Maths For Numerical Reasoning Tests

What You Actually Need to Know Before Sitting a Numerical Reasoning Test

Most people walk into these tests completely unprepared for what the questions actually look like. They study formulas they'll never need and skip over the mental math tricks that would save them three minutes per question. I've watched candidates blow through easy problems at record speed only to completely freeze on a deceptively simple percentage change question because they'd never practiced reading data from a messy table under time pressure. Numerical reasoning tests aren't math tests in the traditional sense. They're speed-reading tests disguised as mathematics. The mathematical operations are usually basic - percentages, ratios, averages, proportions - things you learned in school. The actual challenge is extracting the right numbers from dense tables and charts in under thirty seconds, then performing the calculation correctly before the next question demands your attention. The typical test gives you between forty-five seconds to a minute per question across thirty to forty questions total. That's not a lot of breathing room when you're flipping between pie charts, line graphs, and multi-column data tables.

Practical Maths For Numerical Reasoning Tests

Here's the part nobody tells you: you rarely need a calculator on these tests, and using one incorrectly will cost you more than it saves. Most providers - SHL, Kenexa, Cubiks, Saville - design their questions so that mental math or rough estimation gets you to the right answer faster. The options are deliberately spaced apart. If you calculate 47.3% and one of the choices is 47.2% while another is 51.8%, you're safe. They don't try to trap you between two nearly identical decimals unless the question is specifically testing precision, which is rare. I remember a candidate I was coaching who kept getting tripped up on a specific type of question involving currency conversion within a multi-step table. The table listed prices in multiple currencies with exchange rates scattered in footnotes. She'd convert everything at the start, spend two minutes doing arithmetic that wasn't actually necessary, and then run out of time on the last five questions. The workaround was simple: only convert the specific cells she needed for each question. Most questions only reference two or three data points, not the entire table. Converting the whole dataset upfront is a waste of time that costs you roughly forty-five seconds per question on average. Percentages are the single most common operation and the single most common source of errors. Not because percentages are hard, but because people misread what the question is actually asking. "X increased by 20% from last year. Last year's value was 150. What is this year's value?" That's straightforward multiplication. But then they get hit with the reverse: "X is now 180, which represents a 20% increase. What was the original value?" That requires division, and people second-guess themselves and pick the multiplication answer instead. The fix is to always write out what you know and what you need before touching a number. Takes four seconds. Prevents about half the percentage mistakes I see.

Time-weighted averages show up more often than you'd expect. A company reports revenue across four quarters: £200K, £250K, £300K, £400K. What's the average quarterly revenue? Easy, right? Add them up and divide by four. But sometimes the data is presented differently. Sales figures where some quarters have different numbers of trading days, or staffing costs spread across months with different hour counts. The question tests whether you recognize that a simple average is wrong and you need a weighted calculation. I once saw a test taker miss three questions in a row on this pattern because the data presentation looked deceptively uniform. The numbers were there in the table, just not in the format she expected. Ratio problems follow a similar trap pattern. The question will give you a ratio like 3:5 and then ask for an absolute value based on a total or partial figure. The standard approach is to add the ratio parts (3+5=8), divide the known value by the total parts to find one unit, then multiply back up. This is reliable but slow if you're doing it from scratch each time. The shortcut is recognizing common ratio multiples in the answer choices. If one option is 75 and the ratio parts sum to 8, immediately check whether 75 divided by 8 gives a clean number. If it doesn't, the answer probably isn't 75. This eliminates wrong choices without full calculation and cuts response time from about forty seconds to fifteen.

Data Interpretation Strategy That Actually Works

Table interpretation is where most candidates lose the most time. A single table might contain six columns and twelve rows of data covering multiple years, regions, and product categories. The instinct is to read every cell. Don't. Skim the headers first, understand what each column and row represents, then jump straight to the question. The question will tell you exactly which cells matter. Reading the entire table upfront wastes roughly two minutes per set, which adds up to eight or nine minutes across a full test - time you're unlikely to get back. Charts and graphs require a different approach. Bar charts and line graphs are generally faster to process because the visual comparison is built in. You can see at a glance which category is highest or whether a trend is accelerating. Pie charts are the slowest because they force you to estimate angles or read explicit percentages. If a pie chart shows four segments without labels, you're doing mental geometry, which is unreliable under pressure. Always check whether the chart includes percentage labels before committing to any calculation. If it doesn't, flag the question and move on. You'll save time by not getting stuck estimating whether a slice looks like 27% or 31%. One thing I've noticed repeatedly: candidates treat every question with the same level of intensity. They give a difficult question the same time budget as an easy one and then panic when they realize they've spent three minutes on a single problem. The practical strategy is to categorize questions on sight. If a question requires more than one calculation step or involves converting between units, mark it and come back later. If it's a direct lookup or single-operation problem, answer it immediately. This sequencing alone can add five to seven minutes to your available time by preventing early bottlenecks.

Currency and unit conversions deserve their own mention because they're both common and surprisingly error-prone. The test might give you prices in pounds and ask for the total in euros, with the exchange rate provided as 1.15 euros per pound. The arithmetic is trivial, but candidates routinely multiply when they should divide or vice versa because they're rushing. Write the conversion factor as a fraction and cancel units explicitly. One pound times 1.15 euros per pound gives you euros. The units tell you whether the setup is correct before you do any multiplication. This habit takes a few seconds to develop but catches roughly sixty percent of conversion errors I've seen.

Common Pitfalls and Where These Tests Fall Short

The biggest limitation of numerical reasoning tests as an assessment tool is that they measure test-taking skill more often than they measure actual numerical ability in a work context. A candidate who's practiced these tests extensively can score in the ninety-fifth percentile without having any real analytical experience. Conversely, someone who's strong with data in their day-to-day work but has never seen a SHL-style table question might score poorly. The tests are reliable for filtering out people who can't handle basic quantitative material, but they're not a precise measure of professional competence. Another issue is the artificial time pressure. Real workplace data analysis doesn't happen with a countdown timer. You can look up a conversion factor. You can double-check your work. These tests simulate pressure that doesn't exist in most actual jobs, and the anxiety they generate can disproportionately affect candidates who perform well under normal conditions. Some providers are starting to offer untimed versions or extended time for accessibility reasons, but the timed format remains dominant across most industries. Preparation needs to be targeted. Spending a month memorizing mathematical formulas is wasted effort. Instead, practice with actual test-style questions under timed conditions. Focus on speed-reading tables, recognizing which operation a question requires within five seconds, and estimating answers before calculating precisely. The online practice platforms from the major providers - SHL's own practice materials, Graduates First, JobTestPrep - are adequate for building familiarity. The goal isn't to learn new math. It's to train your brain to stop second-guessing itself and execute the obvious operation quickly.

I'd recommend doing at least twenty practice questions per session with a timer running, reviewing every mistake immediately to understand whether it was a calculation error, a misreading of the question, or a time-management failure. Most errors fall into the second category. The third category is fixable with practice. The first category usually disappears once you slow down enough to write out your working before committing to an answer. It feels slower at first, but it's actually faster than recalculating after you've selected the wrong option because you misread the question.