An abstract reasoning test is a timed, non-verbal assessment that measures how quickly you can identify the rule governing a sequence or group of shapes, and apply it to pick the correct next figure or the odd one out. There is no language, no arithmetic and no domain knowledge involved: only shapes, positions, colours, counts and rotations. Because it strips out education and vocabulary, it is treated as one of the closest available proxies for fluid intelligence: the ability to solve an unfamiliar problem with no prior training. Employers use it to predict how fast a candidate will learn a job they have never done.
Do not stare at the whole figure. Run a fixed scan order (count, then shape, then position, then rotation, then shading, then size) and stop at the first attribute that changes consistently across the sequence. Test the rule against every frame before you answer, use the answer options to eliminate rather than to confirm, cap yourself at about 45-60 seconds per item, and answer every question. The pattern rules are a small, learnable set, which is why practice moves scores fast.
What Is an Abstract Reasoning Test?
You are shown a sequence, a matrix or a set of figures and asked to work out the rule that connects them. The classic item gives you five frames in a row and asks which of five options comes next. Others give you a three-by-three grid with one cell missing, or two groups of shapes and ask which group a new figure belongs to. The rule is always deterministic (there is exactly one answer that satisfies it) but it is never stated. Finding it is the test.
The same underlying skill goes by several names, and the terminology matters when you are working out what to practise. Abstract reasoning, inductive reasoning, diagrammatic reasoning and non-verbal reasoning are used almost interchangeably by publishers, and Raven's Progressive Matrices is the best-known academic instrument in the family. In practice, treat them as one skill: if you can find the governing rule fast, every version of the test rewards you.
- Practise the whole family, not one label: free abstract reasoning test, free inductive reasoning test and free logical reasoning test.
- If your employer runs a full cognitive battery, pair this with logical reasoning and IQ-style cognitive practice.
What the Test Actually Measures
- Fluid intelligence: solving a novel problem with no relevant prior knowledge. This is why abstract reasoning is used for career-changers and school leavers as often as for graduates.
- Rule induction: forming a hypothesis from a few examples, then testing it, the same loop that underpins learning any new system at work.
- Hypothesis discipline: whether you abandon a rule that fails on frame four rather than forcing the data to fit it.
- Working memory: holding two or three simultaneous rules (a rotation and a count and a shading cycle) without losing one of them.
- Speed under uncertainty: the items get harder as you go, and the clock does not care that you are one insight away from the answer.
Abstract reasoning is culture-fair and education-neutral by design: no words, no numbers, no domain knowledge. That is exactly why employers weight it heavily, and why it is one of the few cognitive tests where deliberate practice produces a visible jump. Most candidates have simply never seen the pattern taxonomy before.
The 6 Question Formats
- Next in sequence: five frames, pick the sixth. The most common format by a distance.
- Matrix completion: a three-by-three grid with one cell blank. Rules usually run along both rows and columns, so check both.
- Odd one out: five figures share a rule; one breaks it. Find the rule, not the difference.
- Group A / Group B classification: two sets defined by hidden rules, and a new figure to assign. Aon cut-e's abstract items are frequently built this way.
- Analogy: figure A is to figure B as figure C is to which option? Identify the transformation from A to B, then apply exactly that transformation to C.
- Diagrammatic operators: shapes pass through boxes that transform them (rotate, invert, add an element). You deduce what each operator does, then apply the chain.
The 8 Pattern Rules That Solve Almost Everything
Almost every abstract reasoning item is built from a small number of transformation rules, usually one or two combined. Learn the list and you stop searching blindly.
| Formula | Expression |
|---|---|
| Rotation | A shape turns by a fixed angle each frame (45, 90, 120 or 180 degrees), often alternating direction |
| Reflection | The figure flips across a vertical, horizontal or diagonal axis |
| Progression / counting | The number of elements, sides or dots rises or falls by a constant step (or follows 1, 2, 3, 5 style sequences) |
| Movement | One element travels around the frame by a fixed number of positions each step |
| Shading / colour cycle | Fill states cycle in a repeating loop, for example black, grey, white, black |
| Size change | Elements grow or shrink in a fixed order, sometimes swapping between two shapes |
| Addition / subtraction of elements | One element is gained or lost each frame, or two frames combine into a third |
| Consistency (the odd-one-out rule) | Every figure shares an invariant property (for example an even number of sides) and the answer is the one that breaks it |
The Scan Order: How to Attack an Item
The most common failure mode is not a lack of ability. It is unstructured looking. Candidates stare at the whole figure hoping the rule will announce itself. Instead, run the same scan every time, in the same order, and stop as soon as an attribute changes consistently.
- 1Count. How many elements are in each frame? A count that rises by one is the single most common rule and takes two seconds to spot.
- 2Shape. Does the shape type change, cycle or stay fixed? Are sides being added?
- 3Position. Is an element moving around the frame, along a path, or between corners?
- 4Rotation. Is anything turning? Fix your eye on one asymmetric element and track only that element across all frames.
- 5Shading. Fill, colour, hatching. Is there a cycle, and does its period match the number of frames?
- 6Size. Growing, shrinking, or swapping between two elements?
- 7Combine. Hard items use two rules at once. Once you have one, look for a second before you commit. A wrong answer usually satisfies one rule and breaks another.
- 8Eliminate. Apply the rule to the answer options and discard every option that violates it. Elimination is faster than construction, because you often only need one attribute to kill three options.
Worked Examples
Frame 1: one black triangle in the top-left corner. Frame 2: two black triangles, top-left and top-right. Frame 3: three black triangles, adding bottom-right. Frame 4: four black triangles, adding bottom-left. Frame 5: five black triangles. The fifth appears in the centre.
Identify the rule and pick the sixth frame from the options.
Run the scan. Count first: 1, 2, 3, 4, 5. A clean +1 progression, so the sixth frame has six triangles. Position second: the triangles fill the corners clockwise from the top-left, then the centre. So the sixth element must be the next available position in that clockwise order. Shape, rotation, shading and size are constant, which means the answer is fully determined by count plus position.
Six triangles: four corners, the centre, and the next clockwise position (the top edge midpoint). The engineered distractors are a six-triangle frame with the extra element in the wrong position, and a five-triangle frame with a shading change. Each satisfies one rule and breaks the other.
Figure A: a square containing three dots. Figure B: a triangle containing four dots. Figure C: a pentagon containing five dots. Figure D: a hexagon containing six dots. Figure E: an octagon containing eight dots.
Find the invariant that four figures share and one does not.
Do not look for what is different. Look for what is the same. Count the sides and count the dots. Square: 4 sides, 3 dots. Triangle: 3 sides, 4 dots. Pentagon: 5 sides, 5 dots. Hexagon: 6 sides, 6 dots. Octagon: 8 sides, 8 dots. The invariant is sides equals dots. Two figures break it, but only one breaks it in a way the others do not share: the square has one fewer dot than sides, the triangle has one more.
The relationship holds for C, D and E. In a real item only one figure would break the rule; this version illustrates the trap of settling on a rule before testing it against every figure. Always verify against all five, not the first three.
Sit a timed abstract reasoning test built on the formats employers actually use. The free preview gives you 4 questions per test plus 5 AI coach messages, no payment details needed.
Start the free testPublisher Differences: What to Expect
Abstract reasoning varies more by publisher than any other cognitive test, because publishers disagree about whether it should be a power test (hard items, generous time) or a speed test (easy items, brutal time). Find out which one you are sitting, because the strategies are opposite.
| Publisher | Questions | Time | Difficulty | Notes |
|---|---|---|---|---|
| SHL Verify Inductive | About 18 items | Roughly 25 minutes | Medium-Hard | Next-in-sequence format; comfortable timing, so accuracy is the priority |
| Aon cut-e scales cls / sx | Continuous items | About 6 minutes, roughly 5-12 seconds per item | Very Hard | Group A / Group B classification; pure speed, designed so you cannot finish comfortably |
| Talent Q Elements Logical | 12-15 items | Adaptive, about 75 seconds per item | Adaptive | Difficulty rises as you answer correctly; a wrong answer late is expensive |
| Saville Swift Analysis (diagrammatic) | Combined battery | About 6 minutes for the section | Medium-Hard | Diagrammatic operator chains alongside verbal and numerical in one sitting |
| Raven's Progressive Matrices | Up to 60 items | 20-45 minutes depending on version | Progressive | Matrix completion; items get steadily harder. The academic reference instrument |
| Korn Ferry / Talent Q lineage | Varies | Tight | Hard | Frequently used for graduate and management intakes alongside numerical and verbal |
If your test gives you roughly a minute per item, it is a power test: the rules are complex, and slowing down to verify pays. If it gives you ten seconds per item, it is a speed test: the rules are simple, and hesitation is the only thing that can beat you. Aon cut-e sits firmly in the second camp; SHL and Raven's sit in the first.
9 Strategies That Move Your Score
- 1Always run the scan order. Count, shape, position, rotation, shading, size. A fixed procedure beats intuition, and it stops you freezing on a hard item.
- 2Track one element, not the whole figure. Pick the most distinctive element (the asymmetric one) and follow only that across the frames. Rotation becomes obvious the moment you stop looking at everything.
- 3Verify against every frame. A rule that explains frames 1 to 3 but not frame 4 is not the rule. Most wrong answers come from committing to a hypothesis too early.
- 4Expect two rules on hard items. Once you find one, keep going. The distractors are built precisely so that they satisfy the rule you found first.
- 5Eliminate with the answer options. Do not construct the answer in your head and then hunt for it. Take one attribute you are sure about and delete every option that violates it.
- 6Ignore aesthetics. Symmetry, prettiness and 'balance' are not rules. Test writers deliberately design distractors that look right.
- 7Cap your time and move on. Around 45-60 seconds on a power test, far less on a speed test. Flag it, guess, come back if there is time.
- 8Answer everything. Negative marking is rare, and a guess between five options still has value. A blank never does.
- 9Review by rule type, not by item. Log which rule you missed (rotation, count, shading) and drill that rule specifically. Most candidates have one or two blind spots, not a general weakness.
Mistakes That Cost the Most Marks
- Staring instead of scanning. Unstructured looking is the number one time sink and it produces panic on hard items.
- Locking onto the first rule you find. On difficult items the first rule is usually the decoy, and a second rule decides the answer.
- Missing rotation because you watched the whole shape. Rotation only becomes visible when you fix on a single asymmetric feature.
- Forgetting to check the columns in a matrix. Rules run down as well as across, and the missing cell is often only determined by the column.
- Treating a speed test like a power test. In a cut-e-style test, spending 40 seconds on one item costs you five items. The correct response to a hard item there is to guess immediately.
- Practising the wrong family. Diagrammatic operator chains and next-in-sequence items train different reflexes. Match the practice to the format in your invitation email.
Your 14-Day Preparation Plan
- 1Days 1-2: Baseline. Sit one full timed test cold. Record your score, your seconds per item, and which rule types you missed.
- 2Days 3-5: Rule taxonomy. Drill the eight pattern rules one at a time, untimed. Do a set of pure rotation items, then a set of pure counting items, and so on, until you recognise each on sight.
- 3Days 6-8: Scan-order discipline. Practise applying the scan in the same order on every item, out loud if it helps. Timed, but at a comfortable budget.
- 4Days 9-11: Format-specific practice. Switch to your employer's publisher. If it is a speed format, drill at speed. If it is a power format, drill combined-rule items.
- 5Days 12-13: Error review only. Redo only the items you got wrong. Name the rule you missed on each one, and drill that rule again.
- 6Day 14: Full rehearsal, then stop. One timed test in the morning, then rest. Fatigue costs more on abstract reasoning than on any other test, because working memory is the bottleneck.
MakingMoves.ai covers 50+ test categories with 113,000+ practice questions, and the AI coach names the exact rule behind every item rather than just marking it wrong, which is what turns abstract reasoning practice into learning. The free preview gives you 4 questions per test and 5 AI coach messages. Paid plans are GROW at EUR 19,95 per week, PRO at EUR 49,95 per month, and MAX at EUR 79,95 per month.
Our logical reasoning category covers abstract, inductive, deductive and diagrammatic items, with AI explanations that name the rule and timed mocks that match the real durations.
Explore logical reasoning practiceFrequently Asked Questions
They overlap heavily. Abstract and inductive reasoning are largely interchangeable labels for finding the rule in a set of shapes. Diagrammatic reasoning usually adds process flows and operator boxes that transform an input. Non-verbal reasoning is the umbrella term. Practising any of them improves the others, but practise the exact format your employer names.
You cannot meaningfully raise your underlying fluid intelligence in two weeks, but you can raise your score substantially, because most candidates have never been taught the pattern taxonomy or a scan order. Practice removes the search cost, and the search cost is what the clock punishes.
Scores are reported as percentiles against a norm group rather than as raw marks. Employers do not publish their cut scores, and they vary by role and intake. The workable goal is to be comfortably above the median of your norm group, not to hit a specific number of correct answers.
There is nothing to calculate, so a calculator is irrelevant. Scrap paper is genuinely useful on matrix and operator items, where sketching the transformation is faster than holding it in your head. Check whether your test allows it. Remote proctored tests often do not.
It depends entirely on whether it is a power test or a speed test. A power test such as SHL Verify Inductive typically runs around 18 items in 25 minutes. A speed test such as Aon cut-e gives you a few seconds per item and is designed so that almost nobody finishes. Check your invitation email for the format.
Yes, and if anything they lean on them more, because the format is education-neutral and language-neutral, useful for international intakes and for career-changers with no relevant academic background. It is a standard component of graduate cognitive batteries alongside numerical and verbal reasoning.
Our assessment tips hub covers timing drills, test-day checklists and preparation plans across all 50+ test categories.
Read the assessment tipsIngmar van Maurik is the founder of MakingMoves.ai and Assessment-Training.com. With 10+ years in psychometric assessment design, he has helped over 1 million professionals prepare for job assessments.