Aon Numerical Reasoning: Navigate Data Sheets and Handle Cannot Say

An Aon numerical reasoning question can require very little arithmetic and still be easy to miss. You select the wrong data sheet, use a forecast as an actual result, or treat an unavailable value as zero. The calculation is then neat, fast, and irrelevant.

For statement-based questions, prepare a four-step habit: locate the relevant sheet, identify the exact comparison, calculate only what is supported, and decide whether the statement is true, false, or cannot be determined. This guide uses an original worked dataset to practice that habit. It does not reproduce live assessment questions or promise that every Aon invitation has the same format. Related drills are available in Statistics and Math.

The image is a conceptual illustration, not an Aon screen or the Cedar exercise dataset below.

Confirm which numerical assessment you were invited to take

Aon offers multiple assessment modules. The official candidate preparation page currently lists a numerical reasoning module with 37 tasks and a 12-minute limit, while applied numeracy is described separately. Treat those as module descriptions, not a guarantee about your employer's invitation, accommodations, or combination of tests.

The official numerical reasoning practice PDF uses data sheets and statements with True, False, and Cannot Say responses. Its instructions describe selecting relevant information and navigating the sheets; practice instructions are not the scored session itself. Check the invitation and introductory tutorial before transferring any strategy.

That distinction changes preparation. A statement-evaluation task rewards evidence discipline as well as arithmetic. An applied numeracy task may ask you to calculate a numerical answer instead. Do not spend your preparation memorizing a three-option response method if your actual module asks a different kind of question.

Keep employer-specific claims separate. This article does not establish a company's selection stages, cutoff score, retest rules, or use of negative marking. Where your invitation is silent, do not invent those rules from a practice website or another candidate's experience.

Use a lookup routine that prevents wrong-sheet answers

Read the statement for four anchors: entity, period, measure, and comparison. “Cedar's on-time delivery rate increased by five percent in FY2025” is about Cedar, FY2024 versus FY2025, a rate, and relative percentage growth. It is not asking whether the number of deliveries increased.

Next, find the sheet whose headings supply that measure. Scan titles and column labels before inspecting every cell. Confirm the unit: thousands of dollars, whole dollars, percentages, or counts. Check whether the period is a fiscal year, quarter, month, or forecast. Only then write the calculation.

For the format shown in Aon's numerical practice material, each statement's needed information is associated with one sheet. Practice identifying that sheet rather than stitching unrelated cells into an answer. In the original exercise below, the Costs sheet deliberately does not contain everything needed to establish profit; importing an unsupported revenue assumption would change the task.

Finally, ask whether the comparison can be resolved at all. If the required baseline is absent, arithmetic will not recover it. If the statement is contradicted by the supplied numbers, answer False. If both a true and a false interpretation remain possible under the supplied information, answer Cannot Say.

Original exercise: three Cedar data sheets

Use only the information explicitly provided. All amounts are fictional. FY2026 revenue is a forecast, not an achieved result. A dash means not supplied, not zero.

Revenue sheet period	Revenue, $ millions
FY2023 actual	—
FY2024 actual	1.20
FY2025 actual	1.50
FY2026 forecast	1.80

Operations sheet	FY2024 actual	FY2025 actual
Eligible orders	1,000	2,000
On-time eligible orders	800	1,700

Costs sheet	FY2024 actual	FY2025 actual
Fixed operating costs, $ thousands	200	250
Total variable operating costs	Not supplied	Not supplied
Revenue	Not supplied	Not supplied

Before reading the answers, classify these six statements and note the sheet you used. Avoid writing a formula until you have named the relevant values.

Statement	Your first check
Revenue increased by 25% from FY2024 to FY2025.	Which year is the growth denominator?
Revenue growth from FY2023 to FY2024 exceeded 10%.	Is the baseline supplied?
The on-time rate increased by five percentage points.	Calculate both rates first.
The on-time rate increased by 5% relative to FY2024.	Is percent the same as percentage points?
Fixed operating costs increased by 25%.	Convert units only if necessary.
Operating profit increased in FY2025.	Does this sheet establish total profit?

Work the answers without adding assumptions

The first statement is True. Revenue growth is (1.50 − 1.20) / 1.20 = 0.25, or 25%. Dividing by FY2025 revenue produces 20%, which answers a different comparison. Writing the baseline beside the formula is a cheap way to catch that denominator error.

The second statement is Cannot Say. FY2023 revenue is missing. You cannot infer it from FY2024, FY2025, or a later forecast. Growth could exceed 10%, equal 10%, or be negative, depending on the unavailable baseline.

The third statement is True. The FY2024 on-time rate is 800 / 1,000 = 80%. The FY2025 rate is 1,700 / 2,000 = 85%. The difference is five percentage points. The larger number of orders does not itself establish the rate change; both denominators matter.

The fourth statement is False. Relative growth in the rate is (85 − 80) / 80 = 6.25%, not 5%. The supplied values contradict the claim. This is not Cannot Say, because the required information is present and sufficient.

The fifth statement is True. Fixed costs rose by (250 − 200) / 200 = 25%. Both figures use the same unit, so multiplying each by 1,000 would cancel out of the ratio. Recognizing that saves keystrokes without sacrificing accuracy.

The sixth statement is Cannot Say. A fixed-cost increase is not a profit calculation. Variable costs and revenue are not provided on this sheet. Do not assume variable costs stayed constant, profit margins were unchanged, or missing entries were zero. A confident business story is not numerical evidence.

The answer sequence is therefore True, Cannot Say, True, False, True, Cannot Say. The point is not to memorize it. The point is to attach each answer to a reproducible lookup and calculation.

Prove Cannot Say with two possible worlds

“I don't know” is weaker than “the data permits opposite answers.” For Cedar's missing FY2023 revenue, construct two values consistent with every supplied cell. If FY2023 revenue was $1.00 million, growth to $1.20 million was 20%, so “exceeded 10%” is true. If FY2023 revenue was $1.20 million, growth was 0%, so it is false.

Because the sheet does not rule out either case, Cannot Say is justified. You do not need to invent a likely historical value. In a timed task, noticing the missing baseline is enough; the two-world proof is a training tool for checking that your reasoning is sound.

Do not overuse this method to manufacture ambiguity where none exists. A statement saying FY2025 revenue exceeded FY2024 revenue is directly supported by 1.50 versus 1.20. A statement saying it fell is directly contradicted. Uncertainty about the broader business does not make a narrow, fully specified comparison unknowable.

Also distinguish logical missingness from your own lookup failure. If a value exists on the relevant sheet but you have not found it, the answer is not automatically Cannot Say. During practice, label that mistake “navigation,” not “insufficient information.” Otherwise, you train yourself to stop searching too early.

Handle forecasts, rounding, and category boundaries

Cedar's FY2026 forecast of $1.80 million does not establish that FY2026 revenue actually reached that level. “The forecast is 20% above FY2025 actual revenue” is supported by the provided figures. “Actual revenue rose by 20% in FY2026” is not. The arithmetic is identical; the evidence status is different.

Rounded displays need another check. Suppose two rates are displayed as 7.2% and 7.3%, each rounded to one decimal place. You can describe the displayed difference as 0.1 percentage points. You cannot generally recover the exact underlying difference from those rounded labels. Whether a statement is answerable depends on its wording and the rounding convention supplied.

For totals, inspect category coverage. “All customers” cannot be calculated from a table containing only premium customers unless that relationship is specified. Likewise, adding a regional subtotal to a national total double-counts its contribution. A visible number is not necessarily eligible for your denominator.

Keep conditional language intact. “Among eligible orders” excludes ineligible orders from the rate denominator. “At least,” “more than,” and “exactly” also matter at boundaries. If a calculated value is 25%, a statement saying “more than 25%” is false, even though “at least 25%” is true.

Practice timing by measuring your actual bottleneck

The official numerical module description's 12 minutes and 37 tasks imply roughly 19.5 seconds per listed task if time were spread evenly. That is arithmetic about the published format, not an instruction to spend exactly 19.5 seconds on every question or an assertion that every candidate must finish all tasks.

First practice untimed until the lookup routine is accurate. Then use short timed sets with unfamiliar tables. Record time spent locating the sheet separately from time spent calculating. If most delay is navigation, doing more percentage drills will not solve the main problem.

Use a compact error ledger: statement, selected sheet, required cells, formula, response, and error category. Categories can be wrong sheet, wrong unit, wrong denominator, unsupported assumption, boundary wording, and arithmetic. Retest the category with a new dataset rather than rereading the same answer.

Follow the live instructions about calculators, permitted materials, navigation, and unanswered items. Do not assume that a practice PDF authorizes a tool in every scored session. Prepare your device and read the introductory examples before the timer begins; avoid experimenting with the interface during a scored question.

After a timed set, explain one correct answer without using the calculator. Name the sheet, the baseline, and the evidence supporting the response. This catches lucky answers that would otherwise appear as progress. Then change one condition: remove the baseline, change actual to forecast, or replace “percent” with “percentage points.” Your answer should change only when the new wording changes what is supported. A practice score alone cannot show whether you learned the distinction or remembered the old response sequence. Keep both the original and changed statement in the error ledger so the next session tests the reasoning rather than familiarity.

Turn the skill into broader interview practice

Question	Transferable practice
Calculate Annual Profit of Credit Card Portfolio	Separate revenue, costs, and population assumptions.
Calculate Profitability with Different Pricing Schemes	Track units and compare like-for-like quantities.
Calculate Break-Even Point and Profit Impact Analysis	Explain the denominator and threshold boundary.
Analyze Profit Decline: Data Collection and Hypothesis Testing	Identify what additional data a conclusion requires.
Evaluate a verbal reasoning passage	Distinguish contradiction from insufficient information.

Choose a calculation exercise in Statistics and Math, then deliberately remove one required value. Explain which conclusions still hold and which become Cannot Say. That is more diagnostic than repeating a familiar formula faster.

Sources and Further Reading

• Aon: prepare for your online assessment
• Aon: numerical reasoning practice tasks
• Aon: applied numeracy practice tasks

Official materials checked October 4, 2026. Cedar's data, answers, and error-diagnosis exercises are original illustrations; your invitation controls the actual assessment rules.