Deloitte calls it diagrammatic reasoning. EY and PwC call it inductive reasoning. KPMG calls it logical reasoning. The names shift depending on the firm and the test platform, but the actual test is the same thing in every case: a sequence of abstract shapes with no words, no numbers, and one missing piece you have to identify.
No prior subject knowledge, no equations, no business jargon. Just pattern recognition under a clock. Most candidates underestimate this test — not because it's harder than numerical or verbal reasoning, but because it rewards a specific analytical approach that isn't obvious the first time you encounter it, and that approach is entirely learnable.
What's actually on screen
Each question presents a row of five or six abstract diagrams — circles, triangles, arrows, grids — where each frame is slightly different from the last. Your task is to identify the hidden rule governing the changes and pick the next diagram from four to six answer options. Some test formats use a 3×3 matrix with one tile missing rather than a linear row, but the challenge is identical: find the pattern, apply it, select the answer.
Here's the important thing: the sequences look complex, but they're built entirely from a fixed set of five pattern variables. Everything you'll see on a Big 4 inductive or logical reasoning test is a combination of these.
That's the complete toolkit — nothing outside those five. The questions feel difficult because two or three variables change simultaneously within a single sequence, creating visual noise that obscures what's actually shifting. The candidates who score well aren't faster thinkers; they've trained themselves to work through the five variables systematically rather than trying to read the sequence as a whole impression.
How each firm runs it
Deloitte hosts its logical test within the Cappfinity immersive online assessment. Deloitte tends to favour matrix formats (3×3 grids) more than linear sequences, and the rules often operate independently across rows and columns — a shape might rotate along one axis while its fill changes along another. Identifying which rule applies to which axis is the specific skill Deloitte's version tests.
KPMG uses SHL for its full aptitude battery, with logical reasoning alongside numerical and verbal tests. The format is largely standard next-in-series — linear sequences, one clear rule set per question. Timing is tight (roughly 20 minutes for around 15 to 18 questions), but the questions don't layer multiple axes of change the way Deloitte's matrix format can.
EY runs inductive reasoning through its SHL cognitive assessment alongside numerical and verbal tests. Worth flagging separately: if you applied for a Tax or Law-adjacent role at EY, you may face the Watson Glaser Critical Thinking test instead of or alongside standard inductive reasoning. Watson Glaser is a fundamentally different assessment — it tests argument evaluation and inference, not pattern recognition — and standard inductive prep doesn't transfer to it.
PwC includes inductive reasoning as part of its SHL Verify G+ battery — roughly 12 questions in a 20-minute window. If your route involves Career Unlocked (the Arctic Shores gamified assessment) rather than SHL, pattern recognition features there too, but embedded in mini-games rather than standalone sequences. Check your invitation to confirm which version applies to your role and market.
The approach that actually moves the needle
Look at the answer options before the sequence
This is the most counterintuitive piece of advice, and consistently the most useful. The differences between the answer options tell you exactly which variables the question is testing. If four out of five options show a circle and one shows a triangle, you know shape is part of the rule. You can now scan the sequence specifically for how shape changes — rather than checking all five variables blind. It's efficient pattern analysis, not guessing.
State the rule before selecting your answer
There's a real difference between sensing a pattern and being able to articulate it. "The shape rotates 90° clockwise each step and increases in size every other frame" is a rule you can apply with confidence. "That one looks right" isn't — and under sustained time pressure, instinct gets sloppier as the session continues. The habit of formulating the rule explicitly before clicking keeps accuracy consistent throughout the whole test, not just at the start.
Move on when you're stuck
Visual complexity is usually deliberate noise. The questions that look hardest often track the fewest variables — the extra shapes and irregular outlines in the background are there to slow down candidates who scan instead of analyse. If 40 seconds have passed and the rule still isn't clear, mark your best option and move on. Every question carries equal weight.
Make timed practice non-negotiable
Untimed practice tells you whether you can find the rule. Timed practice tells you whether you can find it at the pace that actually counts. Most candidates see a 15 to 25% accuracy drop the first time they switch from untimed to timed conditions — that gap is what focused prep closes. Two to three weeks of daily short sessions (10 to 15 questions with a timer running) is enough for most people to see meaningful improvement. Use provider-matched material where possible: Cappfinity and SHL present questions differently enough that generic abstract reasoning banks may feel noticeably off from the real thing.
What to expect on the day
The logical or inductive reasoning test sits inside a longer battery — numerical and verbal sections in the same sitting, no saves, no pauses between sections. Across all four firms, the combined assessment typically runs 45 to 65 minutes depending on the route and market. Most platforms give you a 24 to 72-hour window from the invitation to complete everything, so schedule it in one focused block rather than fitting it around other things.
The informal benchmark for advancing is typically the 70th percentile; candidates who reach assessment centres tend to score around the 80th. Neither figure is published officially, but both are consistent across prep platforms and candidate reports across multiple recruitment cycles.
If you're also working on the numerical reasoning section — a completely different skill set, but one that rewards the same systematic discipline — the Big 4 numerical reasoning prep guide covers exactly what to expect from data tables and charts under timed conditions. Worth reading alongside this one before you sit.
