03 — Test Strategy
How to actually sit this paper. Read this the day before, and again in the ten minutes before you log in.
The paper, as best the evidence supports: ~130 questions in 120 minutes across four blocks — quantitative 30, logical 30, verbal 30, domain 40 — with no negative marking. Full reasoning and the competing accounts are in research/03-pattern-synthesis.md §2. Treat the counts as a planning model, not gospel; everything below is written to survive being wrong about them.
1. The three rules that matter more than everything else
Rule 1 — Never leave a question blank. Not one.
There is no negative marking. This is [CONFIRMED] from three independent accounts with no dissent. A blank scores zero. A guess scores zero on average and 0.25 marks in expectation.
Across ~130 questions, guessing everything you cannot solve is worth roughly 8 marks over leaving them blank. That is more than most people gain from a whole extra topic.
So: with 60 seconds left, if anything is unanswered, stop working and fill every remaining box. Do not aim for a sensible spread — pick one letter and blanket it. Reading the questions is what costs time, and you do not have it.
Rule 2 — One minute per question. Internalise the tempo.
130 questions / 120 minutes = 55 seconds per question. Every account that gives both numbers agrees on roughly this pace.
This changes what "knowing" a topic means. You do not have time to derive anything. You have time to recognise a pattern and execute a remembered method. This is why 91-CHEATSHEET.md exists and why every hour estimate in this library is costed to recall, not mastery.
The practical form of this rule: the 90-second abandon. If a question has taken 90 seconds and you cannot see the end of it, mark your best guess, flag it, and move. A hard question and an easy question are worth the same one mark. Spending four minutes on a hard one to gain one mark while three easy ones go unread is the single most common way to lose this paper.
Rule 3 — Three passes, not one.
Navigation between questions is permitted ([CONFIRMED] — three independent accounts, 2017, 2020 and 2026).
| Pass | Time | What you do |
|---|---|---|
| Pass 1 | ~60% of the clock | Answer everything you can do in under 60 seconds. Flag anything slower and move immediately. Do not start a puzzle set you cannot finish |
| Pass 2 | ~30% | Return to flagged questions, hardest-value-first: puzzle sets where cracking the setup wins 3–5 marks at once |
| Pass 3 | last ~10% | Fill every blank. Then, only if time remains, re-check questions where you changed your mind |
Pass 1 is the one people get wrong. Its job is not to score well. Its job is to guarantee that every easy mark in the paper has been seen.
2. Section order — do not take them in the order given
If section navigation is free (2017 account), take them in this order:
| Order | Section | Why here |
|---|---|---|
| 1st | Verbal (30) | Cheapest and fastest marks on the paper. Grammar and vocabulary are recall — 30–40 seconds each. Banks time for later sections and settles your nerves on questions you can actually finish |
| 2nd | Quantitative (30) | Do these while fresh and your arithmetic is sharp. Skip data interpretation on pass 1 — DI sets are slow, and come back to them |
| 3rd | Domain (40) | The biggest block, so it must not be rushed, but it is also the block your degree already covers. Mostly recall |
| 4th | Logical (30) | Last deliberately. It has the highest variance — puzzle sets either crack in 90 seconds or eat five minutes. Meeting them last means a bad set costs you time you have already banked, not time the other sections needed |
If sections are separately timed — the sources conflict on this and it is unresolved — this ordering is unavailable. In that case apply the three-pass rule within each section, and carry no unanswered question past its section's clock.
If reading comprehension appears
Read the questions first, then scan the passage for the answers. Do not read the passage properly. You are not being tested on comprehension, you are being tested on retrieval speed.
3. The logical section needs its own tactic
This tactic is sound whatever the paper looks like, but its evidence is thin. Treat it as good practice, not as a confirmed L&T feature.
Three of the four logical items on an old (c. 2007-08) L&T-era paper share one premise block, and one of them runs the chain backwards. No modern account confirms or denies it, but sets of three to five questions on one setup are standard in Indian campus aptitude papers, so prepare for puzzle sets rather than thirty independent one-liners.
The consequence is a decision rule:
| Situation | Do this |
|---|---|
| You can build the grid/diagram in ~60 seconds | Commit. You are about to earn 3–5 marks from one piece of work. This is the best time-to-marks ratio on the paper |
| The setup is still opaque after 90 seconds | Abandon the whole set. Guess all of them, flag, move on. Do not half-solve it |
| You cracked the setup but one question is odd | Answer it from the grid anyway. The grid is the expensive part and it is already built |
Always draw the grid. Do not hold a seating arrangement in your head. Rough paper is officially permitted — use it.
4. Quantitative tactics
Look for the shortcut before you start calculating. The observed questions reward this heavily:
- The heights-and-distances question is 20 seconds if you know
cot15° − cot30° = 2exactly, and three minutes if you don't. - The successive-percentage question is instant if you know +p% then −p% always loses exactly
p²/100percent.
Use the options. They are four data points you paid nothing for.
- Back-substitute when solving would be slow — especially for age, ratio and work problems.
- Eliminate by magnitude. If the answer must exceed 100, two options often die immediately.
- Eliminate by last digit on multiplication-heavy questions.
- Eliminate by parity.
Approximate aggressively on data interpretation. If options are 23%, 31%, 42% and 58%, you do not need two decimal places — you need the right bucket.
Assume there is no calculator. Policy is unknown; rough paper is confirmed. Every method in aptitude/ is built for pen and paper. If a calculator appears, treat it as a bonus.
5. Domain section — including the bad case
The good case. The paper matches your registered CS discipline. Read technical/00-INDEX.md: core first, then your one variant layer. Mostly one-line recall — OSI layers, port numbers, data-type sizes, OOP definitions, SQL clauses.
The bad case, and it is a real possibility. You open the domain section and it is core engineering — thermodynamics, strength of materials, electrical machines. This is not paranoia: every observed technical item from an L&T Limited source is core-engineering, and AmbitionBox's entire 91-question L&T Limited GET corpus is civil and electrical. See research/02-online-research-log.md §5.
If that happens, follow this exactly:
- Do not panic and do not quit. You have already banked the aptitude sections if you followed §2.
- Answer all 40 anyway. No negative marking means a blind guess on 40 questions is worth ~10 marks. Blank is worth 0.
- Skim for the free ones. Even a mechanical paper carries units, basic physics, graph reading and arithmetic that any engineer can do. Take those properly.
- Then spend every remaining second on the aptitude sections, which are 90 of ~130 marks and are branch-neutral. That is where this paper is won regardless of which domain paper appears.
Write that sequence down. Deciding it now costs nothing; deciding it during the test costs ten minutes of panic.
6. Cut-off strategy
2026 cut-offs are not stated anywhere I could find, but sectional cut-offs appear in every L&T Limited account that mentions them: VIT Vellore 2016 ("sectional cutoff as well as total cutoff"), NIT Jamshedpur 2017 (50% on the composite section, 33% on English), and SVNIT Surat 2020 ("each section have cut off"). Three independent accounts over five years. Treat them as real.
Two things follow:
- Sectional cut-offs may exist, so no section can be sacrificed. A brilliant domain score does not rescue a failed verbal section. This is the main reason the plan refuses to let you skip verbal, even though it feels like the least "serious" section.
- Target ~70% overall with no section below 50%. That clears every cut-off that has ever been reported for this company, and it means the unknown stops mattering. That is the same posture the Accenture plan took for the same reason.
7. Before you log in
| When | Do |
|---|---|
| Day before | 91-CHEATSHEET.md only. No new topics. Sleep |
| 60 min before | Eat. Clear the desk. ID out. Rough paper and two pens out |
| 20 min before | Close every tab and app. Chrome only. Phone out of camera view |
| 15 min before | Log in. Late logins are refused outright |
| First 2 min | Read the instruction screen properly. It is the only place the real question counts, timing and any negative-marking rule will be stated — and if it contradicts this file, the instruction screen wins |
Officially confirmed conditions: Chrome only, screen-share enabled, webcam + mic + speakers on throughout, no mobile devices, live proctor, rough sheets and pen/pencil permitted, no breaks, timer cannot be paused. If the page crashes, re-enter the same URL and credentials after 2–3 minutes; support is 044-48134590 / assess.support@lntedutech.co.in.
8. The one-paragraph version
Verbal first, then quant, then domain, then logical. Three passes: sweep the easy marks, return to the flagged, then fill every blank. Abandon anything at 90 seconds. Commit to a logical puzzle set only if the setup cracks in 60 seconds, then take all 3–5 marks. Guess everything you cannot do, because nothing is deducted. If the domain paper turns out to be core engineering, answer all 40 blind and pour the rest of your time into the 90 aptitude marks. Target 70% overall, nothing under 50% in any section.