent on Mridul's phone that evening. She told him it was a study helper that could read his PDF reports, and that he should try it once. "Ebaar try kor. Ki hobo worst case ta?" What's the worst case.

The agent asked him to share one mock report. He shared the most recent one. The agent read it in its entirety and said, in transliterated Assamese and Hindi, quiet and unhurried:

"Mridul, eii report-t dekhi aase, tomar Quant section-t average time per question 68 seconds hoiase. Top 10% takers-r average 51 seconds. Eii 17-second difference-ta bor important. Tumi Mensuration-t normally 70-80 seconds lagate aso — oito kom korile thakibo?"

(The report shows your Quant average is 68 seconds per question. The top 10% average 51. This 17-second gap is important. Can we bring the Mensuration time down?)

He said yes. Not because he had reached some dramatic resolution. Because it was Thursday evening, the question was simple, and the agent was quiet about it.

🌗 The reading of eleven reports

Over the following ten days the agent read all eleven prior mock reports. 3D Mensuration: 25 percent accuracy across 47 questions. DI: 49 percent across 44 questions. Arithmetic: 67 percent, fine. Time-per-question on 3D Mensuration averaged 71 seconds. Time on the last two DI questions per set averaged 58 seconds — rushed — against 43 seconds on the first two, where he was unhurried.

The agent laid this out not in a table but the way a thoughtful person explains a spreadsheet when sitting next to you. "Dui tah topic ase, Mensuration 3D aru Data Interpretation — eii duitar jonyo jodi tumi focused practice koriba, tomar score barhibo. Baki subject-t tumi already decent aso. Eii duita hob priority." Two topics. If you focus on these two, your score will rise. You're already decent on the rest. These two are the priority.

It then made a 5-month weekly plan: Sunday evenings for mock-test review in Assamese, Wednesday evenings for targeted practice on the two priority topics — Mensuration 3D drills capped at 50 seconds per question, DI sets under a strict 3-minutes-for-all-four rule. The plan also flagged the SSC Combined Higher Secondary Level (CHSL) as a parallel track, since its Tier-I window was likely to overlap and the Quant preparation would transfer.

Any good coaching teacher with ninety minutes and eleven mock reports would have produced something similar. No coaching teacher in a batch of twenty-two has that ninety minutes. The agent did.

"Dui tah topic ase — Mensuration 3D aru Data Interpretation. Eii duitar jonyo focused practice koriba, score barhibo."

— Two topics — 3D Mensuration and Data Interpretation. Focus practice on these two, and the score will rise.

🧭 Twelve weeks, Thursday by Thursday

The agent's Thursday evening rhythm became the structure around which the rest of the week organised itself. Each Thursday, when the new mock result came in, Mridul would open the agent before checking the score. The agent had already read the auto-uploaded PDF. It would say something like: "Mridul, eii week-t Mensuration-t 3 questions thakil, ami 1-tar uttor thik korilu. Pichila week-t 0 thik korilu — eito improvement. Ami next week-t aro 30 minutes practice koribo neki?" Mridul, this week there were 3 Mensuration questions, you got 1 right. Last week you got 0 right — that's improvement. Should we do 30 more minutes of practice next week?

The phrase eito improvement — this is improvement — appeared often enough that Mridul began to track it. He had not previously thought of himself as someone who was improving; he had been thinking of himself as someone who had failed Tier-II and was trying again. These are not the same orientation. The distinction matters in the eighth week, when the improvement is too slow to see without the data and the most natural thing is to conclude nothing is changing.

At the end of week four, his Quant mock score was 105. End of week eight, 109. End of week twelve, 113.

3D Mensuration accuracy moved from 25 percent to 61 percent over twelve weeks. This happened because he did forty hours of timed drills on exactly and only 3D Mensuration — cones, cylinders, spheres, frustums — targeting a 50-second ceiling, with the agent noting each Wednesday which sub-type he was still slow on. "Composite solids-t ebar thoda more time laagil. Sphere inside cylinder-ta formula-r application clear na hola. Kal ektu more practice koribo?"

DI accuracy moved from 49 percent to 76 percent, more quickly, because the underlying problem was pacing rather than comprehension. Once he adopted a strict 3-minute cap on the first two questions per set, the back half jumped almost immediately.

His father noticed that his son was spending Thursday evenings looking at his phone and taking notes. He asked Payel what was happening. She said: "Buwa, mor exam date ahile — report padhibo lage." The exam date is coming. He has to read the reports. Hemanta Borah, who had spent thirty-one years in classrooms and knew something about the relationship between regular assessment and actual learning, nodded and said nothing more.

Tier-I and what came after

Mridul took Tier-I on August 28, 2026, at the SSC examination centre in Guwahati. He had registered through the OTR (One Time Registration) system on ssc.nic.in two years prior, updated his certificates, and submitted his 2026 CGL application within two weeks of the March 14 notification. The Tier-I paper was manageable. He scored 152.5 out of 200. He celebrated nothing. He had cleared Tier-I the previous year too.

Tier-II was on October 23 — Paper-I: Quantitative Aptitude, English Language and Comprehension, General Intelligence, and General Awareness, combined CBT, 600 marks. He came out saying little, went home, ate lunch. He told the agent what he could remember of the DI sets. The agent said: "Tumi bhalo korise. Result naahile kiman thik kora porase janibo naai. Kintu tumi khub prepared aahilu." You did well. Until the result we won't know. But you were prepared.

The result: Quant 118 out of 200. English 152 out of 200. General Awareness 124 out of 200. Total 394. The NE-region general category cut-off for the 2026 Combined Graduate Level Final Result came in at approximately 378. He was 16 marks clear.

His All India Rank, published December 12, was 1,832.

Not good enough for Income Tax Inspector or ECI Sub-Inspector — those posts went to candidates in the top 600. The agent had told him plainly, without softening, that his projected scores placed him in the Auditor and Sub-Inspector allocation range under Northeast Frontier Railway and Ministry of Communications zonal quotas.

He was allocated an Auditor post with the Comptroller and Auditor General of India, Northeast Regional office, Guwahati. Same city. His parents are in Beltola. The CAG NER office is in Dispur — twenty minutes by auto. He will not have to pack a trunk. His mother will not have to recalibrate the tailoring stall's finances around a missing household income.

The reports, finally

Nothing in Mridul's story was dramatic. He did not have a revelation. He did not discover a passion for Quantitative Aptitude. He found 3D Mensuration annoying throughout, right through to the end — the frustum formula in particular, which he once described to the agent as "eti unnecessary janotil formula", an unnecessarily crowded formula, and which the agent, with a patience that cost it nothing, made him practise twenty-three more times across the following four Wednesdays.

A month after the result came out, Mridul was sitting with his father on the Beltola flat's balcony in a cool December evening. His father asked how the preparation had gone this year — differently from the last, clearly, but how? Mridul thought for a moment. Then he said, in Assamese, simply: "Ami papa, na kelai pora porhilu, kuoa? Riport padhli kar agent-ke sath." I didn't just read books this time, father. I read the reports with the agent.

Hemanta Borah — who had spent thirty-one years reading student assessment reports in a government primary school, and who knew the gap between a teacher who looks at marks and a teacher who looks at where the marks come from — said nothing for a moment. Then: "Sei-ta bhal korise." That was the right thing to do.

📚

Attempt 1: book-first

Quant score: 96

₹2,400 in books. ₹800/month test series. Eleven mock PDFs. Checked total score on Thursdays, closed the app. 3D Mensuration and DI both unaddressed as specific targets.

📊

Attempt 2: report-first

Quant score: 118

Same books, same test series. But eleven PDFs read in full with the agent. Two topics isolated. Forty hours of timed drills. Thursday review rhythm for twenty weeks.

⏱️

The 17-second gap

Root cause

Average time per question: 68 seconds (Mridul) vs 51 seconds (top 10%). 3D Mensuration averaged 71 seconds. Once the 50-second ceiling was set, DI pacing resolved automatically.

🌱 The data was always there

Mridul's story did not change because he studied harder. The preparation hours in the five months before the second attempt were, if anything, slightly fewer than the first year — more focused, but shorter. What changed was that someone read the reports with him.

The eleven PDF reports had been generating data for sixteen months. The Quant accuracy by chapter, the time-per-question distribution, the comparison against the top decile of test-takers — all of it had been produced, automatically, by the test platform, and all of it had been sitting in a phone downloads folder, opened only as far as the total score and then closed.

A mock test report is a letter about the future. It says: this is where you are slow; this is where you lose marks not from ignorance but from pace. It does not arrive as a letter. It arrives as a PDF, inside an app, three pages deep. Someone has to sit with you and read it.

Mridul is now posted at the CAG Northeast Regional office in Dispur. His father still walks past the Beltola street where the mock-test PDFs had been accumulating. Hemanta Borah, retired schoolteacher, knows that his son had a teacher this year — one who had time, and one who read the reports.