Numerical reasoning
Roughly 10-12 multi-step questions · ~75-90 seconds per question
What it tests. Fast fraction-to-percentage conversions, compound growth (CAGR), margin expansions and currency translation under load. No advanced calculus.
Worked example. Given three regions with revenue and operating margins, recalculate total combined operating profit after a 5% cost cut in one region and a 20% revenue rise with a 200-basis-point margin drop in another.
Common traps. Red-herring data (5 columns when you need 2); unit misalignments (millions versus billions, percent versus basis points where 1 bp equals 0.01%); chasing exact decimals.
How to handle it. Round aggressively to cross-check order of magnitude, and pre-format scratch paper into labeled grids per question.
Verbal reasoning
Roughly 10-12 passages with several statements each · ~60-75 seconds per statement
What it tests. Reading comprehension and separating explicit textual truth from plausible assumptions.
Worked example. A passage on biosimilar adoption; decide whether "slower US adoption versus Europe is exclusively due to state opt-in mandates" is True, False or Cannot Say.
Common traps. The outside-knowledge pitfall (answering from what you have read elsewhere); missing absolutes like exclusively, always, never or solely.
How to handle it. Read the statement first, scan for anchor keywords, and default to Cannot Say if it requires a logical leap beyond the text.
Logical / inductive / deductive reasoning
Roughly 10-15 matrix puzzles · ~45-60 seconds per puzzle
What it tests. Abstract problem-solving, spatial reasoning and executive function under time pressure.
Worked example. A 3x3 grid where symbols rotate clockwise, change internal fill and alter line thickness on row and column rules; pick the missing bottom-right shape.
Common traps. Fixating on a single variable (shapes) while ignoring color or border counts; sunk-cost time on one puzzle.
How to handle it. Isolate one attribute at a time (rotation, shading, count, position) and set a hard 45-second cutoff before guessing and moving on.
Situational judgment (SJT)
Roughly 15-20 scenarios · No explicit per-item timer; pace at ~60 seconds per scenario
What it tests. Professional communication, prioritization under stress, hierarchy awareness and culture fit.
Worked example. You find a data contradiction at 4:15pm that breaks a model due at 5:00pm; choose the best and worst of options ranging from hiding the flaw to immediately notifying your Project Leader with a fix plan.
Common traps. The hero complex (fixing structural problems alone without telling the team); passive escapism (pushing the problem down the road).
How to handle it. L.E.K. prizes extreme ownership, data integrity, transparency and collaboration; the right answer is immediate, solution-oriented communication with your lead. Spot the worst action first.
Personality questionnaire
Roughly 40-50 statements · Untimed, but answer instinctively
What it tests. Resilience, structured working habits, comfort with ambiguity and competitive drive against a high-performance benchmark.
Common traps. The chameleon trap (faking the ideal partner profile triggers SHL consistency checks and can auto-reject); always picking aggressive, individualistic choices.
How to handle it. Answer through your most organized, analytical, professional self, but do not lie, and let analytical and structured traits surface naturally.
Custom L.E.K. case-style data test
20-30 questions tied to one or two case scenarios · 60-75 minutes total
What it tests. Live case synthesis, information-hierarchy management and decision confidence with no interviewer present, mirroring an Associate on a tight CDD timeline.
Worked example. An agrochemical company with margin compression: identify the line to divest, calculate break-even volume change from a 12% price rise, and synthesize expert quotes to pick a market-entry strategy.
Common traps. Over-analysis paralysis (studying every data point before reading the question); misreading double-Y-axis charts (dollars on the left, growth percent on the right).
How to handle it. Read the question first, then targeted-scan the charts for the values you need; keep a running hypothesis about whether the business is good or bad as a sanity check.