SHL Test Scores Explained: Percentiles, Sten Scores, and What Employers See

Your raw score is the simplest layer of your SHL result: it is the number of questions you answered correctly. If the test has 25 questions and you answered 18 correctly, your raw score is 18. Raw scores are not penalised — there is no deduction for wrong answers, only for unanswered questions (which score zero).

However, raw scores are almost never what employers look at directly. An 18/25 raw score on one version of the SHL Numerical Reasoning test might represent the 72nd percentile, while an 18/25 on a different version might represent the 60th percentile — because the two versions had different average question difficulties. Raw scores on their own are not comparable across test versions or cohorts.

This is why the question "how many can I get wrong and still pass?" does not have a simple answer. A raw score of 15/25 might clear an employer's cut-off or might not — it depends on the percentile that score converts to, which depends on the norm group being used. The percentile rank is what matters, not the raw number.

A percentile rank tells you what percentage of the norm group your score beat. If your result is at the 70th percentile, you scored higher than 70% of the reference population. If you are at the 85th percentile, you outperformed 85% of that group.

The critical word here is norm group. Your percentile rank is not calculated against every person who has ever taken the SHL test — it is calculated against a specific reference population calibrated to match your candidate level and industry context. Common norm groups include: general working population, graduate level, managerial/professional, and finance/quantitative specialist groups.

A sten score (Standard Ten score) is a 10-point scale derived from your percentile rank. SHL uses sten scores as the primary summary metric on many of its employer-facing reports. The sten scale converts your percentile into a single digit between 1 and 10, where 5–6 represents the average range.

Most competitive graduate employers target candidates scoring sten 7 or above (above the 77th percentile). For highly competitive roles in finance, consulting, and technology, sten 8+ (above the 89th percentile) is often required to progress.

Employer cut-off scores vary considerably and are rarely published publicly. However, based on candidate reports and industry data, the following patterns are broadly accurate for 2026 graduate applications:

Some employers also use a banding system rather than a hard cut-off. They categorise candidates into bands (A, B, C, D) based on sten score, and then use band as a filter: all Band A candidates progress, Band B candidates are reviewed holistically, and Band C/D candidates are screened out. The specific band boundaries are set by each employer independently.

Practice test scores are genuinely useful predictors of real test performance — with one important caveat: the calibration of the practice test must match the norm group of your real test. A practice test scored against a general population norm might show you at the 75th percentile; the same performance scored against a finance specialist norm might be the 50th percentile.

When using CareerTestPrep practice tests, our scoring is calibrated to a graduate-level norm group — which is appropriate for most graduate scheme applications. If you are applying to roles where the employer uses a specialist norm group (investment banking, quantitative roles), add approximately 10–15 percentile points of additional target headroom when setting your preparation goals.

• Track your score across multiple tests: A single score tells you less than a trend. Three practice tests showing 58th → 65th → 71st percentile tells you your preparation is working.
• Weight your most recent scores more heavily: Your preparation is most relevant to where you are now, not where you were two weeks ago. If your last three scores are all above 70th percentile, that is your current level.
• Compare like with like: Compare timed practice scores to other timed practice scores. Untimed practice scores are typically 8–15 percentile points higher than your timed real-test performance.

When an employer opens your SHL score report on TalentCentral, they typically see the following information for each test you completed:

• Percentile rank: Your score relative to the norm group, expressed as a percentage (e.g., "72nd percentile").
• Sten score: The 1–10 summary score derived from your percentile rank.
• Band: An A/B/C/D categorisation if the employer has configured banding thresholds.
• Comparison to cut-off: A pass/fail indicator if the employer has set a cut-off score (this appears as a green tick or red cross in some report formats).
• Test completion status: Whether you completed the full test or exited early (which is a significant red flag).

SHL test performance improves reliably with structured practice. The following approach produces the best results:

• Identify your weak test type: Most candidates perform significantly better on one of the three test types than the others. Identify your lowest-scoring type and concentrate your practice there — marginal improvements to your weakest score produce larger percentile gains than improving an already-strong score.
• Practise under timed conditions from day one: Time pressure is the primary difficulty driver on SHL tests. Practising without a timer builds accuracy but not the speed and discipline the real test demands.
• Review every wrong answer in detail: Understand specifically why each incorrect answer was wrong — which formula you misapplied, which rule you missed, which trap you fell into. Then target that specific gap.
• Build pattern recognition, not just formulas: The question formats on SHL tests repeat. After 5–6 practice tests, you should be able to recognise each question type instantly and know which approach to apply without reading the question slowly.
• Use real SHL-format tests: Generic maths or reading comprehension tests do not build familiarity with SHL-specific formats, timing structures, or question phrasing. Use practice tests calibrated to match the actual SHL platform.