SHL Numerical Reasoning Test Practice: Formats and Answers
SHL numerical reasoning test practice with real formats, timings, two worked questions and the scoring rules. Learn what employers actually score you on.
SHL·Aptitude Test Prep Team · Sep 2, 2026 · 8 min read
SHL numerical reasoning test practice works best when you practise the exact format your employer sent you, not a generic maths quiz. SHL runs several numerical products, and they differ in length, item style and pacing. The classic multiple-choice Numerical Ability test gives you 16 questions in 20 minutes. The newer Verify Interactive Numerical Reasoning test gives you 10 tasks in 18 minutes and asks you to build the answer by dragging chart elements instead of picking from four options. Both use workplace data: revenue tables, headcount charts, unit costs, conversion rates. Neither tests maths you have not already met at school. They test whether you can read a table under pressure, pick the two numbers that matter, and finish. This guide shows the formats, two full worked questions, and the pacing that gets you through.
TL;DR
- SHL Numerical Ability (multiple choice) is 16 questions in 20 minutes. Verify Interactive Numerical Reasoning is 10 tasks in 18 minutes.
- The maths is percentages, ratios, percentage change and unit rates. The difficulty is data extraction and time, not arithmetic.
- You are allowed a calculator on SHL numerical tests, so practise with the one you will use on the day.
- Scores are percentiles against a norm group, not raw marks. Around 40 to 50 percent correct is often an average result.
- Interactive items ask you to adjust a chart or slider, so partial credit is possible and blind guessing is worth less.
- A supervised verification test may follow. Your unsupervised score has to hold up.
What is the SHL numerical reasoning test?
The SHL numerical reasoning test is a timed assessment of how well you interpret numerical data and reach a correct conclusion. SHL describes the interactive version as measuring your ability to work with numbers and apply appropriate mathematics in different situations, covering word problems, calculations and chart interpretation. In practice you get a table or a chart, a short question, and a tight clock.
Employers use it as an early sift. Banks, graduate schemes, tech firms and Big Four audit and tax intakes all screen with it before a human reads your CV. If you want a quick calibration before you read further, try five free timed numerical questions and see how your pace compares to the 60 to 90 seconds per question these tests allow.
How long is the SHL numerical test and how many questions?
The name on your invitation email matters. These are the current SHL numerical products and their published lengths.
| SHL test | Questions | Time | Item style |
|---|---|---|---|
| Numerical Ability | 16 | 20 minutes | Multiple choice, tables and charts |
| Verify Interactive Numerical Reasoning | 10 | 18 minutes | Drag and adjust charts |
| Verify Interactive Numerical Calculation | 12 | 10 minutes | Short calculation entry |
| Verify G+ (numerical section) | 8 to 10 of 30 | Part of 36 minutes | Mixed with inductive and deductive |
Two things follow from the table. First, the interactive tests give you more time per item, roughly 108 seconds, because building an answer takes longer than clicking one. Second, if you were invited to Verify G+, numerical is only a third of the test, so drilling numerical alone will not carry you. Read the SHL general ability test guide before you plan that revision.
What does an SHL numerical question actually look like?
Here are two questions in the SHL house style, with the full working.
Worked example 1: percentage change across two regions
A distribution business reports units shipped by region.
| Region | Q1 units | Q2 units |
|---|---|---|
| North | 12,400 | 14,880 |
| South | 9,600 | 8,640 |
Question: how many percentage points separate the North region’s percentage change from the South region’s percentage change between Q1 and Q2?
Working. North changed by 14,880 minus 12,400, which is 2,480. Divide by the Q1 base: 2,480 divided by 12,400 is 0.20, so a 20 percent increase. South changed by 8,640 minus 9,600, which is minus 960. Divide by the base: 960 divided by 9,600 is 0.10, so a 10 percent decrease. The gap between a 20 percent rise and a 10 percent fall is 30 percentage points.
Answer: 30 percentage points.
The trap is the word “points”. A careless reader computes 20 minus 10 and answers 10. Another careless reader divides by the Q2 figure instead of the Q1 figure. Percentage change is always measured against the earlier value.
Worked example 2: revenue share that does not move
A software firm reports revenue by product line.
| Product | 2025 revenue (GBP million) |
|---|---|
| A | 18 |
| B | 12 |
| C | 6 |
| D | 4 |
In 2026, Product A grows by 10 percent and total revenue reaches GBP 44 million. What share of 2026 revenue does Product A represent?
Working. The 2025 total is 18 plus 12 plus 6 plus 4, which is 40. Product A in 2026 is 18 multiplied by 1.10, which is 19.8. Product A’s share is 19.8 divided by 44, which is 0.45, so 45 percent.
Answer: 45 percent.
The instructive part is the comparison. In 2025, A was 18 divided by 40, also 45 percent. Product A grew and its share stayed flat, because the rest of the business grew at the same rate: 22 rising to 24.2 is also 10 percent. SHL writes distractors around exactly this confusion, usually offering an answer near 50 percent for candidates who assume growth must mean a bigger slice.
For more drilling on this family of questions, work through the numerical reasoning percentages and ratios guide.
How is the SHL numerical test scored?
Your raw number of correct answers is converted into a percentile against a norm group of comparable candidates. A percentile of 70 means you scored higher than roughly 70 percent of that comparison group. Published guidance from test-prep publishers puts an average result around 40 to 50 percent of items correct, and a 90 percent result in roughly the top 10 percent of candidates. Employers then set their own cut, which is why the same score passes one graduate scheme and fails another.
SHL also notes that its interactive assessments use a partial-scoring method, which lets the test ask fewer questions while keeping accuracy. That has a practical consequence: on an interactive chart item, a nearly correct configuration can earn something, so never leave an interactive item untouched.
What is the SHL verification test?
SHL separates unsupervised ability tests from a supervised verification test. The verification test is a shorter, monitored assessment that checks whether your original score is genuinely yours. It is usually sat later in the process, sometimes at an assessment centre or under remote proctoring with a webcam.
This is the reason to avoid help on the first sitting. A large gap between your unsupervised and supervised scores is the exact signal the verification test exists to catch.
How should you practise in the final week?
Volume alone does not move a numerical score. Structure does.
- Sit one full-length timed set every other day, not five sets in one evening. Fatigue distorts your accuracy reading.
- Log every wrong answer into one of three buckets: misread the data, wrong method, ran out of time. The bucket that dominates is the only thing worth training.
- Rehearse the calculator you will use, including the percentage and memory keys. Fumbling the device costs more than weak mental arithmetic.
- Practise skipping. On a 20 minute test, two minutes lost on one stubborn item can cost you three easy items at the end.
- Read the axis labels and the unit line first, before the question. Thousands versus millions is the most common silent error.
The general pacing habits are covered in more depth in the numerical reasoning test tips guide, and the wider process of taking any online sift is in how to pass online assessment tests.
What mistakes cost most marks?
Candidates rarely fail on arithmetic. They fail on reading. The four recurring errors are using the wrong base for a percentage change, mixing a table footnote unit with the chart axis unit, answering the question they expected rather than the question asked, and refusing to abandon a hard item. A fifth is subtler: bringing outside knowledge. If the passage or table does not contain a number, it is not available to you, even if you know the real figure.
Frequently Asked Questions
Can I use a calculator on the SHL numerical reasoning test?
Yes. SHL numerical tests are designed on the assumption that you have a calculator. Use a basic physical one you have practised with rather than an unfamiliar on-screen tool.
What is a good score on the SHL numerical test?
There is no universal pass mark. Your percentile is compared to a norm group and the employer sets the cut. Aiming for the top third of candidates is a sensible working target for competitive graduate schemes.
How many times can I take the SHL numerical test?
That is the employer’s decision, not SHL’s. Many employers allow one attempt per application cycle and treat scores as valid for around a year, so a weak result can follow you into the next round of applications.
Is the interactive version harder than the multiple choice version?
It is not harder mathematically. It is harder to rush, because you build the answer rather than recognise it. You cannot eliminate options, so guessing is far less useful.
Should I practise verbal and inductive tests too?
If your invitation names Verify G+, yes, since numerical is only one of three strands. See the SHL verbal reasoning test guide and the SHL inductive reasoning test guide.