McKinsey Solve:
the complete preparation guide.

Three ways to use this guide

Choose the sections that match your stage of preparation.

What McKinsey Solve is and where it sits in the process

McKinsey Solve is an online assessment built as a set of short games set in nature: islands, oceans, research labs. You do not need to know anything about business or about video games. What the games observe is how you handle information: which data you look for, how you combine conditions, how you decide when the picture is incomplete and the clock is running.

Most candidates take it after the application screen and before the first interviews. If the result is positive, the next step is usually the case interview and, at McKinsey, the Personal Experience Interview (PEI). Some candidates, depending on office and role, may follow a different sequence, and your recruiter is the right person to confirm yours.

What McKinsey states officially

McKinsey publishes a short FAQ about the assessment. The September 2026 FAQ confirms the practical rules below.

The FAQ also says that no preparation is needed. That is true in the narrow sense that the games teach you their rules in the tutorials. In practice, candidates who have already seen the interface and felt the clock can spend their attention on the problem rather than on the controls, which is the reason practice helps. Familiarity is the point, not memorising answers, because the scenarios change for every candidate.

What is not official

The game names, individual timings and format details below come from candidate reports and public preparation sources. Your invitation and the instructions in the assessment take precedence. McKinsey does not publish its scoring model or a pass mark.

The three games at a glance

The table summarises the games as candidates currently report them. The times are not published by McKinsey, so use your invitation as the reference for your own assessment.

Three different things are easy to confuse here, and it helps to keep them apart while you prepare:

• The real test is what McKinsey runs. Only McKinsey knows its exact content, and it can change it at any time.
• Your invitation is the only official signal of what you will get, because it states the length of your assessment.
• Our simulator reproduces the games as candidates report them, so that you can practise the format, the interface logic and the timing. Its scenarios are original, and a result in practice is not a prediction of your official result.

Elena, McKinsey Solve candidate: after practising the three games, nothing in the real assessment surprised her.

Three habits that carry across the games

The games look different, but they reward the same few habits, and it is worth building them before you specialise in any one game.

• Start from the objective. In Redrock, read the question before the data. In Sea Wolf, read the conditions before the microbes. In the Lab, read what the project needs before deciding about people.
• Write things down before you act. The figure and its unit, the list of conditions, the commitment you made yesterday. Paper is quicker than memory under time pressure.
• Decide with a reason you could state. McKinsey says it may ask you at random to explain your logic. A choice you can explain in one sentence is usually a choice made in an orderly way.

None of this is a trick. These habits simply leave less room for the errors that cost most points, and they make the second and third runs of any game much faster than the first.

Redrock Study: data, calculations and a report

Redrock Study usually comes first. According to candidate reports, it runs for about 35 minutes on a single timer and has two parts. In the Study, you read about a research question on a fictional island ecosystem, save the data you need in an on-screen journal, answer a few analysis questions, and complete a short report that includes choosing a chart. In the Cases, you answer six short, independent questions, each with its own small set of data.

Navigation is reported to be one-way: once you move on, you cannot go back to fix an earlier step. That is why the habits you build in practice matter more than the difficulty of the maths, which stays at the level of percentages, ratios, averages and growth rates.

Where candidates lose points

Most lost points in Redrock do not come from hard arithmetic. They come from saving too much data and then not finding the right figure, from using a value in the wrong unit, or from spending so long on the Study that the six Cases are rushed. A short routine helps: read the question first, decide which figure you need, check its unit, and only then calculate.

The full method, two longer worked exercises (including the chart in the report) and a pacing table are in the Redrock Study practice guide.

The Sea Wolf: constraints and combinations

The Sea Wolf, also called Ocean Cleanup, is reported to last about 30 minutes and to cover three ocean sites. At each site you choose microbes for a treatment, and the final team has to satisfy several conditions at the same time. Candidates describe conditions of two kinds: average values of some attributes that must fall within a range, and traits that must be present or must be avoided.

The good news is that every site repeats the same steps. Once you have a reliable way of working through the first site, the second and third go faster. The difficulty is that it is easy to lose time comparing microbes before you know which ones are allowed at all.

Write the conditions down first

Before you look at any microbe, write the conditions of the site on paper and separate them into two groups. The first group holds the hard conditions that a microbe or a team either meets or does not, such as a trait to avoid. The second holds the conditions you can only check on the team as a whole, such as an average. The first group lets you exclude microbes at once, and the second tells you what to compare among the ones that are left.

In the real game the pool is larger and not every site allows a perfect team, so a good result within the time is better than a perfect first site and two rushed ones. McKinsey's ban on applications also applies here: the check above is done by hand, without spreadsheets or solvers. A full site worked through by hand, with two combinations compared, is in the Sea Wolf guide.

Sustainable Futures Lab: judgement with incomplete information

Sustainable Futures Lab focuses on judgement rather than calculation. It asks you to make decisions about a project and the people working on it, with less information than you would like. What it puts at the centre is judgement: what you find out before deciding, how you set priorities, and how you treat people and resources over time.

Candidates currently report two formats:

• A 20-minute format with one ranking task and twelve situational questions.
• A 30-minute format in which you lead a small research team for three simulated days. Each day has the same stages: you gather a limited amount of information, assign researchers to workstations with a reason for each move, respond to requests from the team, and reflect on how each member is doing.

Which format you get is not something you can choose, and the sources do not allow anyone to guarantee it. The length in your invitation is the most useful clue, as the next section explains.

Consistency is not the same as never changing your mind

In both formats, the decisions are connected. If you moved someone with a reason yesterday, or promised a researcher something, that commitment is part of today's picture. At the same time, new information can make a different decision the right one. The difference between a considered change and an erratic one is the reason behind it.

This is an illustration of reasoning, not a scoring rule: McKinsey does not publish how the Lab is assessed. The two formats, the stages of the 30-minute version and a longer worked decision are covered in the Sustainable Futures Lab guide.

The 85 and 95 minute versions explained

McKinsey does not publish the length of the assessment or of each game. It states only that the length is in your invitation email. Everything below comes from candidate reports and preparation providers, which in 2026 describe two three-game versions in circulation.

The important point is in the first two columns. Redrock and Sea Wolf are reported to be the same in both versions, so the ten minutes of difference are entirely in the Lab. The 30-minute Lab started appearing in candidate reports from late August 2026, from the United States, the United Kingdom and other regions, while the 85-minute version is still reported by several sources after that date.

Whatever your version, each game is reported to have its own timer, and the tutorials before each game are untimed. Use the tutorial time to understand the objective and the controls calmly, rather than rushing to start the clock.

The official rules, and what they mean on test day

The rules in McKinsey's FAQ are short, and candidates often skip them. They are worth reading, because breaking them can cost much more than a low score.

1. Use the untimed tutorials. Each game starts with one. Read the objective, try the controls, and only then start the timed part.
2. Use no outside help. Applications, websites, AI tools and pre-prepared notes are not allowed, and the FAQ says violations may exclude you from consideration. That includes spreadsheets and solvers for Sea Wolf, whatever some guides suggest.
3. Work alone, and do not record. Recording or taking screenshots of the assessment is prohibited.
4. Report technical problems while they happen. Contact McKinsey's support through the channels in the FAQ or your invitation. Support may add time or reset your link, but only if you contacted it at the time.
5. Ask for accommodations through your recruiter. If you have a disability, a health condition or a specific learning difficulty, speak to your recruiter before you start. Extra time is one example of an accommodation.

Candidates widely report working on blank paper during the games, since the ban covers notes prepared in advance. If you are unsure about anything specific to your case, ask your recruiter before you begin.

PSG, Imbellus, Problem Solving Game: one test, several names

If you search for this assessment you will meet several names. They refer to the same thing, or to earlier versions of it, so they do not need separate preparation.

• Problem Solving Game, often shortened to PSG, is the name used by many candidates and providers for the game-based format.
• Imbellus is the game-assessment company that developed the early versions, so older articles call it "the Imbellus test".
• Digital assessment or online assessment is the generic term used in some invitations and forums.
• McKinsey Solve is the current name, and the one used in this guide.

Before the games, McKinsey used a paper-based Problem Solving Test (PST) for many roles. It is no longer the format candidates describe.

How McKinsey Solve is scored: what is known and what is not

McKinsey does not publish how McKinsey Solve is scored, and it does not publish a pass mark. Any precise threshold, percentile target or pass rate you read online is a guess, and this guide does not repeat them.

What is widely reported by candidates and preparation providers is that the assessment looks at two things:

• The product: the result you reach, such as the correct answers in Redrock or the conditions your Sea Wolf team meets.
• The process: how you get there, because the games record your actions while you play.

Nobody outside McKinsey knows how the two are weighted, and claims that one matters much more than the other contradict each other. The practical consequence is simple: work in an orderly way and aim for good answers, without trying to game either side.

The September 2026 FAQ confirms that your results remain valid for one year and that you may take McKinsey Solve once every 12 months, measured from the date you complete it. Your result is one element of your application. Ask your recruiter how it applies to a future application or another role.

A preparation plan you can adapt

The aim of preparation is not to play as many times as possible. It is to find where you lose time or points, fix that, and check that the fix worked. That gives you a cycle with three steps.

1. Play a first timed run of each game. Do not prepare beforehand beyond reading the rules. The point is to see where the clock gets tight and where you hesitate.
2. Practise the point that cost you most. That might be the unit check in Redrock, the order of conditions in Sea Wolf, or how you use information requests in the Lab. Work on one thing at a time.
3. Play again and compare. Look at the same step that went wrong before. If it improved, move to the next weakness; if not, find out why before playing again.

A few reviewed runs are worth more than many runs in a row. Playing five times without looking at what went wrong mostly trains you to repeat the same mistakes faster.

Pacing: starting points, not rules

Each game has its own clock, and time saved in one game is reported not to carry over to the next. The splits below are starting points reported by other candidates. Use them on your first runs, then adjust them to your own pace.

In every game, an answer is usually better than a blank. If the clock is almost over, make your best choice with what you have and move on.

How to review a run in ten minutes

A review does not need to be long. What matters is that it ends with one thing to change on the next run. After each run, answer three questions in writing.

1. Where did the time go? Note the stage or site where you were slowest, and whether you finished.
2. Where did the points go? For each mistake, name the cause: a wrong input, a missed condition, a decision made before checking the information.
3. What will I do differently next time? One change only, specific enough to check: "write the unit next to every saved figure", "prioritise teams without the unwanted trait, then compare their full attributes".

Keep these notes in one place. After three or four runs they show whether the same mistake keeps coming back, which is the signal to practise that step on its own before playing the full game again.

A calendar for one week, or more

Candidates commonly report having about a week between invitation and deadline, but check your own. The table works for a week and stretches easily if you have longer. If you have more than two weeks, repeat the middle cycle once per game rather than adding more runs at random.

If you have only two or three days, keep the same logic and compress it: one timed run per game, one focused fix on the weakest one, and a final run of all three in order.

Common preparation mistakes

Most mistakes in McKinsey Solve preparation are about method and time management, not about maths. These are the ones to watch for.

• Playing many runs without reviewing them. Without a review, you do not know what changed between one run and the next, so you cannot fix it.
• Practising only the game you like. The game you avoid is usually the one where a few hours make the biggest difference.
• Practising without the clock. The games feel very different under time. Untimed play is useful for learning the rules, but not for the rest of your preparation.
• Saving every figure in Redrock. A full journal is harder to search. Start from the question and save what it needs.
• Comparing Sea Wolf microbes before writing down the conditions. Exclude what is not allowed first, then compare what is left.
• Changing Lab decisions without a reason. Change your mind when new information justifies it, and be able to say why.
• Studying retired games. Ecosystem and Plant Defense are no longer reported, so time spent on them is time lost.
• Using tools the rules prohibit. Spreadsheets, solvers and AI tools are not worth the risk, and a quick check by hand is enough.

How to use feedback between runs

Playing a scenario shows you the format and where you feel pressure. It does not always tell you why you lost points, and the reason decides what you practise next. In Redrock, a wrong answer can come from the wrong input, the wrong formula or the wrong unit. In Sea Wolf, a low score can come from one condition missed at one site. In the Lab, the question is whether your decisions hold together across the scenario.

This is what the detailed reports in the CasePreparation simulator are for. Here is what each one shows, using excerpts from real reports of our practice scenarios.

Redrock: answers, corrections and your journal

The Redrock report puts each of your answers next to the correct one, with the solution written out step by step, and checks your journal: which figures you saved, which you needed, and which you saved but never used. Read it by asking one question: did I lose this point on the input, the formula or the unit? Then practise that, not Redrock in general.

Sea Wolf: time per site and conditions met

The Sea Wolf report shows how your time was spread across the three sites and, for each site, the team you sent with every condition checked against its target. In the excerpt, the averages and the desired trait are all met, but two microbes carry the trait to avoid, so one condition is lost. Two patterns are common in these reports: one site that eats the time of the next, and one type of condition that is missed again and again. Each points to a different fix: a time limit per site in the first case, an earlier exclusion step in the second.

Sustainable Futures Lab: how your decisions read

The Lab report gives you a practice score on five skills used by the simulator (prioritisation, decision-making under uncertainty, interpreting messy information, balancing trade-offs, and team and stakeholder effectiveness), together with written feedback on your decisions: what went well, what to change, and a few principles to carry into the next run. The scores are the simulator's own measure, not McKinsey's. Their use is to show which kind of situation to think through before you play again, for example the questions where you acted before checking the information you had.

Beatriz, McKinsey Solve candidate: the report after every run, and the detail on timing, showed her what to change.

What the free trial includes, and what full access adds

The free trial and full access do different jobs. The free trial is for getting familiar with the three games and seeing where you struggle. Full access is for working on those points with more scenarios and with reports that explain what went wrong.

Full access includes a 100% money-back guarantee if you do not pass McKinsey Solve, subject to the eligibility and refund conditions. These include completing at least five full runs of each game before your test and providing your official result within the required timeframe. Explore the package on the McKinsey Solve practice page.

Amara, McKinsey Solve candidate: she completed the five runs per game the guarantee requires, and the guarantee was the reason she bought.

After McKinsey Solve: the case interview and the PEI

Passing McKinsey Solve opens the door to the interviews, and candidates often report that the first round comes within a few weeks. That is not much time if you start case preparation only after your result. The case interview tests things the games do not: structuring a problem out loud, doing maths while talking to an interviewer, and turning analysis into a recommendation. The Personal Experience Interview asks for real stories from your past, chosen and practised in advance.

The skills overlap only in part. Redrock helps with reading data and simple maths, and the Lab with judgement, but neither replaces practice with a person asking you questions. If your test is still a week or two away, spending a little time on cases now will make the weeks after your result calmer.

The complete guide to case interview preparation covers the case and the fit interview in detail. If you want practice with a former interviewer, Franco also offers one-to-one interview coaching, with a written report after each session.

Frequently asked questions

Where to start

Start by playing each game once under time, before reading more. You will learn more from twenty minutes of play than from another hour of articles, and you will know which of the deep dives to read first: Redrock Study, The Sea Wolf or Sustainable Futures Lab.

About Franco

Franco Borgonovo

Former BCG Principal · 10-year tenure · Official BCG coach

Franco spent ten years at BCG, from intern to Principal, interviewing and selecting candidates as a global lead recruiter. He is still an official BCG coach for new hires, and today works one to one with candidates on applications, McKinsey Solve and interviews.

More about Franco →