Numerical reasoning tests are the most widely used cognitive assessment tool in graduate and professional hiring, particularly in banking, consulting, finance, and the public sector. They test your ability to interpret numerical data under time pressure, not advanced mathematics, but the applied analytical thinking that underlies good professional judgment. This guide covers what numerical reasoning tests actually assess, the main question formats you'll encounter, the preparation strategies that make a measurable difference, and the common mistakes that cost candidates scores they should have got.
What Numerical Reasoning Tests Actually Measure
A common misunderstanding: numerical reasoning tests don't test how good you are at mathematics. They test how accurately and quickly you can extract and interpret information from numerical data, charts, tables, graphs, financial summaries. The maths involved is usually GCSE-level arithmetic: percentages, ratios, proportions, basic statistics. The difficulty comes from the time pressure, the data volume, and the question format, not from the complexity of the calculations.
What good numerical reasoning performance actually requires:
- Quick identification of the relevant data in a table or chart (filtering out noise)
- Accurate percentage and ratio calculation without errors under time pressure
- Reading multi-stage questions that require combining information from multiple sources
- Calm, systematic working under a time limit that makes most candidates rush and make avoidable errors
The tests are predictively valid for professional performance because these skills reflect the kind of quantitative reasoning used in analysis-intensive roles: extracting the right numbers from a data set, performing the right calculation, and arriving at an accurate conclusion quickly.
Main Question Formats
The most widely used numerical reasoning tests (SHL Verify, Korn Ferry, Talent Q, Cubiks, Saville Assessment) use similar question structures:
Data interpretation from tables. A table presents company figures, market share data, financial results, or population statistics. Questions ask for specific values, percentage changes, ratios, or comparisons. The challenge: the relevant numbers are buried in a larger data set and the time limit requires quickly finding and using them.
Graph interpretation. Bar charts, line graphs, pie charts, and combination charts with multiple series. Questions typically involve reading off values, calculating changes between periods, or comparing categories. Multi-series charts require attention to the correct line or bar.
Financial ratios and percentages. Questions involving profit margins, year-on-year growth, market share calculations, cost-to-revenue ratios. These are standard GCSE calculations but require accuracy and speed in context.
Multi-step problems. Questions that require two or three calculations in sequence, often combining data from multiple parts of the table or chart. These are the questions that catch candidates who rush, each step needs to be done correctly for the final answer to be right.
Preparation Strategies That Work
The research on practice effects for numerical reasoning is fairly consistent: practice can improve scores by 5–15% through the mechanisms of format familiarity, reduced test anxiety, and identification of your weak areas. It doesn't significantly improve underlying quantitative ability in the short term. This means preparation should focus on maximising the gains from familiarity rather than trying to learn new mathematics.
Use official or publisher-matched practice materials. SHL practice questions are more representative of SHL tests than generic numerical reasoning exercises. Each publisher has format quirks, question length, table complexity, chart types, that you benefit from encountering in practice rather than in the live test.
Practice under timed conditions. The time pressure is the primary difficulty. Doing practice questions without a timer produces unrealistic performance data and doesn't develop the pacing discipline you need. Use your target test's actual time per question (typically 45–75 seconds for SHL-style tests) from the start of your practice.
Review your errors with precision. For every question you get wrong, identify exactly where the error occurred: reading the wrong row/column, a calculation error, a misunderstood question, or a conceptual misunderstanding. Different error types need different remediation.
Speed up percentage and ratio mental arithmetic. There are a small number of calculation types that appear repeatedly. Practising quick percentage calculations (especially finding what percentage one number is of another, and calculating percentage changes) until they're fast and accurate pays disproportionate dividends.
Time Management in the Test Itself
Time management during the test is where most candidates lose avoidable points. Common patterns:
- Spending too long on a difficult question at the expense of easier later ones, skipping and returning is often better than spending four minutes on one question
- Rushing multi-step calculations and making arithmetic errors, the cost of speed here outweighs the benefit
- Reading the question after looking at the data instead of before, reading the question first tells you which numbers you're looking for, reducing the time spent processing irrelevant data
A systematic approach: read the question first, locate the relevant data, perform the calculation carefully, select the closest answer. "Closest answer" is important, numerical reasoning tests are often multiple choice with options clustered around plausible values, so rough calculations sometimes lead you to the wrong option.
Calculator Policy
Most online numerical reasoning tests (those administered via SHL's OPQ platform, Korn Ferry, etc.) permit on-screen calculators. Some assessment centres administer paper versions without calculators. Find out the specific policy for your assessment. If a calculator is available, use it for multi-step calculations, mental arithmetic errors under pressure are common and avoidable. If not, practise mental arithmetic for the most common calculation types.
Your broader cognitive profile, how you process quantitative information alongside other cognitive dimensions, matters for roles that use numerical assessments in selection. Take the free IQ test to see your cognitive strengths across numerical, verbal, and abstract reasoning.
Frequently Asked Questions
Is it possible to significantly improve numerical reasoning test scores with practice?
Yes, within limits. Format familiarity, reduced anxiety, and targeted arithmetic practice reliably improve scores. The effect is typically 5–15%, which can be meaningful if you're close to a cut-off. Substantial improvement in underlying quantitative ability, the thing tests are trying to measure, takes longer than typical pre-test preparation time. Candidates who score in the bottom 20% on an initial practice test are unlikely to reach the top 30% through practice alone; candidates who score in the middle have more room to improve through preparation.
What's the difference between numerical reasoning and numerical ability tests?
Numerical ability tests are typically speed tests of basic arithmetic: how quickly can you add, subtract, multiply, and divide? They're more common in administrative and operational role selection. Numerical reasoning tests present data in context and require judgment about what to calculate and how to interpret the result, more relevant for professional roles where data interpretation is part of the job. The preparation strategies overlap but numerical reasoning requires more focus on data interpretation and less on raw calculation speed.
How are numerical reasoning tests scored?
Most professional-grade numerical reasoning tests produce norm-referenced scores, your result is expressed as a percentile relative to a comparison group (often graduates, management candidates, or the specific professional population the test is designed for). This means a raw score of 70% correct might be 85th percentile if the test is hard and your competition scored lower on average. Employers typically set cut-off percentiles (40th, 60th, 70th) rather than absolute scores. The cut-off for graduate roles at major investment banks is typically around the 70th–80th percentile on their comparison group.
Do numerical reasoning tests disadvantage candidates from certain backgrounds?
There is evidence for differential performance by educational background, socioeconomic background, and in some studies, gender, though the gender effects are smaller and less consistent than the educational background effects. The differential is primarily attributable to differential exposure to quantitative problem-solving and test-taking in educational settings. Test publishers work to minimise this through differential item functioning analysis, but some differential remains. Some employers supplement or replace numerical tests with work-sample tests for this reason.
How long before a test should I start preparing?
Two to four weeks of focused practice (30–45 minutes per day) is typically sufficient to get most of the preparable gains. Starting earlier gives you more time to identify and address specific weak areas; starting much earlier but doing sporadic practice is less effective than intensive focused preparation closer to the test date. The day before: do one timed practice set to maintain familiarity, then rest, fatigue is a more common performance limiter on test day than inadequate preparation.