Aptitude Test Scores Explained: Percentiles, Norm Groups and What Banks Actually Require

Every application season, thousands of UK candidates finish an online aptitude test, receive either a rejection or an invitation, and never find out what their score was or what would have been enough. Employers do not publish pass marks. Test providers do not reveal cut-offs. And most preparation advice skips the question entirely. This article explains how aptitude test percentiles and norm groups actually work, what is genuinely known about the thresholds at competitive employers, and why the whole system makes one particular piece of information — your real current level — more valuable than any amount of blind practice.

Aptitude test percentiles: your raw score is not your result

On almost every major aptitude test — SHL, Aon (cut-e), and their peers — the number of questions you answer correctly is only the input. What the employer sees is a percentile: your performance ranked against a comparison population called a norm group. A result at the 72nd percentile means you performed better than 72% of the people in that comparison group.

This is called norm-referenced scoring, and it has a consequence candidates rarely appreciate: there is no fixed "pass mark" in the sense of a percentage of correct answers. Answering 75% of questions correctly could be an excellent result against one norm group and a mediocre one against another. The test is not asking "can you do this maths?" It is asking "can you do this maths better than the other people we are considering?"

Norm groups: the invisible variable

The norm group is chosen by the employer from options the test provider offers — general population, graduates, managerial candidates, sector-specific pools, and so on. The same raw performance lands at very different percentiles depending on that choice. Compared against the general adult population, a numerate finance applicant looks exceptional; compared against other finance applicants — a heavily self-selected, well-practised group — the same person may be merely average.

This is why investment banking tests have a reputation for being brutal even though the underlying maths is GCSE-level. The questions are not harder than average; the comparison group is. Banks calibrate against highly numerate applicant pools, so the effective bar sits far above what the question difficulty alone would suggest. It is also why hearing "the test was easy" from a friend who sat the same test for a different employer tells you almost nothing.

What aptitude test percentile banks actually require

No major bank, consultancy or accounting firm publishes its cut-off, and any guide quoting an exact official threshold is inventing precision that does not exist. What we have instead are consistent estimates from across the preparation industry and candidate communities, and they cluster tightly enough to be useful:

• Competitive graduate employers are widely estimated to sift somewhere around the 70th–80th percentile on cognitive tests.
• For top-tier investment banking and the most selective consulting firms, the commonly cited working assumption is the 80th percentile as a floor, with some estimates putting the safe target at the 85th or higher.
• The Big 4 and large corporate schemes are generally estimated to sift lower than front-office banking — though "lower" still means comfortably above average, against an above-average pool.

Treat every one of those numbers as an industry estimate, not a published fact — firms adjust thresholds by role, by division and by year, and some use test results as one weighted input rather than a hard gate. But the strategic picture the estimates paint is reliable: at competitive UK finance and consulting employers, roughly speaking, you need to be in the top fifth of an already strong applicant pool. Rejection rates support this — it is widely reported that the majority of applicants to large graduate schemes are filtered out at or before the testing stage.

Why most candidates have no idea where they stand

Here is the uncomfortable structural problem. The information that matters is relative — your percentile against a strong applicant pool — but everything a typical candidate does obscures it.

Free practice tests offer unlimited retakes, so scores inflate with repetition: your fifth attempt at a familiar question bank measures your memory of that bank, not your ability. Practice sites score against unknown or generous baselines, so a flattering "85%" may mean very little against a finance-calibrated norm group. And practising at home, untimed or loosely timed, removes exactly the pressure the real test applies. The result is a candidate who has done hundreds of questions, feels prepared, and still has no honest estimate of the one number employers will judge them on.

The fix is conceptually simple: measure yourself once, properly, before you start training. One sitting, real time limits, unfamiliar questions, no retake — the same conditions as the real thing. That single measurement tells you three things blind practice never will: roughly where you sit relative to the standard, which section (numerical, verbal, abstract) is dragging you down, and how much preparation time the gap actually requires. Then, when application season arrives, your practice is targeted at a known weakness instead of spread thinly over everything.

The practical takeaway

Aptitude test scoring is relative, opaque and calibrated against strong applicant pools — three facts that together explain most of the anxiety around it. You cannot control the norm group, the cut-off or the other applicants. You can control whether you walk into application season knowing your real level, and whether your preparation time goes where the measurement says it should.

Before you start preparing, find out where you actually stand. The Acumen Logic Benchmark is a free, one-shot diagnostic — 20 questions under real timed conditions, taken once, like the real thing. You get a percentile, a section-by-section breakdown, and an honest starting point — which is exactly the piece of information this entire system is designed to withhold from you.