Classification — called "Odd One Out" in most SSC question papers — is built on one simple idea: four items share a defining property, and one item does not. Your job is to name that one misfit.
The analogy that works best in coaching class is this: imagine you are sorting laundry. Shirts, trousers, a jacket, and a kurta all go in the "clothes" pile. A shoe lands in the same basket by accident. You spot the shoe immediately because it belongs to a different category entirely. That is the cognitive move you make in every Odd One Out question.
What makes this tricky in CGL is not the concept — it is the deliberate ambiguity setters build in. A question about Delhi, Mumbai, Chennai, Kolkata, and Lucknow looks like a geography question. Five cities, all in India — what separates them? The setter is testing whether you know the "four metros" classification. You need the right classification axis, not just any classification.
This is the core insight: items can always be grouped in more than one way. The question rewards the grouping axis that leaves exactly one item outside. Your skill is finding that axis quickly.
Three broad categories of Odd One Out appear in SSC CGL:
One more thing before the deep dive: in SSC CGL, Odd One Out questions are among the fastest in the Reasoning section. A well-trained solver should close most of them in under 20 seconds. That speed is the payoff for learning the framework below properly.
Every Odd One Out question, regardless of type, can be cracked with this four-step process.
Step 1 — Scan for the obvious axis. Read all five items. Check whether one item jumps out immediately as being from a different real-world category. Mango in a list of flowers, School in a list of house rooms, Cow in a list of big cats. If one item is clearly from a different domain, you are done in 5 seconds.
Step 2 — If not obvious, list properties of each item. For each item, write down 2-3 properties mentally:
Step 3 — Find the property shared by exactly four items. This is the axis. Once you identify it, the fifth item that lacks this property is the odd one out.
Step 4 — Cross-check by elimination. Verify that the four items you grouped together actually all share that property without exception. Setters sometimes place a near-miss — an item that almost fits the group but fails on a subtle criterion.
Look at the animals question: Tiger, Lion, Leopard, Cheetah, Cow. A surface-level read gives you "mammals — all five are mammals." That axis does not help. Go one level deeper: wild vs. domestic. Still four vs. one — Cow is the odd one. Go one more level: carnivore vs. herbivore. Tiger, Lion, Leopard, Cheetah are carnivores; Cow is herbivorous. Any of these deeper axes reaches the same answer.
But contrast this with the metros question: Delhi, Mumbai, Chennai, Kolkata, Lucknow. All five are Indian cities. All five are state/UT capitals. The distinguishing axis is the "four metros" classification — a specific, named grouping in Indian geography. Here, GK knowledge is required. You cannot logic your way in without knowing the classification exists.
The takeaway: for semantic questions, first try logic (category depth), then fall back on GK. Never skip the GK possibility.
When you see a number-based Odd One Out, run this mental checklist in order (fastest checks first):
For the question 7, 11, 13, 17, 21: one scan for primality is enough. 7 — prime, 11 — prime, 13 — prime, 17 — prime, 21 — 3×7, composite. Done.
For shapes, the fastest axis is almost always: straight-sided polygon vs. not. Triangle, Square, Pentagon, Hexagon all have straight sides and are closed polygons. Circle has no straight sides. The moment you identify "polygon" as the grouping property, Circle is the misfit.
Secondary axes for geometry: number of sides divisible by a specific number, regular vs. irregular, convex vs. concave. These are rarer in CGL but appear occasionally.
Setters sometimes design questions where two different axes seem to each produce an odd one out. This is the hardest variant. Here's how to handle it: the "correct" axis is the one that groups exactly four items with a precise, named, recognizable property. A vague or stretched property is usually the wrong axis. When in doubt, go with the most specific and commonly recognized classification.
Before reading options, look at all five items and ask: "Does any item belong to a completely different domain (living vs. non-living, object vs. place, plant vs. animal)?" If yes, that item is the answer. No further processing needed. This handles roughly 40% of SSC CGL Odd One Out questions. Standard method (checking all properties): 30s. Domain Jump: 5s.
When given a set of odd numbers (or numbers that look like primes), check each for divisibility by 3, 7, 11, 13 in that order — these are the most common composite "imposters" in SSC questions. 21 = 3×7, 33 = 3×11, 77 = 7×11, 91 = 7×13. Memorize these disguised composites. Recognizing 21 as composite takes 2s with this pattern vs. 10s of trial division for an unprepared solver.
For geometry questions: does the shape have a countable number of straight sides? Yes → polygon. No → not a polygon (circle, ellipse, semicircle). In any mixed set of polygons plus one curved shape, the curved shape is always the odd one out. Step count: 1 question, 1 property check, 1 answer. No enumeration of sides needed.
Animals/plants questions that are not immediately obvious: go down the classification ladder — Kingdom → Class → Order → Family → Genus. Stop at the first level where four items cluster and one does not. Big cats (Tiger, Lion, Leopard, Cheetah) cluster at Family Felidae; Cow does not. Domestic animals cluster differently from wild ones. This ladder prevents you from stopping too early at "all are animals" and missing the real axis. Saves 15-20s vs. random property testing.
Some classification axes are named groupings you must know, not logic you can derive: four metros (Delhi, Mumbai, Chennai, Kolkata), Panchmahal, BRICS, SAARC, peninsulas, landlocked states. When a question involves cities, countries, or political/geographic entities and logic alone does not separate them, immediately switch to GK mode. Flagging this mentally in 3s prevents 20-30s of fruitless logical analysis.
Walk into the exam hall with this decision tree running automatically:
Is one item obviously from a different real-world category? → Mark it. Done (5s).
No? → Are these numbers? → Run Prime/Composite check first, then Even/Odd, then Perfect Square. Mark the misfit (10-15s).
No? → Are these geometric shapes? → Polygon vs. non-polygon. Mark (5s).
No? → Are these entities with a known named grouping (cities, countries, historical periods)? → Switch to GK mode. Recall the standard classification. Mark (10-15s).
Still no? → List 2 properties of each item. Find the property held by exactly four. Mark (20-30s).
Rule of thumb: if you have not converged in 30s, eliminate the most obviously "different domain" option and move on. Do not let one question steal time from three others. In SSC CGL Reasoning, speed across all questions matters more than perfection on any single one.
Why this question: The most foundational Odd One Out type — items from two distinct natural categories mixed together.
Solving path: Domain Jump Scan — Rose, Lotus, Jasmine, Marigold are flowers. Mango is a fruit. Different domain, different answer. 5 seconds.
Why this question: Tests whether you look past the surface category "all are places" to the sub-category "parts of a house."
Solving path: All five are "places." Go one level deeper — Kitchen, Bedroom, Bathroom, Hall are rooms within a house. School is a separate institution. Axis: "parts of a house." Answer: School. 10 seconds.
Why this question: Classic animal classification — tests whether you know the "big cats / Felidae" grouping vs. domestic animals.
Solving path: All are mammals — unhelpful axis. Wild vs. domestic — Cow is domestic, rest are wild. Carnivore vs. herbivore — Cow is herbivorous, rest are carnivorous. Either axis identifies Cow. Depth Ladder confirms Family Felidae for the other four. Answer: Cow. 10 seconds.
Why this question: Polygon vs. non-polygon — a pure geometric classification that tests one specific conceptual boundary.**
Solving path: Triangle (3 sides), Square (4 sides), Pentagon (5 sides), Hexagon (6 sides) — all closed figures with straight sides = polygons. Circle — no straight sides, no vertices = not a polygon. Polygon Litmus: 1 check, done. Answer: Circle. 5 seconds.
Why this question: Tests a specific GK-based named classification (four metros) that cannot be derived purely by logic.**
Solving path: All five are Indian cities, all are major cities. Logic alone stalls here. GK Flag triggers — recall "four metros of India": Delhi, Mumbai, Chennai, Kolkata. Lucknow is a state capital but not in this named group. Answer: Lucknow. 8 seconds with GK Flag trained.
Why this question: Numerical Odd One Out — tests prime number identification under exam pressure.**
Solving path: Prime Sieve — 7: prime. 11: prime. 13: prime. 17: prime. 21: 3×7 = composite. Only one composite in the set. Answer: 21. 8 seconds.
Stopping at the broadest category. Writing "all are animals" or "all are places" and getting stuck — you must go deeper until exactly one item falls outside. Broad categories almost never produce the odd one out directly.
Forcing a numeric axis onto a non-numeric question. When the question has five cities or five animals, some students start counting letters in the names or assigning numeric codes. This wastes 30+ seconds and rarely works. Trust the semantic/GK axis first.
Ignoring the "named group" possibility for geography/polity items. Four metros, eight union territories, SAARC nations, BRICS members — these are specific classifications. If five items look like "all are the same type," the setter almost always has a named sub-group in mind. Learn these classifications as vocabulary.
Confusing "reptile" with "snake." Crocodile and Python are both reptiles, but only Python (and Cobra, Viper, Boa) are snakes. Stopping at "all are reptiles" misses the odd one out. This is the Depth Ladder mistake applied to biology.
Assuming the first plausible axis is the correct one. In deceptive questions, you might find an axis that groups three items together and leaves two outside. That is not the answer. The axis must group exactly four and exclude exactly one.
Not verifying the group of four. After identifying your odd one out, always check that the remaining four actually all share the property you identified — without exception. A single counterexample means you have the wrong axis.