How to Solve Numerical Reasoning Questions
Learn the methods behind percentage change, ratios, tables, charts, averages, rates, finance questions and data sufficiency — with worked examples and the common mistakes to avoid. Learn it, then drill it.
The maths on a numerical reasoning test is rarely the hard part — it’s GCSE-level arithmetic. What separates candidates is method and speed: knowing the one reliable way to set up a percentage change, a ratio share or a reverse-percentage problem, and executing it without second-guessing under a clock that gives you barely a minute a question.
Each guide is built to be used, not just read. Try a question first to surface what you don’t know, then learn the method, study the worked examples, and see the named mistakes the test is designed to catch — so you recognise the trap before it costs you. Then drill the area and prove it. If you don’t yet know which areas to study, take the free diagnostic first; it points you straight at your weakest patterns.
% of, % change, reverse %, successive %, percentage points, shares of a total.
Read the guideRatio questions — sharing in a ratio, comparing rates, direct & inverse proportion, fractions of a quantity.
Read the guideReading values from a table, totals, differences, and locating the right row/column.
Read the guideData interpretation: reading bar, pie and line charts and combining values off them.
Read the guideAverage questions: mean, median, range and weighted averages.
Read the guideSpotting the rule in a sequence and continuing it.
Read the guideInterest, break-even, depreciation, margins, growth (CAGR), yield, index numbers, real-terms.
Read the guideSpeed-distance-time, currency, per-capita and work-rate problems.
Read the guideBasic probability, complement, expected value, combined events, Venn diagrams.
Read the guideSolving linear equations and simultaneous equations.
Read the guideNo-calculator fluency: the four operations, fraction/decimal/percent conversion, percentage shortcuts.
Read the guideTrue / False / Cannot-say judgements from given data.
Read the guide