In this guide, we cover the complaint I hear more than any other — and the one I lived myself.
In Sem 1, I walked out of every paper feeling good. Concepts? Clear. Definitions, examples, pros-cons, conclusion — all there, neat and concise, two-and-a-half pages per long answer. Then results: 6s and 7s. Meanwhile the guy next to me, who openly "wrote whatever came to mind for 7 pages", got 8s. I was furious for a week — at him, at the examiner, at the system. Then a senior told me the sentence that changed my grades forever: "Your answers were correct. They just weren't checkable." It took me one full semester to understand what that meant, and this post is that understanding, complete.
What matters here: this exact fight — length versus quality — explodes on the DU reddit every exam season. One camp says examiners only count pages. The other says quality-only toppers exist. Both camps are telling the truth about different halves of the same checking process. Whether you sit in lectures every day or prepare from the printed material at home, the examiner holding your copy behaves identically — and once you see the copy through their eyes, the whole debate dissolves.
From three years of writing, getting marked, and comparing copies with seniors across courses, here is the complete mechanics of DU answer writing: what the checker actually does in 90 seconds, the structure that captures step-marks, the visuals that buy pages honestly, and the garnish (quotes, cases, data) that separates 7s from 9s.
How examiners read your answer copy
A DU examiner checks dozens of copies in a sitting, in bulk, against a deadline. Your answer is not read like an essay — it is scanned for decision points. Three frameworks to keep in mind:
- Scanning rewards structure, not density. Headings, underlines, bullet points, and diagrams are visible in a scan; a brilliant paragraph buried in a grey wall of text is not. Two answers with identical knowledge score differently the moment one is scannable and the other isn't. This is the entire "pages" mystery — pages aren't marks, but structured pages are.
- Step-marking pays for parts. Long answers split into headed sections earn marks section by section — intro, three body points, diagram, conclusion. Miss one section and you lose that section, not the answer. A single-block answer is all-or-nothing: one weak paragraph poisons the examiner's impression of the whole page.
- Relevance is the multiplier. Length × structure × relevance. Pages of off-topic flow ("everything I know about this topic") irritate a fast checker and cap your score no matter the length. The 7.5-CGPA trap from the forums is exactly this: extending sheets with repetitive, non-relevant filler. Relevant length scores; irrelevant length annoys.
The length-vs-quality debate, settled honestly
| Answer type | What the checker sees | Typical outcome |
|---|---|---|
| Short + brilliant (2 pages, perfect content) | Correct but thin; step-marks have nowhere to land | 6–7 (this was Sem-1 me) |
| Long + structured + relevant (5–6 pages) | Every section ticks a step-mark; visuals confirm effort | 8–9 (the actual topper zone) |
| Long + repetitive filler (7–8 pages, same point × 4) | Irritation by page 4; impression capped early | 6–8 with a ceiling (the "bullshit" strategy) |
| Short + vague (1–2 pages, generic lines) | Nothing to reward | 4–5, sometimes ER territory |
Read that table twice, because it ends every forum fight: the "fill 7–8 pages" crowd and the "quality only" crowd are both half-right. Pages without structure hit a ceiling; quality without length has no landing area. The winning answer is long because it is structured — each heading naturally grows 3–4 honest lines, each visual earns a page, and relevance keeps every line scorable. Length is the result of structure, never the goal by itself.
The skeleton: every long answer, every subject
Memorise this order until it's reflex. For a 10-mark answer target 3–4 pages; for 15–18 marks, 4–6 pages (confirm your paper's marks-per-question from PYQs on Sulaksh PYQs first — patterns differ across courses):
- Introduction (4–5 lines). One crisp definition + one line framing what the answer will do. Never start with "In this answer I will discuss..." — start with the concept itself. First impressions form here, in seconds.
- Main body (headings + sub-headings). Three to four headed sections, each 5–8 lines. Headings are the step-marks made visible — underline them. One-line paragraphs under each heading beat one giant paragraph covering all three points.
- Visual (one minimum). Diagram, flowchart, table, or timeline — drawn with a scale, labelled fully. A visual breaks the grey wall, proves command, and honestly fills half a page of marks in five minutes. Details below.
- Garnish (one per answer). A thinker quote, case name, Article number, dataset, or current example — the difference between an 8 and a 9. Subject-wise garnish guide follows.
- Conclusion (3 lines). Direct verdict answering the question as asked — not a summary, a judgment. Examiners remember endings; give them one.
Visuals: the honest page-buyers
Every subject has a native visual — learn yours and practise it till it's automatic:
- PolSci / History / Hindi / English: timelines, thinker-comparison tables, cause-effect flowcharts. A Rowlatt Act → Jallianwala → Non-Cooperation chain beats a paragraph of dates.
- Eco / Commerce: labelled diagrams with axes + one data table (Budget figure, scheme outlay). Curves drawn freehand wobble — use the scale.
- Science / Geography: labelled cycles and maps. A nitrogen cycle from memory is half the marks before a word is read.
- SEC / VAC practicals: screenshots-described steps, calendars, metric tables (see the SEC SMM guide).
Practise each visual twice from memory during revision — if you can't draw it without the book, you don't own it yet. One visual per long answer is the floor, not the ceiling.
Garnish by subject (the 8-to-9 jump)
Structure gets you to 8. Garnish takes you to 9. One precise reference per answer:
- Theory papers: quote thinkers by name — but only when sure. A wrong attribution ("as Marx said" for a Gandhi line) damages more than no quote helps. Seniors phrase it as a rule: if you don't exactly remember who said what, refrain rather than invent.
- Law / Constitution / EVS papers: Article and Act numbers with years (see the EVS acts method and Constitution guide). "Article 21" alone is a claim; "Article 21, expanded in Maneka Gandhi (1978)" is an argument.
- Eco / Commerce: one current figure per answer — a Budget number, a scheme, a rate. Data signals you're reading the world, not just the textbook.
- All papers: answer the question asked, not the one you prepared. Underlining the question's key verb (analyse / discuss / evaluate) before starting keeps garnish relevant instead of decorative.
Exam-hall tactics (the last 10%)
- Rank, then write. Use reading time to mark every question ✅/⚠️/❌ and attempt strong-first (see the sprint's reading-time strategy). Confidence compounds; opening with your weakest answer poisons the whole copy.
- Time-box per answer. 30–35 minutes per long answer with a written deadline; last 15 minutes buffer. One masterpiece + one blank loses to two complete goods — always.
- Attempt everything. A structured 2-page attempt at a half-known question earns 4–5 step-marks. A blank earns a zero and signals surrender. There is no negative marking — blanks are the only unforgivable answer.
- Underline sparingly, margin cleanly. Underline headings, key terms, and verdict lines — not whole paragraphs. Over-underlining is visual noise; selective underlining is a second skeleton for the scanner.
The engine behind all of this is the 3-pass PYQ method — timed solving is where skeletons become reflex — and the strategy wrap-up in the 9+ CGPA playbook.
Frequently asked questions
Q1: How many pages per answer, honestly?
10 marks → 3–4 structured pages; 15–18 marks → 4–6. But pages are the output: plan 4 headings + 1 visual and the pages fill themselves. Chasing page-count directly produces filler, which caps scores.
Q2: Are diagrams really compulsory?
Not officially — practically, yes. One labelled visual per long answer is the highest marks-per-minute move in the hall. Subjects without drawable content get tables and flowcharts instead; the function (breaking the grey wall) is identical.
Q3: What if I quote a thinker wrongly?
Don't. A wrong attribution actively damages credibility — examiners know their thinkers. Unsure? Paraphrase the idea without the name, or use a case/article reference you're certain of instead.
Q4: Hindi medium — same rules?
Identical. Structure, visuals, and garnish work the same in Hindi; underline headings the same way. Just confirm technical terms in your medium in advance so you don't stall mid-answer searching for words.
Q5: I know everything but can't extend without repeating — help?
You're missing headings, not knowledge. Take any answer you "can't extend" and force 4 headings + 1 visual onto it — each heading demands 3–4 lines of distinct content, which kills repetition structurally. Extension is an architecture problem, not a knowledge problem.
What to do next
Stop choosing sides in the length-vs-quality war — build answers that are long because they're structured and relevant. This week, take one past paper and write two answers with the full skeleton: planned headings, one visual, one garnish, verdict conclusion. Compare them with your old style. The difference on the page is the difference the examiner sees in 90 seconds.
Questions on this topic? Write to us via Contact — we reply with the relevant PYQ hub and official links.