Numerical Critical Reasoning Test: Complete Guide 2026

Numerical Critical Reasoning (NCR) is a more advanced form of numerical aptitude assessment than the standard SHL Numerical Reasoning test. While both formats measure your ability to work with data, the NCR test goes further — demanding multi-step reasoning, interpretation of complex business scenarios, and the ability to draw conclusions from ambiguous or layered data sets.

The term "Numerical Critical Reasoning" refers to several related but distinct test variants. SHL's older Series 7 and NT-series tests — sometimes called the harder end of SHL's numerical catalogue — are characterised by denser text, more complex data presentations, and questions that require genuine interpretation rather than simple data retrieval. Deloitte uses a specific NCR test in their own hiring process that has become well-known among graduate applicants. More broadly, any advanced numerical test used by consulting and finance firms that requires multi-step reasoning and business judgement falls within this category.

The critical difference is in the word "critical." In a standard numerical reasoning test, you are largely reading data from a table or chart and applying one calculation. In an NCR test, you must critically evaluate which data is relevant, understand the business context, perform multiple linked calculations, and arrive at a defensible conclusion — all under time pressure that is typically more severe than standard tests.

The Three Variants You'll Encounter

• SHL Series 7 / NT-series: SHL's harder numerical formats, deployed for roles requiring strong analytical capability. These feature multiple data sources per question (a table alongside a passage of text, for example), and often require you to interpret what the data means rather than simply calculate a single figure. They were developed for roles at a higher level of analytical demand than entry-level screening tests.
• Deloitte NCR test: Deloitte uses a specific Numerical Critical Reasoning assessment in their graduate and experienced hire processes. It is well-documented by candidates and is considered one of the harder employer-specific numerical tests in professional services. It combines standard table-based data with more complex multi-step scenarios, and is benchmarked against an analytically strong norm group.
• General advanced numerical reasoning: Investment banks, management consultancies, and high-demand finance roles frequently use tests from providers including cut-e, Korn Ferry Talent Q, and custom platforms that test NCR-style skills under different names. If a test description mentions "multi-step," "business scenario," or "critical reasoning with data," you are in NCR territory regardless of the exact label.

Understanding which variant you face matters because the preparation emphasis differs slightly. For Deloitte's NCR, industry knowledge and comfort with financial metrics helps. For SHL's harder numerical formats, speed and accuracy with compound calculations is paramount. This guide covers all three, with particular emphasis on the Deloitte NCR given how frequently candidates encounter it.

NCR-level numerical testing is the norm rather than the exception at the most analytically demanding graduate employers. The following firms either use a named NCR test or deploy numerical assessments that operate at NCR difficulty:

Other Key Employers

If you are applying to any of these firms, NCR-level preparation is warranted regardless of whether the test is formally labelled "NCR." The analytical demands are comparable, and the preparation strategies in this guide apply across all of them.

NCR tests use several recurring question types that differ in meaningful ways from standard numerical reasoning. Understanding each type — and the approach each demands — is one of the highest-leverage preparation steps you can take.

1. Percentage Change Calculations

These appear in most NCR tests. They require you to calculate the percentage change between two figures drawn from different parts of a data set — sometimes requiring you to first derive the figures themselves before calculating the change.

A question might show a table of revenue figures for five product lines across four years. You are asked: "By what percentage did the combined revenue of Product Lines A and C change between 2021 and 2023?" This requires summing two rows for 2021, summing two rows for 2023, then applying the standard percentage change formula: (new − old) / old × 100. A standard test would give you the aggregate directly. The NCR format requires the intermediate step.

2. Ratio and Proportional Analysis

Ratio questions ask you to compare two or more quantities and express the relationship — often as a ratio, fraction, or proportion of a total. In NCR tests, the figures needed are rarely presented directly. You must calculate each component first.

For example, a question might present operating cost data for four business divisions and ask: "In 2022, what was the ratio of Division A's costs to Division B's costs, expressed to two decimal places?" The complication is that costs may be presented as percentages of a total, requiring you to first convert to absolute figures before computing the ratio.

3. Compound Annual Growth Rate (CAGR)

CAGR questions are a signature feature of NCR tests used by consulting firms. They require a specific formula — CAGR = (End Value / Start Value)^(1/n) − 1 — and the ability to interpret the result in context.

A typical question shows a table of operating costs for four divisions across three years. You are asked: "What was the compound annual growth rate of Division B's costs from 2021 to 2023?" This requires extracting the correct figures (Division B's 2021 and 2023 values), calculating the ratio (End/Start), applying the CAGR formula with n = 2 years, then selecting the correct answer from options that are closely spaced. Getting the formula direction wrong — or using n = 3 instead of n = 2 — produces a plausible-looking but incorrect answer. This is a deliberate trap in well-designed NCR tests.

4. Profit, Loss, and Margin Analysis

These questions present financial scenario data — revenue, costs, overheads, tax rates — and require you to calculate a profitability metric. The challenge is that the data is layered: you may need to calculate gross profit first, then apply an overhead figure from a different part of the table, then calculate the net margin as a percentage of revenue.

For example: "A company's revenue was £480m in 2023. Cost of goods sold represented 62% of revenue. Overhead costs were £58m. What was the operating profit margin?" This is a three-step calculation: derive COGS (£480m × 62% = £297.6m), subtract overheads (£480m − £297.6m − £58m = £124.4m), then express as a margin (£124.4m / £480m = 25.9%). The answer choices will include values corresponding to making each intermediate error.

5. Currency Conversion and Index Problems

These questions require converting values between currencies or reindexing data to a base year. Exchange rates are provided in a separate table, and you must correctly identify which rate to apply and in which direction. Index problems require understanding that an index value of 115 means 15% above the base (not 115% above).

6. Multi-Source Interpretation Questions

Unique to NCR tests, these combine a data table with a short written passage. The passage provides context (e.g., "The company announced a 12% restructuring charge in Q3") that modifies the figures in the table. You must integrate both sources to answer correctly — the table alone or the passage alone gives an incomplete picture.

This is the question type that most clearly distinguishes NCR from standard numerical reasoning. It is also related to the skills tested in Watson Glaser critical thinking — the ability to combine information from multiple sources to reach a defensible conclusion.

Many candidates prepare for NCR tests using standard numerical reasoning practice materials and are caught off guard by the additional complexity. Understanding exactly where NCR diverges from the standard SHL Numerical Reasoning format allows you to focus preparation on the specific gaps.

NCR questions reward systematic technique over raw speed. Candidates who develop and consistently apply a small set of deliberate strategies consistently outperform those who rely on intuition alone.

1. Scan All Data Before Reading the Question

Spend 10–15 seconds scanning the full data set before reading the question. Note what each table/column/chart contains, identify the units (£m vs £k, %), and spot any footnotes. This prevents the costly mistake of using the wrong data source when under time pressure. Many NCR questions are specifically designed to catch candidates who read the question first and grab the first plausible-looking figure.

2. Estimate Before Calculating

For most NCR questions, you can eliminate two or three answer options through estimation before doing any precise arithmetic. Round all figures to one significant figure, perform the mental calculation, and see which answer option is in the right ballpark. This turns a 5-option question into a 2-option question, dramatically improving accuracy even if your final calculation is slightly off.

3. Write Out Intermediate Steps

In multi-step NCR questions, write down each intermediate result on your rough work paper before proceeding to the next step. Holding two or three numbers in working memory under time pressure creates errors. The few seconds it takes to write "Step 1: £297.6m" before calculating Step 2 is consistently faster overall because it eliminates the costly errors that require starting over.

4. Track Units Through Every Step

Unit errors are the most common source of avoidable mistakes in NCR tests. The data may present values in £m, but ask for an answer in £k; exchange rates may be expressed as GBP/USD rather than USD/GBP. Before writing your final answer, confirm that the unit you have calculated matches the unit the question asks for. This check takes under five seconds and prevents some of the most avoidable incorrect answers.

5. Eliminate Impossible Answers First

NCR answer options are designed to catch specific errors. Identify what a clearly wrong answer looks like (e.g., a negative number when the question asks for a growth rate from rising figures, or a value well outside any plausible range) and eliminate those immediately. This narrows your effective search space and reduces cognitive load.

6. Know Your Core Formulas Cold

You should be able to apply the following formulas without pausing to recall them: percentage change, percentage of a total, CAGR, profit margin, markup vs margin distinction, index calculation, and currency conversion. Practise these until they are completely automatic. Under NCR time pressure, retrieving a formula from memory costs time and attention you cannot afford.

Time pressure is one of the defining features of NCR tests. At 25 questions in 20 minutes, you have an average of 48 seconds per question — compared to 60 seconds in a standard SHL test. The complication is that NCR questions individually take longer due to their multi-step nature. This means you must actively manage time, not just work quickly.

The Triage Approach

Not all NCR questions take the same time. Some questions present straightforward data that requires a two-step calculation; others present dense multi-source scenarios that take significantly longer. A strong time management strategy involves quick triage: in the first few seconds of each question, assess whether the data is simple or complex. If simple, proceed and aim to complete in under 40 seconds. If complex, allocate 60–70 seconds — but never more than 90 seconds on a single question.

The "Flag and Return" Rule

If you are stuck or confused after 60 seconds, flag the question, make your best guess, and move on. Returning to it with fresh eyes after completing the remaining questions is almost always more effective than spending 2–3 minutes on a single question mid-test. Most NCR platforms allow question flagging — use it actively, not as a last resort.

You cannot develop genuine NCR time management by reading about it. You must practise under timed conditions repeatedly. Use our timed practice tests and track not just your accuracy but your time per question. The target is a consistent average of under 50 seconds with accuracy above 70%.

NCR test performance is frequently limited by a small number of recurring mistakes. Identifying and eliminating these mistakes in practice has a disproportionate impact on your score.

Mistake 1: Using the Wrong Year or Column

NCR data tables are designed with multiple columns that look similar. Under time pressure, candidates frequently pull data from 2022 when the question asks for 2021, or from the wrong division. Prevention: before extracting any figure, re-read the specific row/column identifiers the question specifies. This takes five seconds and eliminates one of the most common error types.

Mistake 2: Confusing Margin and Markup

Profit margin is expressed as a percentage of revenue: (Profit / Revenue) × 100. Markup is expressed as a percentage of cost: (Profit / Cost) × 100. NCR tests frequently feature questions where candidates who confuse these two will find their incorrect answer available as a distractor option. Always check whether the question asks for margin or markup, and apply the denominator accordingly.

Mistake 3: Misapplying the CAGR Formula

The most common CAGR error is using the wrong value for n (the number of years). For data spanning 2020 to 2023, n = 3 (not 4). The formula computes the rate of growth per year over the period between the two endpoints, so you count the gaps between years, not the number of years listed. Practice with explicit attention to this distinction until it is automatic.

Mistake 4: Ignoring Units in the Question

Data tables commonly present figures in £millions, but the question may ask for an answer in £thousands (or vice versa). Answer options are calibrated to catch the unit-conversion error — the "wrong unit" answer will appear plausible. Build a habit of checking the question's requested unit against the table's unit before finalising your answer.

Mistake 5: Not Checking the Reasonableness of the Answer

After calculating, spend two seconds asking: "Does this answer make sense in context?" A profit margin of 150% or a CAGR of 95% for a stable business division should trigger a recheck — they are mathematically possible but contextually implausible. This sanity check catches sign errors and formula inversion mistakes that calculation alone does not catch.

Mistake 6: Over-Preparing for the Wrong Test Format

Candidates who have only practised standard SHL numerical reasoning sometimes feel confident going into an NCR test and are blindsided by the additional complexity. Confirm the test format in your invitation and ensure your practice materials match the actual test. For Deloitte specifically, use practice materials that explicitly replicate the Deloitte aptitude test format.

Three weeks of structured preparation is sufficient to meaningfully improve your NCR score — provided preparation is active (practising and reviewing errors) rather than passive (reading about the test). This plan assumes approximately 45–60 minutes per day.