Numerical reasoning
Part of the roughly 70-question blended set · No hard timer, but total time is scored; aim under 2 minutes on complex data tasks
What it tests. Mental mathematical stamina, computational accuracy, data extraction and interpreting financial ratios without explicit guidance.
Worked example. From a table of revenue, EBITDA and net debt for four firms across three years, calculate the absolute difference in EBITDA margin (EBITDA divided by revenue) between two firms, rank all four by leverage (net debt divided by EBITDA), and type the exact value rounded to two decimal places.
Common traps. Spending five or six minutes perfecting one calculation (which hurts efficiency), and rounding or format errors such as entering 14.5% instead of 0.145, which scores zero.
How to handle it. Set up a clean scratchpad with formulas ready; if a calculation would take five minutes, make an educated estimate, input it and move on.
Verbal reasoning
Part of the blended set · No hard timer; maintain a steady workflow
What it tests. Command of professional business English, contextual logic, editorial accuracy and synthesis of dense information.
Worked example. A draft client memo on a cross-border regulatory shift contains ambiguous phrases, a misused 'affect' versus 'effect' and logical jumps; click the flawed sections and select professional replacements from a contextual menu.
Common traps. Correcting statements using outside news knowledge rather than the text provided, and missing subtle tone shifts that alter diplomatic clarity.
How to handle it. Read the whole passage once for the core objective, then edit as a first-year analyst reviewing a document for an MD: precision, formal tone, absolute clarity.
Critical and logical reasoning
Part of the blended set · No hard timer; do not stall on abstract sequences
What it tests. Non-verbal reasoning, structured problem-solving, abstract pattern recognition and isolating logical fallacies.
Worked example. Premise 1: all technology acquisitions by Team X exceed £500m. Premise 2: some Team X acquisitions are cross-border. Proposed conclusion: some cross-border Team X acquisitions exceed £500m. Decide True, False or Cannot Say strictly from the premises (here it follows, so True).
Common traps. Confusing probable real-world truth with logical certainty, and getting stuck on visual sequences.
How to handle it. Convert verbal logic into symbols or Venn diagrams on your scratchpad, and remember 'Cannot Say' is a valid position when the parameters do not lock an outcome.
Situational judgement test
Scenario-based, within the blended set · No hard timer
What it tests. Commercial awareness, accountability, communication style, long-term decision-making and values alignment.
Worked example. At 17:30 you find a structural formula error that shifts a valuation by 8% the night before a pitch, after your associate has left. The strongest path emails the associate immediately, outlines the overnight fix plan and delivers the corrected deck, rather than hiding the error or leaving it.
Common traps. The 'robot analyst' trait of always working alone without communicating, and passive options that hide errors or shift responsibility.
How to handle it. Prioritise open communication, transparent problem-solving and team goals over individual reputation; never conceal an analytical error.
Behavioural strengths and personality
Within the blended set · No hard timer
What it tests. Behavioural consistency, authenticity, risk appetite, continuous learning and capacity for sustained workloads.
Worked example. Indicate where you sit between 'I thrive when managing multiple ambiguous tasks simultaneously' and 'I prefer clear, highly detailed instructions before starting a project'.
Common traps. Contradicting yourself by guessing what the firm wants (which triggers an unreliability flag), and choosing only middle options, which yields a flat profile.
How to handle it. Answer honestly but from your most professional, focused self, and be decisive rather than neutral.