Inductive Reasoning Test: 15 Questions, 15 Minutes

The same test has five different names — make sure you are preparing for the right one

This is the single most common source of wasted preparation. Employers name the test after whichever provider they bought it from, and the names do not match:

• SHL calls it Inductive Reasoning
• Kenexa / IBM calls it Logical Reasoning
• Cubiks calls it Abstract
• Saville calls it Diagrammatic Analysis
• Talent Q calls it Elements Logical
• Many general ability batteries call it Non-verbal Reasoning

All six ask the same thing: infer an unstated rule from examples, then apply it. If your invitation email says diagrammatic or abstract and you have been searching for that term, you have been reading half the available material. Search inductive as well.

One real difference worth knowing: SHL's version is usually strictly timed per section with no going back, while some batteries let you revisit questions. Check your invitation for that detail specifically — it changes your whole time strategy.

The rule families that actually appear

Almost every item is built from one of six rule types, often two combined. Learning to scan for them in a fixed order is faster than staring at the sequence hoping to see it.

1. Rotation. An element turns by a constant angle each step — 45°, 90°, 120°. Check direction: consistent clockwise is common, but alternating direction is the standard trap.

2. Progression. Something increases or decreases steadily — size, number of sides, line thickness, position along an axis. Count rather than eyeball; a 4-sided to 5-sided change is easy to miss at speed.

3. Counting. The number of elements, intersections, or enclosed areas follows a sequence. This is the family most people skip, and it is where the harder items hide.

4. Alternation. Two states swap on a fixed cycle — filled/unfilled, left/right, present/absent. If a two-step cycle does not fit, try three.

5. Movement. An element travels a fixed path — around a perimeter, along a diagonal, one cell per step. Watch for wrap-around, where it leaves one edge and re-enters the opposite one.

6. Addition and subtraction. Elements accumulate or disappear at a fixed rate, or two frames combine into a third by overlaying and cancelling what they share.

Scan in that order. Rotation and progression are visible fastest; counting and combination take longest, so leave them for when the obvious families have failed.

When two rules both fit — the situation that costs the most time

You will regularly find a rule that explains every frame but does not match any answer option. This is not a mistake in the test, and it is not you being wrong. It is the format working as designed.

The convention across providers is that the intended rule is the simplest one that accounts for every element in every frame. Your rule probably explains most of the elements and quietly ignores one.

The practical fix takes ten seconds: pick the frame with the most going on, list its features one by one — shape, count, shading, orientation, position — and check your rule against each. The feature your rule cannot explain is the one carrying the real pattern.

If two rules genuinely both fit all frames, choose the one requiring fewer separate operations. A single rotation beats rotation-plus-recolouring.

Time strategy: this is a speed test wearing a reasoning test's clothes

Most inductive papers give you roughly 45–75 seconds per item, and the scoring rewards volume. The people who score well are rarely the ones who solve the hardest item — they are the ones who did not lose four minutes on it.

• Set a hard cap of about 60 seconds per item. At the cap, take your best guess and move on. There is almost never negative marking on these tests, so a blank and a wrong answer cost the same.
• Do the scan in order. Rotation, progression, counting, alternation, movement, combination. An unstructured stare burns time without narrowing anything.
• Use the options as data. If four options share a feature and one does not, that feature is usually the discriminator — work backwards from it.
• Expect difficulty to be uneven, not ascending. Many providers randomise, so a hard item early does not mean the rest will be harder.

If your test allows revisiting, mark and skip aggressively on the first pass. If it does not — SHL's usually does not — then a guess at the cap is the only correct play.

How employers actually score it

Your raw score is converted to a percentile against a norm group, and the norm group matters more than the raw number. Being compared against general applicants is very different from being compared against graduates in a technical discipline.

There is no universal pass mark. Employers set their own cut-off, and the same performance can clear one and fail another. If you are told you did not pass, that is information about the threshold, not about your reasoning.

Two things are reliably true across providers: unanswered items score zero, and speed contributes as much as accuracy because the papers are designed so that almost nobody finishes comfortably.

A worked example: one sequence, the rule, and why each wrong option is wrong

Three frames. Frame 1: an outlined triangle in the top-left corner, pointer at 0°, one dot beneath it. Frame 2: an outlined square in the top-right corner, pointer at 90°, two dots. Frame 3: an outlined pentagon in the bottom-right corner, pointer at 180°, three dots. Which figure is frame 4?

Scan the five dimensions in a fixed order. Rotation: 0° → 90° → 180°, so +90° clockwise each step. Count: 1 → 2 → 3, so +1. Fill: outlined throughout, so no rule. Position: top-left → top-right → bottom-right, one corner clockwise. Shape: triangle → square → pentagon, one side more each step. Four rules, all confirmed on both transitions — the key check, since a rule that fits only frames 2→3 is not a rule.

Apply each once more to frame 3: pointer 270°, four dots, still outlined, bottom-left corner, hexagon. Answer: an outlined hexagon in the bottom-left, pointer at 270°, four dots.

• Outlined hexagon, bottom-left, pointer at 90°, four dots — rotation run backwards. The commonest slip; it comes from reading the frames right to left.

• Outlined circle, top-left, pointer at 0°, five dots — every rule applied twice. Comes from counting the question mark as a frame.

• Filled hexagon, bottom-left, pointer at 270°, four dots — a dimension changed that had no rule. Comes from stopping the scan early and assuming the rest 'must' change too.

• Outlined pentagon, bottom-right, pointer at 180°, three dots — frame 3 repeated. The time-out pick.

Every wrong option is one step from the answer on exactly one dimension. That is how the questions above are built, and it is why the report can name the slip behind each miss rather than just marking it wrong.

A practice routine that works in a week

• Day 1: Do one full timed paper cold, before reading anything. You need a baseline and you need to find out which rule family you are blind to. Most people are blind to counting.
• Days 2–4: Practise untimed. Untimed practice is where you build the scan habit; timed practice only measures it. For every item you miss, write down which rule family it used.
• Days 5–6: Return to timed, with the 60-second cap enforced strictly. Getting comfortable abandoning an item is a skill, and it has to be practised.
• Day 7: Rest. Fatigue costs more on speeded tests than one more practice paper gains.

The test above is timed and gives a per-type breakdown after you submit, so you can see which rule family is costing you rather than just a total. Nothing is marked until you finish.