The four test families
| Test | What it measures | Typical timing | Biggest trap |
|---|---|---|---|
| Numerical reasoning | Data interpretation under time | 60-75 sec per item | Reading the wrong table column |
| Verbal reasoning | Inference from text | 50-60 sec per item | Using outside knowledge |
| Logical / inductive | Pattern recognition | 45-60 sec per item | Chasing one exotic rule |
| Situational judgement | Workplace behaviour | Untimed or generous | Answering as your ideal self |
Twelve tips, in order of impact
- 1Practise under the exact time limit. Untimed practice builds false confidence.
- 2Learn percentage change, reverse percentage and index numbers cold; they cover most numerical items.
- 3Read the question before the chart, so you know which number you are hunting.
- 4Estimate first, then confirm. Many items can be answered from magnitude alone.
- 5Never spend more than 1.5x the average time on one question.
- 6For verbal tests, answer only from the passage; "cannot say" is a real answer.
- 7In inductive tests, check the simplest rules first: rotation, count, colour, position.
- 8Use the on-screen calculator you will have in the real test, not your phone.
- 9Take the real test in the morning, rested, on a laptop with a wired connection.
- 10Do one full mock 48 hours before, not the night before.
- 11Log every wrong answer by cause: misread, method, or speed. Fix the largest bucket.
- 12For SJTs, be consistent; the key rewards ownership plus escalation, not perfection.
A worked numerical example
Revenue rose from 480m to 552m. What was the percentage increase? Compute the change, 72m, then divide by the base, 480m, giving 15%. The trap version asks what revenue was before a 15% increase to 552m, which is 552 / 1.15 = 480m, not 552 x 0.85. Reverse percentage errors are the single most common numerical mistake in finance testing.
Four-week practice plan
| Week | Focus | Volume |
|---|---|---|
| 1 | Technique: percentages, ratios, index numbers | 30 min daily, untimed |
| 2 | Timed sets by topic | 2 x 20-item sets daily |
| 3 | Full-length mocks under exam conditions | 3 mocks, reviewed line by line |
| 4 | Error log remediation and taper | 1 mock, then light drills |
Why practice works better than talent here
Psychometric tests measure reasoning speed within a format, and formats are learnable. The first time you see a numerical reasoning table, you spend fifteen seconds understanding the layout before you begin. After twenty practice questions, you understand it in two. That reclaimed time is spread across every question in the paper and is usually worth more than any improvement in underlying ability you could achieve in the same period.
This is also why cramming fails. Reasoning speed responds to repeated short exposure, not to one long session. Twenty minutes a day for two weeks will move your score materially further than eight hours on a Sunday, and it leaves you fresher on test day.
Numerical reasoning: the arithmetic is not the hard part
Most numerical questions are simple arithmetic wrapped in a comprehension problem. The errors that cost marks are almost never calculation errors; they are reading errors. Using the wrong row, missing a unit written in the table header, or answering the percentage change when the question asked for the absolute change. Before you calculate anything, read the question stem twice and identify exactly which number the answer should be.
- Check units and scaling in table headers before touching the calculator.
- Estimate first; if your computed answer is far from the estimate, you misread something.
- Learn percentage change and ratio manipulation until they need no thought.
- Skip and return rather than sinking ninety seconds into one item.
Verbal reasoning: answer only from the passage
Verbal reasoning punishes general knowledge. The correct answer is the one the passage supports, not the one that is true in the world. If a statement is plausible but the passage never establishes it, the answer is cannot say. Candidates lose most of their marks here by importing what they already believe, and the fix is to point at the sentence that justifies your choice before you select it.
Test-day conditions
Small logistics decisions have outsized effects. Sit the test at the time of day you think most clearly. Use a real calculator rather than a phone. Close every other tab. Have paper for working. And do not sit it immediately after a lecture or a shift, because these tests are sensitive to fatigue in a way that most academic exams are not.
Key takeaways
| Point | Detail |
|---|---|
| Timing is the skill | Accuracy without pace fails the cut |
| Track error causes | Misreads are cheaper to fix than method gaps |
| Simulate the platform | Interface familiarity converts to marks |
| Do not game the SJT | Consistency checks catch it |
Frequently asked questions
Can you improve at psychometric tests?
Yes. Scores on numerical and logical tests typically improve substantially over three to four weeks of timed, reviewed practice, mostly by reducing misreads and improving pace.
What is a good score on a numerical reasoning test?
Banks usually benchmark against a graduate norm group and set the pass around the upper third, though the exact threshold varies by programme and applicant volume.




