Treat classification as the core skill for SBI PO Prelims: name every puzzle family, DI presentation and English question function before you touch it, then decide in seconds whether to solve, park, or leave. This guide teaches the named categories, works through seating and DI decisions step by step, ranks arithmetic by measured cost, and finishes with a weekly triage drill scored against a concrete rubric.
The sectional clock changes what 'prepared' means
Because English, Quantitative Aptitude and Reasoning Ability are assessed in separate timed blocks, readiness means making fast selection decisions inside each block, not just knowing more formulas or rules.
In most exams, a strong candidate can bank time in a favourite subject and spend it elsewhere. SBI PO Prelims removes that option: each section runs in its own window, so an over-invested puzzle in Reasoning is simply lost time, not time you can lend to Quant. Your preparation should therefore rehearse the decision — solve, park, or leave — as deliberately as the solving itself.
Build this into every practice session from day one. Before you touch a question, label it: which puzzle family, which DI presentation, which English function. Then state your decision and the reason in one line. Over weeks this creates a personal map of which categories you solve reliably and which you consistently defer — knowledge that is far more useful on exam day than an extra hundred solved questions done in random order.
Reasoning puzzles: classify the family before you start drawing
Puzzles and seating arrangements fall into recognisable families — linear rows, circular facing, floor and box stacking, scheduling, comparison-based. Identifying the family determines whether the set is worth your clock.
Treat classification as a distinct skill. A single-row linear arrangement with direct positional clues is structurally different from a double-row facing puzzle, where 'opposite' relationships add a constraint layer. Box-stack and floor puzzles depend on relative ordering clues ('above', 'immediately below'), while scheduling puzzles fix events against days or months. Each family has a typical setup cost: comparison-based and multi-variable stacking puzzles usually need the most deductions before the first answer becomes reachable.
Scenario 1 — a plausible mistake: with the Reasoning clock running, a candidate opens a seven-box stacking puzzle with three variables (colours, cities, professions), draws the stack, and makes deductions one by one. Fifteen minutes of effort later, two of five questions are answered. The better decision: in the first scan, count variables and recognise that multi-variable stacking sets carry high setup cost; mark the set, move to a single-variable linear seating, and return only if time remains. Why it matters: in a section with its own clock, one high-setup set can absorb time that would have converted three low-setup sets into answers. The trap is not difficulty — it is unfavourable cost per question.
Seating arrangements: use structure, not guesswork, to place people
Solve seating puzzles by fixing the frame first — row direction or facing orientation — then placing the most constrained people before the loosely constrained ones, and testing each deduction against the frame.
Worked mini-example: five friends A, B, C, D, E sit in a single row of five positions, all facing north, with positions numbered 1 to 5 from the viewer's left. Clues: C sits second from the left end; B sits immediately right of A; E is not adjacent to C; D sits at one of the ends. Fix C first — position 2 is settled by definition. The negative clue on E then does real work: E cannot occupy positions 1 or 3, so E must be at 4 or 5. Now test the two possible A–B pairs. If A and B take positions 3 and 4, only positions 1 and 5 remain for D and E; E must be at 5, so D takes the end at position 1 — arrangement D, C, A, B, E. If A and B take positions 4 and 5, the remaining positions 1 and 3 both fail (D is not at an end in position 3, and E at 3 would sit beside C), so that branch collapses. The lesson: anchor fixed-position clues first, apply negative clues to shrink the options, then case-test the relational clue — the order of operations, not speed of drawing, decides how long the set takes.
Compare this with a circular arrangement where some people face inward and some outward: here 'left of' and 'right of' reverse depending on facing, so a single misapplied direction corrupts every subsequent deduction. That is why the frame check — confirming facing orientation before any placement — is a named step worth rehearsing on its own. If a circular set keeps producing contradictions after two attempts, that is structural evidence of a misread frame, and parking the set is the correct response, not redrawing faster.
Data Interpretation: choose the set before choosing the numbers
DI sets differ by presentation — table, bar, pie, line, caselet — and by calculation demand. Scan all sets first, then commit to the one whose questions match approximation-friendly arithmetic.
Worked example of the right habit: a pie-chart set gives a distribution of 7200 units across five segments; one question asks for the difference between two segments of 22% and 18%. A plausible mistake is computing both values exactly (1584 and 1296, difference 288) when the options are spaced far apart — 4% of 7200 approximated as 4 × 72 = 288 takes seconds. Conversely, a caselet embedded in dense paragraphs often requires extracting ratios before any arithmetic starts, so its extraction cost is the real price, not the calculation. Judge each set on extraction plus calculation cost together.
Scenario 2 — a plausible mistake: a candidate starts a line-graph set on production across months because line graphs 'look easy', but every question demands successive percentage changes computed exactly, eating the clock. The better decision: compare all available sets in the first scan; a table set whose questions ask only for ratio comparisons and 'highest value' identification converts faster even though tables look duller. Why it matters: the presentation format signals nothing about cost — the question verbs do. Train yourself to read the questions of a set before reading its data.
Arithmetic topics: rank them by setup cost, not by familiarity
Core arithmetic — percentage, ratio, average, profit and loss, simple and compound interest, time and work — rewards percentage-fraction conversions and unit-based methods over long formula chains.
Two habits separate fast arithmetic from slow arithmetic. First, percentage–fraction conversions: knowing that 12.5% is one-eighth, 16.66% is one-sixth, and 37.5% is three-eighths turns most percentage questions into fraction multiplication. Second, ratio-based methods in time-and-work and partnership problems: expressing work rates as ratios avoids the fraction bookkeeping that slows lcm-based approaches. Practise conversions as a standalone daily drill with a timer, because in the exam their value is speed, not correctness alone.
Rank your topic list by your own measured cost, not by how well you think you know it. In a timed drill, time yourself on five questions each of percentage, time and work, boats and streams, mixtures, and ages. Expect a couple of these topics to run noticeably longer per question for you — whichever pair that is becomes your defer list for the exam. This ranking is personal and must be rebuilt each week, because it changes as your practice does.
English: read RC questions by function, and fix errors by rule
Reading comprehension questions split into factual, inference, vocabulary-in-context, and tone/function types; grammar questions test named rules — agreement, tense, modifiers, preposition use — each with a checkable test.
Triage RC by question function. Factual questions can often be answered from a located sentence without reading the full passage; inference questions require understanding the paragraph's argument, so they cost more and deserve a full read of the relevant paragraph; vocabulary-in-context questions should be answered by substituting each option into the original sentence, never by guessing the word's standalone meaning. Skim the questions before the passage, then read strategically: opening paragraphs for the topic and stance, topic sentences for structure.
Error-spotting questions are rule-retrieval tasks. Build a short named list: subject–verb agreement after collective nouns and intervening phrases; tense consistency in conditional structures; misplaced modifiers ('only' placement); parallelism in lists; preposition use after common verbs and adjectives. For each rule, write one sentence you got wrong and the corrected version beside it — the pair is a far better revision unit than the rule alone. Cloze tests work the same way: check tense continuity across the passage and logical connectors between adjacent sentences before choosing individual words.
A triage drill with a self-check rubric, and your preparation sequence
Run a weekly triage drill: one timed section where the only task is labelling and deciding, not solving. Score it against a rubric, then fold results into a staged preparation sequence.
Practical exercise: take any practice set of mixed Reasoning puzzles and DI sets. Give yourself a fixed personal scan window — say five minutes — in which you label every item (family, presentation, question verbs) and mark solve/park/leave. Do not solve anything. Then check: did your 'solve' set actually contain the fastest items when you solved them afterwards? Rubric — give yourself one point each for: every item labelled before any solving; no 'solve' item that later took over 50% more time than a parked one; your three fastest solved items were all marked 'solve' in the scan; you can state in one line why each parked item was parked. Four out of four is the milestone; repeat weekly until stable.
Adaptable sequence: weeks one and two, learn the classification lists (puzzle families, DI presentations, RC functions, grammar rules) untimed, and run the triage drill without time pressure. Weeks three and four, add timed section practice where the decision habit is mandatory, and rebuild your personal arithmetic topic ranking. Ongoing, alternate full-section timed sets with error-log review — one corrected worked example per error, written out completely. Treat scores as a by-product; the log of decisions is the primary evidence of progress.
| Item type | Attempt now if... | Park or leave if... |
|---|---|---|
| Linear seating, direct clues | Frame is clear on first read | Two contradictory placements after one attempt |
| Multi-variable stack or comparison puzzle | Early in the section and low-variable | Three or more variables with only indirect clues |
| Table DI with ratio or identification questions | Question verbs are simple | Every question needs exact multi-step computation |
| Caselet DI | Extraction is quick and verbs are simple | Extraction itself requires chained inferences |
| RC factual and vocabulary questions | Target sentence is locatable quickly | Only inference questions remain late in the clock |
| Error spotting | A named rule fires on first read | Nothing identifiable within a first read |
References and further reading
Use these references to explore the concepts and check the latest information from the relevant organizations.