Where do these Bain practice questions come from?

They are original items written in the style of both assessments, not real Bain, SOVA or TestGorilla content, built to mirror the question types and distractor logic candidates report.

Bain Assessments / Numerical Reasoning

The one section in both Bain tests

Bain Numerical Reasoning Practice Questions: SOVA and TestGorilla

Ten worked questions in the real formats, with every wrong option explained. Numerical reasoning appears in the Bain SOVA test and in the Bain TestGorilla assessment, which makes it the highest return thing you can practice.

Updated September 2026 14 min read Original questions in the style of both tests

10Worked practice questions

~50 secRealistic pace per item

6 typesCover almost everything asked

Both testsSOVA and TestGorilla

Overview

Bain SOVA or Bain TestGorilla? The math is the same

Bain runs two assessment platforms and assigns them by office and role. The Bain SOVA test is more common across EMEA and the Middle East. The Bain TestGorilla assessment is reported mainly in US offices and some European ones. Your invitation email names which one you are sitting.

The good news for preparation is that numerical reasoning is the one section that appears in every reported configuration of both. The question types, the distractor design and the arithmetic are close enough that practicing for one prepares you for the other. What differs is the pressure model.

	Bain SOVA	Bain TestGorilla
Clock	No countdown on screen, but time per item is recorded and feeds the score	Hard countdown per block, visible, and it does not pause
Length	Around 15 items in the numerical section	Roughly 12 items in about 10 minutes
Going back	No return once a section is submitted	No return once an item is submitted
Scoring	Percentile ranked against other applicants	Accuracy weighted, read alongside your CV
Calculator	Generally permitted	Reports differ, so prepare without one

Practice at pace either way: roughly 45 to 50 seconds per item is the working target for both. On TestGorilla the clock enforces it. On SOVA nothing stops you going slower, which is exactly why candidates do, and why their scores suffer for it.

Question types

The six types that cover almost everything

Across both Bain assessments the numerical items cluster into six shapes. Recognizing which one you are looking at within the first few seconds is most of the speed advantage.

Type	What it looks like	The usual trap
Percentage change	Two figures, one growth or decline rate to compute	Dividing by the new figure instead of the old one
Share of total	A table of categories, one as a proportion of the whole	Using a subtotal as the denominator
Margins	Revenue and a cost line, gross or operating margin out	Reporting the cost rather than what is left
Compounding	A rate applied across two or more periods	Adding the rates instead of multiplying them
Weighted average	Two or more groups of different sizes	Taking a simple average and ignoring the weights
Reverse percentage	A post-change figure, the original asked for	Applying the rate forward instead of backward

The ten questions below cover all six, in roughly increasing difficulty. Give yourself 50 seconds each before revealing the answer, and do them on paper rather than with a calculator.

Practice

Questions 1 to 3: the basic shapes

Straightforward arithmetic, built with the same distractor logic the real Bain items use. If any of these take more than 50 seconds, that is the signal to drill this type rather than move on.

Question 1 · Percentage changetarget 40 sec

Data

A packaging business reported revenue of $240 million in 2024 and $288 million in 2025.

By what percentage did revenue grow?

Correct: B. The change is 48. Divide by the starting figure: 48 over 240 is 20 percent.

A is the same change divided by the new figure, 48 over 288, which is the single most common error in this section. C states the absolute change as if it were a percentage. D gives 2025 as a percentage of 2024 rather than the growth on top of it. Anchor on the base year before you divide anything.

Question 2 · Share of totaltarget 45 sec

Exhibit

Units sold by channel, thousands

Channel	Units
Retail stores	420
Online	310
Wholesale	190
Export	80

Online accounted for what share of total units?

Correct: C. Total is 420 plus 310 plus 190 plus 80, which is 1,000. Online is 310, so 31 percent. The round total is deliberate: check whether the numbers sum to something clean before reaching for a calculator.

A divides online by the first three channels only, a denominator error you make by reading the table too fast. B is retail's share, the right method on the wrong row. D is online as a share of the two largest channels combined.

Question 3 · Gross margintarget 45 sec

Data

A beverage company reported revenue of $6.4 million last year. Cost of goods sold was $4.16 million.

What was the gross margin?

Correct: A. Cost of goods is 4.16 over 6.4, which is 65 percent, so 35 percent of revenue is left. Margin is always what remains, not what was spent.

B is the cost ratio reported as the margin, the classic inversion. C is gross profit in dollars, which is correct arithmetic answering a question that was not asked. D divides gross profit by cost of goods rather than by revenue. Read whether the question wants a percentage or a dollar figure before you start.

Practice

Questions 4 to 7: the ones that separate candidates

Two steps rather than one, which is where the 50 second budget starts to bite.

Question 4 · Compoundingtarget 50 sec

Data

A regional carrier moved 850,000 passengers in 2023. Volume grew 10 percent in 2024 and a further 10 percent in 2025.

Approximately how many passengers did it move in 2025?

Correct: D. 850,000 times 1.1 is 935,000. Times 1.1 again is 1,028,500. Two 10 percent increases compound to 21 percent, not 20.

A adds the rates and applies 20 percent in one go, which is the trap the question is built around. B stops after one year. C applies 30 percent, as though the rate ran for three years. Whenever a rate repeats across periods, multiply rather than add, and expect the correct answer to sit slightly above the naive one.

Question 5 · Weighted averagetarget 50 sec

Exhibit

Average selling price by segment

Segment	Units	Price
Premium	200	$90
Standard	800	$40

What is the average selling price across all units?

Correct: B. Revenue is 200 times 90, which is 18,000, plus 800 times 40, which is 32,000. Total 50,000 over 1,000 units is $50.

A is the simple average of 90 and 40, ignoring that standard outsells premium four to one. C and D are plausible near misses that reward a guess but not a method. The sanity check that saves you here: the answer must sit closer to the larger segment's price, so anything near the midpoint is wrong by construction.

Question 6 · Reverse percentagetarget 45 sec

Data

After a 25 percent discount, a product sells for $60.

What was the price before the discount?

Correct: C. $60 represents 75 percent of the original. Divide: 60 over 0.75 is 80. Check it forward, since 25 percent of 80 is 20, and 80 minus 20 is 60.

A adds 25 percent to 60, which is the reflex answer and is wrong because the discount was taken off a larger base. D adds 20 percent. Reverse percentage items are worth recognizing on sight, because the reflex answer is always among the options.

Question 7 · Two step sharetarget 50 sec

Data

A market is worth $2.4 billion. The client holds a 15 percent share. Management expects the market to grow 10 percent next year while the client's share falls to 12 percent.

What will the client's revenue be next year?

Correct: A. The market grows to 2.64 billion. Twelve percent of that is 316.8 million.

B applies the new share to the old market, forgetting the growth. C is this year's revenue at 15 percent of 2.4 billion. D grows the market but keeps the old share. Every distractor here is one of the two steps done correctly and the other skipped, which is exactly how the real items are built: the trap is never bad arithmetic, it is a missing step.

Practice

Questions 8 to 10: the back end of the section

Items typically get harder as the section runs, and these are the shape of what arrives once you are already short on time.

Question 8 · Break eventarget 55 sec

Data

A subscription business has fixed costs of $1.8 million a year. Each subscription sells for $240 and carries $60 of variable cost.

How many subscriptions are needed to break even?

Correct: B. Contribution per subscription is 240 minus 60, which is 180. Fixed costs of 1,800,000 divided by 180 is 10,000.

A divides by the full price of 240 and ignores variable cost. C divides by the variable cost. D is a near miss that comes from rounding the contribution to 150. Break even items are always fixed cost over contribution, and the only judgment involved is spotting which figure is the contribution.

Question 9 · Index readingtarget 55 sec

Exhibit

Output index, 2022 equals 100

Year	Plant A	Plant B
2023	108	96
2024	115	108
2025	121	126

Which statement is supported by the exhibit?

Correct: D. Against a 2022 base of 100, Plant A reaches 121 and Plant B reaches 126, so B grew more over the period.

A is the trap worth internalizing: an index measures growth from each plant's own base, not volume, so it can never tell you which plant produced more. B misreads a rise from 115 to 121 as a fall. C ignores Plant B dropping to 96 in 2023. Whenever a table is indexed, the only comparisons available are growth comparisons.

Question 10 · Multi step with a distractortarget 60 sec

Data

A retailer operates 240 stores averaging $1.8 million of revenue each. It plans to close 15 percent of stores and expects the remaining stores to pick up one third of the closed stores' revenue. Head office costs of $40 million are unaffected.

What will total revenue be after the closures?

Correct: C. Total revenue today is 240 times 1.8, which is 432 million. Closing 15 percent removes 64.8 million, leaving 367.2 million. One third of the closed revenue, 21.6 million, transfers across, giving 388.8 million.

A stops before the transfer. B is the starting figure, which sits there for anyone who runs out of time and grabs the number they already calculated. D subtracts the head office costs, which the question tells you are unaffected and which are about profit rather than revenue. The unused figure is the tell: if a question hands you a number you never touched, check whether you missed a step or whether it was placed there to be ignored.

From Case Prep · Bain Assessment Simulators

Ten questions on a page is not the same as ten under a clock

Reading an explanation is the easy part. The Bain SOVA full mock and the Bain TestGorilla sprint put the same question types behind a timer, with no way back and no feedback until you finish. The first section of each is free.

numerical-block 06:12

Items 1 to 4Done

Items 5 to 8Live

Items 9 to 12Ahead

7 / 12Answered

48 secAverage per item

FreeFirst section

Q7. A market is worth $2.4bn and the client holds 15 percent. The market grows 10 percent while the client's share falls to 12 percent. What is the client's revenue next year?

A$316.8 million

B$288.0 million

C$360.0 million

One pass only No going back ~50 seconds per item

83%Accuracy

Percentages92%

Compounding67%

Multi step75%

Accuracy by question type Every item explained

Sit the real blockTwelve items against a countdown, or the hidden clock if you are prepping SOVA.

Find your slow typeAccuracy broken out by question type, so you know which of the six is costing you.

Fix and repeatTen sittings, rotating question bank, full reasoning on every item.

Open the Bain simulators Free math drills →

Method

Estimating fast enough for 50 seconds

Fifty seconds includes reading the stem, reading four options, calculating and clicking. Very little of it is available for arithmetic, so the candidates who finish are not faster at math. They just do less of it.

Read the options before you calculate

If the options are 15, 35, 65 and 96, you need one significant figure, not three. Computing 34.7 percent when 35 is the only option in range spends 30 seconds to gain nothing. Option spacing tells you how precise to be, and checking it takes two seconds.

Round to something you can divide

Move 62 over 310 to 62 over 300 and you get roughly 20.7, close enough to identify 20 percent. Round both numbers in the same direction and the error mostly cancels. Where options are tight, round once and then adjust.

Use the anchors you already know

Ten percent, then halve it for five, then halve again for 2.5. A third is 33 percent, a quarter is 25, an eighth is 12.5. Most Bain numerical items resolve into combinations of these rather than into long division.

Sanity check the direction

Before selecting, ask whether the answer should be larger or smaller than the figure you started with, and roughly by how much. A weighted average must sit nearer the bigger group. Two compounded increases must exceed their sum. That check catches most wrong answers in under three seconds.

The rule that rescues the most marks: cap yourself at 45 seconds. If you are not on a path to an answer, take the best option and move. On the Bain TestGorilla assessment an unreached item scores as wrong, and on the Bain SOVA test a slow crawl costs you on the recorded time component. Either way, finishing beats perfecting.

Pitfalls

The five traps that recur in both Bain tests

Dividing by the new figure
Percentage change is always measured against the starting value. This single error accounts for more lost marks in Bain numerical reasoning than any other, and the wrong answer it produces is always among the options.
Answering the calculation the data suggests
The exhibit often supports several plausible sums. Read the question before the table, and confirm which row, which year and which unit it names.
Adding rates that should compound
Two 10 percent increases make 21 percent, not 20. Any time a rate repeats across periods, multiply.
Treating an index as a volume
An index shows growth from each series' own base. It can never tell you which series is larger, only which grew more.
Using every number provided
Bain items routinely include a figure that is there to be left alone. An unused number is not evidence you missed a step, though it is worth one second of checking.

Preparation

A one week drill plan

Numerical reasoning is the section where a week of work moves your result most, because the skill is mechanical and the gains are immediate.

Day 1. Diagnose
Work the ten questions above with a stopwatch. Record the time on each. Anything over 60 seconds names the question type you are weakest on.
Days 2 to 4. Drill the weak type
Twenty minutes daily on that single type until it is automatic, then a short mixed set to keep the others warm. The math drills run these formats against a clock.
Day 5. Estimation only
Work a set where you are forbidden from calculating exactly. Force yourself to round and check option spacing. This is the habit that buys the seconds.
Day 6. Full timed block
Twelve items in ten minutes, no pausing, no going back. Whatever your accuracy, note how many you reached.
Day 7. Review and stop
Label every miss as misread, ran out of time, or did not know. Those three need different fixes, and only the third needs more math.

Format specifics are in the Bain SOVA guide and the Bain TestGorilla guide. If you are still waiting on your invitation, the Bain deadlines page has the timing for each role.

FAQ

Bain numerical reasoning: frequently asked questions

How hard is the Bain numerical reasoning test?

The arithmetic is straightforward: percentages, ratios, margins, growth rates and weighted averages. The difficulty comes from the pace and from distractor options built around common misreads rather than from the math itself.

How many numerical questions are in the Bain assessment?

The Bain SOVA test reports around 15 items in the numerical section. The Bain TestGorilla assessment reports roughly 12 items in about 10 minutes. Both work out near 45 to 50 seconds per question.

Can I use a calculator on Bain numerical reasoning?

A calculator is generally permitted on SOVA. Reports differ for TestGorilla, and some candidates describe having none. Prepare with paper and mental math so a calculator is a bonus rather than the basis of your pacing.

Is the numerical section the same in SOVA and TestGorilla?

The question types and distractor design are close enough that practice transfers between them. What differs is the pressure model: TestGorilla shows a hard countdown per block, while SOVA shows no clock but records your time per item.

What is a good score on Bain numerical reasoning?

Neither platform reports a section score to you. On SOVA the cognitive sections are percentile ranked, with the 70th to 80th percentile commonly described as the practical range. The target you can act on is finishing every item with your accuracy intact.

Where do these practice questions come from?

They are original items written in the style of both assessments, not real Bain, SOVA or TestGorilla content. They are built to mirror the question types and the distractor logic candidates report, so that working them teaches the same habits.

Keep going

The rest of the Bain process

Bain SOVA Test Guide All five sections, the hidden clock, and how percentile scoring works. Read the guide → Bain TestGorilla Guide Four hard timed blocks, both reported versions, and the pacing math. Read the guide → What score do you need? Why the published pass rates disagree, and the percentile worth targeting. Read the guide → Consulting math drills The same formats against a clock, free with any account. Start drilling → Bain application deadlines Every window for AC, ACI, MBA and advanced degree candidates. Check the dates → Bain salary data Base, bonus and total comp by level and region. See the numbers →

Bain SOVA or Bain TestGorilla? The math is the same

The six types that cover almost everything

Questions 1 to 3: the basic shapes

Questions 4 to 7: the ones that separate candidates

Questions 8 to 10: the back end of the section

Estimating fast enough for 50 seconds

Read the options before you calculate

Round to something you can divide

Use the anchors you already know

Sanity check the direction

The five traps that recur in both Bain tests

A one week drill plan

Bain numerical reasoning: frequently asked questions

The rest of the Bain process

Ten questions down. Now do them against a clock.