Bain online test (TestGorilla): format, scoring and how to prepare
Which test Bain sends depending on the office, the Latin America and 4-section formats, how it’s scored, how to solve each question type and a one-week prep plan.
Practice the Bain test for freeBefore the case interviews, Bain screens candidates with a short online test. It takes about 30 to 40 minutes, you answer against the clock and, at many offices, it runs on TestGorilla. This guide explains which test you may get, what each section looks like, how it’s scored and how to prepare.
This guide combines what Bain publishes on its careers site with what prep guides and candidates describe. Bain doesn’t publish the questions or the cutoff score, and the format changes by office and year. The instructions that count are the ones in Bain’s email: if something here doesn’t match, follow the email.
What it is and where it fits in the process
Bain calls it the digital assessment. It comes after you apply and before your first interview. What Bain says about it on its official page:
- After applying, you get an email telling you whether you’ll take Sova or TestGorilla. In some markets, the link arrives about a week after you apply.
- It takes about 30 to 40 minutes and measures problem solving, communication and analytical skills.
- You can use a calculator, pen and paper. The calculator is only for the test, not for the interviews.
- AI tools are prohibited, as are screen recording, screenshots and software that transcribes or gives feedback.
- TestGorilla is done in one sitting. Sova can be split up, as long as you finish before the deadline.
- If you don’t complete it by the deadline in the email, your application isn’t considered. Hearing back can take up to 60 days.
The test is one part of the evaluation, together with the rest of your application. A strong result won’t save a weak résumé, but a poor one can stop a strong résumé before it reaches an interviewer.
Which test you get, by office
Bain doesn’t use the same test everywhere. This is what’s known as of late 2026:
| Offices | Test | Source |
|---|---|---|
| Santiago, São Paulo, Bogotá, Buenos Aires | TestGorilla | Bain’s application pages |
| Middle East, Greater China and Southeast Asia | TestGorilla | Bain’s application pages |
| Most US offices and Amsterdam | TestGorilla | Candidate reports |
| Most of Europe, including London in the 2026-27 cycle | Sova | Candidate reports and prep guides |
Older guides mention Pymetrics (cognitive games) and HireVue (video). Bain used them at some offices, including Santiago in 2022, but its 2026 page no longer names them. Your invitation email is what counts.
The Latin America format
For its process in Chile and Brazil, Bain published a “Gorilla” test followed by two rounds of two case interviews. In Chile the test was in Spanish. The modules it published for Brazil were:
| Section | Questions | Time |
|---|---|---|
| Critical thinking | 16 | 12 min |
| Numerical reasoning | 12 | 10 min |
| Attention to detail (text) | 16 | 12 min |
That’s 44 questions in 34 minutes: about 45 seconds per question in critical thinking and attention to detail, and 50 in numerical reasoning. The Portuguese version added an English module (20 questions in 12 minutes). For Chile, Bain didn’t list the modules in that announcement, so they are most likely the same, but check your email.
The other version: 4 sections
At US and other offices, candidates report a different version, with 51 questions in about 39 minutes:
- Numerical Reasoning: 12 questions in 10 minutes, with tables and short problems.
- Business Judgment: 15 questions in 10 minutes. Each one describes a business situation and gives two options; you pick the one a good consultant would take first.
- Leadership & People Management: 12 questions in 10 minutes on how to lead someone given their skill and confidence level.
- Problem Solving: 12 questions in 9 minutes with rules, schedules, priorities and rates.
If your email describes this version, the numerical part of this guide still applies. For the other three, the simulator has the full 4-section format, and here’s how to approach each one.
Business judgment: two options
Both options usually sound reasonable. What sets the better one apart is almost always one of these:
- Understand first, then act. Locating the problem or confirming an odd data point comes before proposing a solution.
- Speak up early and honestly. A mistake, a delay or data that contradicts the hypothesis goes to your manager as soon as you know, with its impact.
- Hypotheses and priorities. Start from a hypothesis to focus the work, and do first what matters most for the recommendation.
- No ethical shortcuts. Improperly obtained confidential information isn’t used, and a relevant cost isn’t hidden.
You have 40 seconds per question: read the situation once, ask yourself what a good consultant would do first, and choose.
Leadership: the style depends on the person and the task
The questions are based on situational leadership: the right style depends on how well the person can do that particular task and on their motivation or confidence.
| The person… | Style | What you do |
|---|---|---|
| Can’t do the task yet but is eager | Directing | Give clear instructions and follow up closely |
| Can do part of it, still makes mistakes and has lost heart | Coaching | Keep directing, but explain why and encourage them |
| Can do it but lacks confidence or motivation | Supporting | Listen, decide together and back them up |
| Can do it and is committed | Delegating | Hand over the task and agree on the outcome |
Watch for two traps. Someone who is excellent in one field still needs directing if the task is new to them. And someone who knows the job but has lost confidence won’t be helped by step-by-step instructions. Some versions of the model describe the levels a little differently, so focus on two questions: can they do it? do they want to, or feel able to?
Problem solving: apply the rules to the letter
These are short problems: calendars to cross-check, prices with conditions, tasks that must fit a time window, work rates and time zones. Almost every mistake comes from skimming a rule: “more than 5 years” doesn’t include someone with exactly 5, “or part of” is charged as a full unit, and a booking that ends at 9:00 leaves the room free at 9:00. Write the rules down before calculating. In work-rate problems, remember that two people together are always faster than the faster one alone.
Rules and scoring
These rules come from TestGorilla and from what candidates report:
- One question at a time, no going back. If you move on without answering, the platform warns you, but once you continue, that question is closed.
- A timer for each section. You can rest between sections; within a section, the clock doesn’t stop.
- Questions come from a bank. Two candidates don’t see exactly the same ones.
- Anything you haven’t submitted when time runs out is lost.
- Wrong answers cost nothing. A wrong answer is worth the same as a blank: zero.
The score the employer receives is a percentile: your result compared with other people who took the same test. Harder questions weigh more. Bain doesn’t publish the cutoff, so the practical goal is to get as many right as possible and leave none blank.
Critical thinking
This is the most varied section. Four question types show up.
Syllogisms and “true, false or can’t tell?” You get two or three statements and a conclusion. The golden rule is to use only what the text says, even if the conclusion sounds reasonable in real life.
All certified suppliers passed an audit. Some suppliers from Valparaíso passed an audit. Conclusion: some suppliers from Valparaíso are certified.
The answer is can’t tell: the Valparaíso suppliers who passed the audit might not be certified. The most common mistake is marking “true” because the conclusion seems logical. Another classic trap: “if it rains, the match is called off” doesn’t say rain is the only reason to call it off.
Number series. First look at the differences between consecutive numbers. If they aren’t constant, try the differences of the differences, multiplication, two interleaved series (odd and even positions) or adding the two previous numbers.
Ordering puzzles. Five people, several clues (“A arrived before B”, “C arrived right after D”) and a question about one position. Draw five slots and start with the clues that fix a position or chain two people together. You don’t need the full order if you already know who is in the position being asked about.
Arguments. You get a short line of reasoning and are asked what assumption it relies on, what strengthens it or what weakens it. Three ideas solve almost all of them:
- The assumption is what has to be true for the conclusion to hold. Try negating it: if the conclusion collapses, that’s the assumption.
- What weakens most is usually another cause that explains the same result (another change that happened at the same time, a pilot that doesn’t represent the rest).
- A correlation doesn’t prove a cause. If two things go together, look for a third factor that explains both.
Numerical reasoning
These are tables and short business problems: growth, market share, margins, break-even, exchange rates, averages. The math is simple; what’s hard is setting it up correctly with 50 seconds per question. The wrong options are often exactly the typical mistakes, so even if your first calculation matches an option, check the base.
An example:
A product’s price rose 20% in March and fell 20% in June. It cost US$1,000. What does it cost now?
It doesn’t go back to US$1,000. The drop applies to the already higher price: 1,000 × 1.2 × 0.8 = US$960.
The most frequent mistakes:
- Dividing by the wrong base. Growth is calculated on the starting value, and “A is X% larger than B” is calculated on B.
- Mixing up percentage points and percent. If the margin goes from 20% to 25%, it rose 5 points, not 5%: in relative terms it rose 25%.
- Averaging percentages or prices without weighting. If many more units of one product were sold, its price weighs more in the average.
- Leaving out a row. In market share, the total includes “other companies”.
- Forgetting the variable cost. Break-even is fixed cost ÷ (price − variable cost), not fixed cost ÷ price.
Use the calculator for long sums, but estimate first: if you expect “a bit over 30%” and the calculator shows 300%, something went wrong.
Attention to detail
Comparing lists, codes and addresses, checking a form against an email, counting invoices that meet a condition. It looks easy, and that’s why people lose points by rushing. What works:
- Compare in short chunks of three or four characters, not the whole line at once. Typical changes are two swapped digits (4839 → 4389), a similar-looking letter (n for m, l for i) or a missing accent.
- Check field by field, not row by row, when comparing two tables: first all the names, then all the codes, then all the amounts.
- Read the condition twice before counting. “More than US$5,000” doesn’t include an invoice of exactly US$5,000, and “Comercial Andes” isn’t the same as “Comercial Andina”.
- Tally on paper as you count, with one mark per match. Recounting from scratch costs time.
Managing your time
You have less than a minute per question. Three rules:
- Don’t get stuck. If a question takes twice the pace, pick your best option and move on. The others are worth the same.
- Never leave one blank. Since wrong answers cost nothing, a guess beats a blank. And since you can’t go back, the decision is made on the spot.
- Check the clock halfway. In critical thinking you should reach question 8 with about 6 minutes used; if you’re behind, speed up on the question types that come fastest to you.
One-week prep plan
- Day 1: take the full test in the simulator in test mode, in the format you’ll get, with no prep, to see where you start.
- Days 2 and 3: practice your weakest section in learning mode. Read the explanation of each mistake until you understand why the right option is right.
- Days 4 and 5: practice the other sections of your format, one per day, and review the typical mistakes in this guide.
- Day 6: another full test in test mode. Compare your section scores with day 1.
- Day 7: rest or do just one section. Set up where you’ll take the test: computer, Chrome, calculator, pen and paper, and a place without interruptions.
After the test
If you pass, the interviews come next. For Latin America, Bain describes two rounds of two interviews each. The interviews combine a case with questions about your experience. There’s no calculator in the case, so it helps to train with the mental math trainer. To practice solving a case step by step, also see the case guide: it’s McKinsey’s, but the math, structure and synthesis apply just the same.
CasePracticeFree is an independent project, not affiliated with Bain & Company or TestGorilla. The examples in this guide and the simulator questions are our own.