McKinsey Solve Game: Newest Updates & Guide (May 2023)

Check out the only, fully-playable McKinsey Solve Test (Problem-Solving Game) Simulation in the entire market, with the new 2023 Redrock Study Task.

With that out of the way, let's continue to learn about the test, shall we?

What is McKinsey Solve (or Problem-Solving Game)?

McKinsey Solve is a gamified, pre-interview screening test 

McKinsey Solve (formerly called Problem-Solving Game, Digital Assessment, or colloquially the "Imbellus Game") is a gamified test designed by Imbellus for the McKinsey & Company. 

In the McKinsey recruitment process, the Solve Game sits between the resume screening and the case interviews, serving the same purpose as the paper-based tests – ruling out the “unfit” candidates to save time and resources during the expensive case interview phase.

Solve has entered trial since 2017 (back then it was known as the Digital Assessment) and has been rolling out extensively in 2020. Since then, Solve had replaced the paper-based Problem Solving Test in every McKinsey office.

The test is mandatory for candidates applying in all practices: General, Operations & Implementation, Research & Analytics, Digital, etc.

Note: As this is a gamified test, in this article, the two terms “game” and “test” will be used interchangeably when referring to the McKinsey Solve.

The new gamified test is supposedly crack-proof

Now, why did McKinsey change the test format from a paper-based test to a game? Keith McNulty, McKinsey’s Global Director of People Analytics and Measurement, put it this way:

“Recruiting only knows if candidates got the right answer, not how they approached the question. Plus, there’s a large amount of strategy, preparation, and luck involved in multiple-choice tests, and if you use them in the selection process, it reinforces the status quo—at a time when you are looking to widen the scope of candidates you’re hiring.”

So essentially, McKinsey is trying to create a test/game that is impossible to game (ironic, isn’t it?).

But in fact it can be broken down into bite-size pieces

With field reports from hundreds of real test takers, we have gathered enough insights to break down the McKinsey Solve into bite-size pieces, which are fairly consistent across candidates. Using those insights, we can derive working overall approaches to the game.

In this article, we will cover:

• Technical details of the test: time limit, number of tasks and mini-games, assessment criteria

• Break-down of each mini-game: description, underlying logic, and recommended strategy

• Test-taking tips to maximize your chances

• Similar games for practicing the McKinsey Solve Game

It is important to keep in mind that since neither Imbellus nor McKinsey publicizes the exact details of the criteria/mechanisms used in-game, the insights in this article – reported by our correspondents – may not reflect 100% of the in-game elements.

What is the McKinsey Solve like?

The McKinsey Solve Test or Digital Assessment has a time limit of 60-80 minutes. The candidate is asked to solve 2 out of 6 possible mini-games. Both the final results and the process are assessed, and if the candidate is found to possess similar skills and tendencies to a McKinsey consultant, they are offered an interview.

For a more detailed guide on the technical details of the game, please check out the McKinsey PSG Simulation (All-in-one) package.

Figure 1: Overview of McKinsey Solve / McKinsey PSG

Time limit is 71 minutes

As of April 2021, the reported time limit for the McKinsey Solve is exactly 70 or 71 minutes, with 35 minutes recommended for the first game (Ecosystem Building), and 35 minutes for the second game (Redrock Study), or 36 minutes (Plant Defense). Time spent on tutorials is not counted towards the limit.

Ever since the start of the game, there have been variations in time limit reports, however, these tend to stay between 60-80 minutes. This variation depends on the length of each mini-game.

Pre-2023, i.e. with Plant Defense mini-game: Actual time allocation depends entirely on the candidate’s decision – however since the first game is much more predictable, we recommend playing this quickly to allow more time for the second game. With a proper approach, the first game should take only 15-20 minutes, with time for a double-check taken into account.

Summer 2023 onward, i.e. with Redrock Study mini-game: The Ecosystem Building mini-game is now allocated a fixed 35-minutes, and the Redrock Study another 35. That means even if you finish the first game early, there is no additional time for the second game.

Candidates should also make the most out of the tutorial time – try to guess the objective of the mini-game, and think of an overall approach before beginning a mini-game. You can also use that time to make necessary preparations, such as pen and paper, or maybe a light snack to keep yourself energized.

Each candidate has to solve 2 out of 6 mini-games 

As of June 2023, 6 mini-games are confirmed for the McKinsey Solve Test: Ecosystem Building, Redrock Study, Plant Defense, Disaster Management, Disease Management, Migration Management. The 2 main mini-games that nearly all candidates will encounter are the Ecosystem Building Game and the Redrock Study Task. 

Our reports indicate that 100% of the McKinsey Solve Test will have Ecosystem Building in the first game slot. For the second game slot, right now, 80-90% of the candidates will have the Redrock Study Task, while 10-20% will have the Plant Defense game (before this, the ratio for the second game was reversed). This means McKinsey is gradually phasing out Plant Defense in favor of the Redrock Study Task.

The first one, Ecosystem Building, is similar to city building games - except with animals instead of buildings - where you have to build an ecosystem with a number of species.

In the Redrock Study Task, your mission is to solve ONE large study using on-screen tools then move on to answering 10 smaller cases with a similar topic.

The other 3 games - Disaster Management, Disease Management, Migration Management - are alternatives that McKinsey previously used for beta testing. They no-longer appear in the McKinsey Solve test in 2023.

• Disaster Management involves identifying the natural disaster occurring in an ecosystem and moving the whole system to another location to minimize damage. This mini-game appeared occasionally from 2020 to 2021.

• Disease Management is about identifying an infectious disease, figuring out its rules of infection, and predicting its spread within an ecosystem. This mini-game appeared occasionally from 2020 to 2021.

• Migration Management is about directing a group of animals from one point to another such that it loses the least amount of resources and animals. This mini-game appeared occasionally from 2021 to 2022.

For the latest insights on the game - Redrock Study, check out the section below or our designed simulation package for this mini-game. 

The next part will be about how candidates are assessed – if that’s not in your interest, you can skip straight to the mini-game and strategy guide using this link.

Every keystroke and mouse movement will be assessed

Each candidate will be assessed using both product scores (i.e. the final results) and process scores (i.e. how they get those results).

Product scores are determined by your level of success in achieving the objectives of the mini-games.

In the first mini-game, while there is no 100% right answer, some solutions will be better than others. You will be given this information through a report screen. For the second mini-game, the final results are definitive fact-based and data-based answers. There will be right and wrong answers, but McKinsey will not inform you how many correct answers/actions you get.

• Mini-game 1: How many species survive? 

• Mini-game 2: Did you pick the right data points? Are your calculations and reports correct? Did you choose a suitable graph to display the data?

Process scores, on the other hand, are dictated using data on your patterns during the whole problem-solving process – every keystroke, every click, and every mouse movement will be assessed.

The process and product scores are combined to form a profile of problem-solving skills and capabilities. And while there is no official statement from McKinsey about which candidates they select, it is likely that the more you resemble a high-performing consultant at McKinsey, the higher your chances will be.

Candidates are assessed on five core dimensions

Your problem-solving profile is drawn using the five following dimensions:

• Critical thinking: the ability to form a rational judgment from a set of facts

• Decision-making: the ability to select the best course of action among options

• Meta-cognition: the ability to use strategies to make learning information and solving problems easier (e.g., testing hypothesis, taking notes)

• Situational awareness: the ability to determine the relationships between different factors and to project the outcomes of a mini-game

• Systems thinking: the ability to understand cause & effect relationships involving several factors and feedback loops (e.g., anticipating several orders of consequences)

The good news is that all the skills assessed are generally not evaluated by themselves, which means training one skill will probably also drive up your assessment scores in others. This is absolutely crucial because you won’t have to go into every nitty-gritty task just to squeeze out some extra score.

Furthermore, while all capabilities must be presented for success, some metrics are considered to be more impactful than others. From this Imbellus research paper, we could deduce that Critical thinking, Situational awareness, and Systems thinking are the fundamental skills that all successful candidates need to possess.

Meanwhile, Decision-Making and Meta-Cognition skills mastery are the advanced skills that will transform candidates from good to great ones.

Figure 2: Median Construct Percentile through McKinsey Recruiting Pipeline (Source: Imbellus)

The test measures telemetry data to calculate the five dimensions

While it is hard to pinpoint exactly the telemetry data gathered since Imbellus does not fully disclose this information, one way of framing this is by each stage of the problem-solving process itself.

Based on our findings from real candidates, we believe the telemetry could be assorted into the following sets, each directly influencing the key activities during the stages from identifying the problem to delivering the next-step recommendation.

Problem Identification: your systematic thinking pattern

• Methodological vs. abstract

• Big-picture thinking vs. detail-oriented

• Example telemetry: prioritization and focus tendency, clicking and decision pattern

Quantitative analysis & data synthesis: the ability to translate data into insights

• Drawing relationship between data

• Filter out correlated or irrelevant information

• Example telemetry: data focus pattern, time spent on quantitative task

Hypothesis-crafting: bringing insights into actionable hypothesis

• Putting emphasis on a certain approach / methodology from insights

• Example telemetry: duration of the transition from analysis to decision-making, disrupted status quo period

Decision-making: coherence in actions and thinking

• Random selection or well-thought out decisions based on analysis

• Decisiveness in carrying out actions with the chosen tactics

• Reaction under growing time pressure – panic clicking vs. calm and focus

• Example telemetry: factors connecting each selection, time spent deciding between options

Next-step recommendation: learning and reflection

• Ability to adjust existing strategy and preference for tried-and-true method in presence of new data set or shifting conditions

• Progressive learning and reflection with failures and successes

• Example telemetry: number of clicks, scrolling speed, time spent on certain data blocks

Breaking down the test – Redrock Study Task

Mini-game overview & description

The Redrock Study Task began appearing as early as July 2023. Then in March 2023, it received an update which divided the Task into 2 Parts which we will see below.

The first part of the mini-game, also the most important one, consists of ONE large study with a main objective and a set of supporting data. This part is divided into 3 main phases: INVESTIGATION, ANALYSIS, AND REPORT.

• Phase 1 - INVESTIGATION: Your task is to skim through the case description, identify the objective and necessary data points, then collect them into an on-screen Research Journal.

• Phase 2 - ANALYSIS: Using a provided calculator, you process the data points to answer 3 quantitative questions. These answers will be used to fill in the report in phase 3. Your calculation history will be recorded.

• Phase 3 - REPORT: With the results calculated from phase 2, your main job is to complete the textual and graphical report (you have to choose which type of graph to use).

In the second part, you have to answer 10 cases with a similar topic to part one (i.g. If your part 1 case is about clothing sales, the mini cases will also be about clothing sales). Though, our user reports show that the topic is purely cosmetic and does not affect the final assessments.

As of July 2023, we have only received reports of Single-select Multiple choice questions (that is, choose an answer out of A, B, or C) and Numerical-answer questions. There have been no signs of open-ended questions.

As for the time limit, the whole task is given a total of 35 minutes for both parts. While there are no official time constraints, we recommend spending 25 minutes on the first part, and 10 minutes on the second part to optimize your outcome.

Breaking down the study in Part 1

In the first part of the Redrock Study Task (we’ll refer to this as the study or case), the study’s flow is designed to test candidates’ logical and reasoning skills. If you don’t follow the logic carefully, the algorithm may be unable to recognize your thinking process, and view you negatively. Here, we have broken the study down into 4 aspects.

Game aspect 1: understanding the study

This refers to the first phase of the Redrock Task, which is INVESTIGATION. To truly grasp what you need to do, you must first clearly identify the case's objectives. Then, your next task is to understand all the data points presented within the case, to identify which ones can be used to answer the objective.

In general, all information presented on the screen is needed towards understanding and solving the case. But some are less important than others. Background information and test instructions are usually text-based data that you can’t select or move around. They only serve to give you an overview of the case, like the case’s theme, and don’t need to be collected. 

By contrast, important data points are highlighted and presented in boxes on the screen. You can click and drag these boxes around to work inside the case. Among these movable data points, there are 3 types of crucial information that you need to find:

• Case objectives: These are text based data, informing you about the goal that you must solve in the case. It usually sits at the top of the case, right after the instruction. 

• Calculation instructions: These are data points telling you which math formula you must use and which numbers to choose. They are often long texts/sentences that describe the relationships (higher/lower/etc.) between subjects in the case.

• Numbers: These make up the largest portion of the data points in the case. They usually appear in charts/diagrams (bar chart, pie chart,...), tables, or sometimes in-between texts. You have to collect these numbers into the journal to calculate in the next phase. Only a small percentage of these numbers (10-15%) are actually important to the case.

Figure 3: Data points in the study

In general, the rule of thumb is that once you have collected the case’s objectives, you must identify which math formula to use. Only then can you gather suitable numbers that the calculation requires. Also, only a handful of data points are necessary to solve the case, so pick wisely.

Game aspect 2: collecting data points

You can drag any movable data point in any phase of the Redrock test into the Journal to “collect” it. In the Research Journal, each collected “data point” will show up as a card, with its own label and description. Data in the Journal can be used to feed into the Calculator, or into “answer inputs”, (blank spaces under the questions).

Some data comes with appropriate labels for its contents, but some do not. All data labels can be manually changed – we recommend doing so if the default label does not adequately describe the contents. Appropriate labeling will speed up your analysis later, since it allows you to quickly identify the relevant data.

Once collected, each data point can also be highlighted by using the “I” button (presumably for “important”) on the left of its label. Toggling on this button will cover the whole data point in an orange tint. We recommend highlighting information that is needed during the ANALYSIS (or calculation) phase.

Inside the Research Journal, you can move these data points up and to organize them from top to bottom. It’s possible that McKinsey will look at how you organize the data. We’ll give some insights on that later. The specific sorting method is still receiving changes, so we’ll update it as we go.

Figure 4: The research journal, which is always present on the left of your screen

Game aspect 3: processing the data points for insights

During the second phase of the game, you will be provided with 3 quantitative questions that directly relate to the game’s objective. Each one has 2-3 sub-questions with an answer input gap requiring an answer from the calculator. To answer these questions, you have to feed the collected numerical data points into an on-screen calculator, then drag the results into the appropriate gap.

The calculator has a simple interface, similar to a phone’s digital calculator, with basic operators like *,+,-,/. It’s safe to assume that the math involved are usually simple calculations (similar to most candidates' reports). Though they lack the ‘%’ button for percentage calculation.

We recommend that you perform all calculations on the provided calculator, as all your operations are recorded in a history log. So, we assume that how you work towards the answers will also weigh on the final results.

A recommended workflow is to drag the data points from your research journal into the calculator’s input screen to perform the operation. Then you’ll need to drag the result and drop them into the blank space in the question. You should avoid typing the number on your keyboard as it may lead to unfortunate typos.

Here are a few confirmed question types and calculations during phase 2 of part 1:

• Basic operations (add/subtract/multiply/divide): Basic operations don’t often sit alone. They usually have to be involved in complex questions.

• Simple percentages and ratios: They require you to calculate simple ratio, percentages and fractions. For example: “What is the percentage of population growth between 2021-2022?” (Provided data: Population number in 2021, Population number in 2022)

• Compound percentage questions: They require you to calculate multiple ratios and percentages in a row. For example: “What is the population number at the end of 2023?” (Provided data: Population number at the start of 2022, Population growth rate for 2022, Projected increase in population growth rate for 2023 compared to growth rate for 2022)

One important thing to note, as reported, the results that you get from these questions are almost always needed in the REPORT phase. There’ll be a review screen so ALWAYS collect your answers into the journal.

Game aspect 4: completing the case report

The Report phase is the last part of the Redrock Study Task. It consists of two parts: Summary and Data Visualization.

• Summary involves filling in the blanks of a text-format report, using numbers already given and produced in the previous phases, and expressions such as “higher”, “lower”, “equal to”, etc. The blanks in this phase will likely be somewhat like the answer inputs in the Analysis phase.

• Data Visualization involves choosing the correct type of chart and filling in the numbers to produce a meaningful chart for the report. For this step, a difference between the Redrock Study and the old McKinsey PST is the lack of compound chart type. This drastically reduces the difficulty, as you only have to work with simple chart types like bar or pie charts.

Figure 5: Screenshots of questions for the report phase

Mastering the Redrock Study

From what we can see, the Redrock Study Task is more similar to its Problem-Solving Test predecessor than a game. That makes the tips to this task a bit different from the previously-popular Plant Defense game. There’s no instant formula that can guarantee the best chance of survival (maybe this is why Plant Defense got canceled), rather, you must act and think like a McKinsey consultant.

Tip 1: Show a top-down and structured approach while collecting data

A good McKinsey consultant always takes a top-down approach when analyzing a problem, and recruiter often favor candidates with this trait. During the Study, McKinsey can assess this trait when you collect and arrange data.

Always collect the objectives first. They are the central problems of the case, and represent the highest level of your issue tree. You must always collect them into the Research Journal. If they are too long, you can always note down a summary on a piece of scratch paper.

Figure 6: Study's objectives

The next step is to identify the math formula. This type of data governs which calculation formula you need to use, and in turns, which numbers to collect next. We’ll call this the relational data. The objectives will determine the relational data points you need.

Finally, collect the necessary numbers. These are the ones needed for calculating and filling in the final reports. Collect only the ones you need by analyzing the objectives and relational data. Don’t collect all data points erratically, as this showcases that you have no structured thinking.

Tip 2: Label and organize data

As stated before, once collected into the journal, each data point will have a label and description. Some data points already have good labels, some do not.

It’s possible that McKinsey can recognize good labels, so we suggest always changing the label and description of a data point when necessary. Good label can seem good to an algorithm, and it can also help you analyze them more conveniently. We have a few suggestions as to what constitute a good label:

• What is the timeframe? (“Is this data for 2020, or 2021?”)

• Which subjects are concerned? (i.e., the things represented by rows and columns in a spreadsheet, or axes on a chart).

• Is there anything else I need to keep in mind? (i.e., the footnotes or any auxiliary information that accompanies a chart/table) 

As for arranging data, try to keep it consistent and top-down. “Overview” data points should be placed above the “granular” ones.

For example, keep the objectives at the top of your research journal, and below them are relational data points. Numerical data points from the same table should be placed together, and beneath the relational data that refers to them. McKinsey MIGHT take this as a sign that you are a structured person, if not, it will help you solve the case easier.

Tip 3: Avoid going back and showing indecisiveness

The game allows you to go back and forth freely between each phase to collect more data points. While this is great for when you make a mistake or need to double check, we don’t recommend doing so.

This behavior signals that the candidate does not understand each section fully and is uncertain about the task. And in phase 1, McKinsey’s instruction clearly states that you should collect all and only relevant data before moving on. It’s possible that moving back and forth can be viewed negatively by the algorithm.

Tip 4: Choose the correct chart-type (bar/line/pie)

We have written an entire guide on how to chart like a McKinsey consultant, so be sure to check it out before attempting this task. But in short, you need to choose the correct type of chart that best describes a certain type of data, in the McKinsey way.

Part 2 cases tear down

Since this part of the test has only been introduced recently, we are still in the process of interviewing and synthesizing insights. More information will be updated later as things develop.

TEST FLOW, FORMAT AND DIFFICILTY

There are 10 cases in Part 2, each has a question with directions, text information and data exhibits. Each case also has an onscreen tool to assist you. You must solve the cases sequentially, that means you can’t skip forward and must answer one case before the next.

All 10 cases will follow the same theme/topic with the Part 1 study. But from candidate reports, it’s safe to assume that the theme does not play any part in the answer, and each case is self-contained (which means you don’t need numbers of another case to get the answer).

The word count to the 10 cases can vary between 100 and 400 words. They only require a fundamental level of quantitative or reasoning skill to solve and don’t require advanced mathematical skills. But most of our candidates struggle to solve them within 10 minutes, so be careful. 

QUESTION TYPES

The questions types that we have seen from candidate reports generally mirror those in part 1. We categorize them into five main types: Word Problems, Formulae, Verbal Reasoning, Critical Reasoning, and Visualization. We also deduced the rate at which these questions appear part 2.

• Word problems (50%) are math exercises that require candidates to read the text and exhibit data to solve

• Formulae (20-30%) are a similar question type to word problems, but the candidate only needs to identify the formula used for calculation.

• Verbal Reasoning (7-8%) and Critical Reasoning (7-8%) are single-select multiple choice questions requiring candidates to choose a “true” or “false” statement among 3-5 options.

• Visualization (10%) requires the user to choose the correct type of chart to illustrate the given data.

Part 2 has a near identical format to a traditional Problem-Solving Test (except for the on-screen tool like a calculator similar to Part 1’s). Thus, to save time, we only recommend getting familiar with the interface and mastering fundamental knowledge for a McKinsey consultant (like the issue tree, MECE, etc.) which we covered many times before.

Watch more: McKinsey PSG Explained

Breaking down the test – Ecosystem Building

Mini-game overview & description

In the Ecosystem Building mini-game, you have to create an ecosystem with 8 species from a list of 39. There are three key objectives:

Here’s a detailed description of data and metrics in the mini-game, and how they relate to the objectives.

Objective 1: Terrain Match

There are two scenarios on which you must build the ecosystem: “the Mountain” and “the Reef”. 

Each location in the Mountain world has the 8 following specifications: Elevation, Temperature, Wind Speed, Humidity, Cloud Height, Soil pH, Precipitation, Air Pressure.

Each location in the Reef has the 7 following specifications: Depth, Water Current, Water Clarity, Temperature, Salt Content, Dissolved Oxygen, Wind Speed.

Terrain specifications have very little correlation.

Each species also has a few required terrain specifications – if these terrain requirements are not met, the species will die out. These requirements are often not exact numbers, but ranges (e.g: Temperature: 20-30 C). 

All 39 species are organized into 3 equal groups using their terrain specs – I call them “layers”. Species of the same layers have exactly the same terrain specs.

Objective 2: Food Chain Continuity

Each species has a few natural predators (Eaten By), and prey (Food Sources) – see below for exceptions.

The species are divided into producers (which are plants and corals, which consume no calories), and consumers. Consumers can be herbivores (plant-eating animal), carnivores (animal-eating animal), or omnivores (eats both plants and animals).

Producers always have the Food Sources as “sunlight” or other natural elements, i.e. they do not have prey. Some consumers are “apex animals”, meaning they do not have natural predators (can be recognized by empty the “Eaten By” specs). These have strategic implications in building the food chain. 

 Objective 3: Energy Balance

Each species has a “calorie needed” and a “calorie provided” figure. A species lives if its calorie needed is less than the sum calorie provided of the species it eats (so it has enough energy to survive) and its calories provided is higher than the sum calorie provided of the species that eat it (so it’s not eaten to extinction).

Two caveats apply here: a species often don't eat all of its prey and is not eaten by all of its predators. There are certain rules for priorities (see the “Feeding Overlap” issue) and more often than not, predators and prey will interact on a one-to-one basis.

In old versions of the game, each species will be placed on a group basis, with the number of individuals in each group ranging from 20 to 60. In these versions, calorie specs are “per individual”, so you have to perform the math to get the true consumption and production figures of the whole species.

New versions discarded this “per individual” feature, presenting the calorie specs for the whole species as one, but there is no guarantee the old feature won’t be re-introduced.

As of game-flow, the candidate is free to switch between choosing location and species during the mini-game. There is also a time bar on the top of the screen.

Old reports indicate that once you’ve submitted your proposed ecosystem, you would receive a scorecard in the end, showing how it actually plays out. Key measurements might include calories produced and consumed, and the number of species alive.

However, recent reports have indicated that results aren't displayed at the end. In either case, it is safe to assume that the underlying principles remain the same.

Cracking the mini-game

The biggest challenges in the Ecosystem Building mini-game are task prioritization and data processing – most test-takers report that they are overwhelmed by the amount of data given, and do not know how to approach the problem. However, the second problem can be mitigated by reading the rules very carefully, because McKinsey provides specific and detailed instructions in the tutorials.

To overcome both challenges at the same time, first, we need to know the “eating rules” (i.e. how species take turns to eat) and then we can develop a 3-step approach to meet those challenges.

Figure 8: Description of Ecosystem Building game interface

EATING RULES AND FEEDING OVERLAP

In the McKinsey PSG Ecosystem mini-game, species take turns to eat and get eaten, in accordance to very specific and comprehensive rules:

1. The species with the highest Calories Provided in the food chain eats first.

2. It eats the species with the highest Calories Provided among its prey (if the eating species is a producer, you can assume it automatically bypass this step, as well as steps 3-5).

3. The eating species then “consumes” from the eaten species an amount of Calories Provided that is equal to its Calories Needed, which is at the same time substracted an amount equal to the Calories Provided taken from the eaten species.

4. If there are two “top prey” species with the same Calories Provided, the eating species will eat from each of them an amount equal to 1/2 of its Calories Needed.

5. If the Calories Needed hasn’t been reduced to 0 (i.e.: satisfied), even if the eating species has consumed all the Calories Provided of the first prey the eating species will move on to the next prey with the second-highest Calories Provided, and repeat the above steps; the prey that has been exhausted its Calories Provided will be removed permanently from the food chain and considered extinct.

6. After the first species have finished eating, the cycle repeats for the species with the second-highest Calories Provided, then the third-highest, etc. until every species has already eaten. Note: in every step where species are sorted using Calories Provided, it always uses the most recent figure (i.e. the one after consumption by a predator).

7. At the end of this process, all species should have new Calories Provided and Calories Needed, both smaller than the original figures. A species survive when its end-game Calorie Needed is equal to 0, and Calorie Provided is higher than 0.

Let’s take a look at an example – try applying the rules above before reading the explanation, and see if you get it right:

Figure 9: Example of a food chain in Ecosystem Building minigame

Now, here’s how this food chain is resolved:

• The three producers automatically have their Calories Needed satisfied and does not need to eat anything.

• The first species to eat is an animal – the Mouse. It eats equally from Grass and Mushroom, which have equal Calories Provided, an amount of 2,000 each. The Mouse’s Calories Needed reduces to 0, while the Calories Provided for Grass and Mushroom reduce to 3,000 each (Grass and Mushroom survive).

• The second species to eat is the Squirrel. It should have eaten Grass, but Grass’s new Calories Provided is only 3,000, so the Squirrel picks Nuts instead. Squirrel’s Calories Needed becomes 0, while Nuts’ Calories Needed becomes 500 (Nuts survive).

• The third species to eat is the Snake. It eats the Mouse, reducing its own Calories Needed to 0 while taking 2,000 from the 3,000 of the Mouse’s Calories Provided. (Mouse survives).

• The third species to eat is the Snake. It eats the Mouse, reducing its own Calories Needed to 0 while taking 2,000 from the 3,000 of the Mouse’s Calories Provided. (Mouse survives).

• The fourth species to eat is the Fox. It eats the Squirrel, reducing its own Calories Needed to 0 while taking 2,000 from the 2,500 of the Squirrel’s Calories Provided. (Squirrel survives).

• The last species to eat is the Tiger. It eats the Snake first, taking away all of the Snake’s 1,500 Calories Provided, then proceeds to take 500 from the Fox’s 1,200, so that its Calories Needed can be reduced to 0 (Snake becomes extinct, Fox survives).

• The Tiger is not eaten by any other animal (Tiger survives).

Figure 10: Solution of a food chain in Ecosystem Building minigame

With these rules in mind, let us go through a 3-step process to building a food chain:

Step 1: Select the location:

• Use a spreadsheet or scratch paper to list the terrain specs and calories provided of the producers of the mini-game.

• Skim through the data to see which terrain specs remain the same across all species, and which ones change. Only changing terrain specs are relevant (there should be 2 of them), the rest are merely “noise” intended to cause information overload.

• Calculate the sum calories provided for the producers of each layer. The layer with the highest calories provided is likely to be the easiest to build the chain.

Step 2: Build the food chain:

• Look through the data to list the consumers with compatible terrain requirements in your spreadsheet.

• Pick the apex predator with the lowest calorie needed as the starting point of the food chain.

• Build the food chain top-down like an issue tree, by listing the food sources of the top predators. Continue drilling down until you reach the “base” level of corals and plants. Ideally the food chain should contain 3-4 levels, and 8 species.

• Alternatively, you can build the food chain in a bottom-up manner, by looking at the “Eaten By” specs of each species, until you reach the top predators. Our reports indicate that in real test conditions, this approach can be just as fast as the top-down one.

• During the whole process, try to prioritize species with high calories provided, and low calories needed. This should maximize the chance of calorie surplus in the food chain, and leave room for new additions should the first chain not meet the required 8 species.

• If you finish the chain short of the required 8 species, work bottom-up to find gaps (i.e unused surplus calories), and plug in those gaps with predators or plant-eating animals.

• The whole process should be done on a spreadsheet/scratch paper to facilitate calculations.

Step 3: Triple-check and adjust:

• Go back to the beginning of the process and check if every species of your food chain is compatible with the chosen location.

• Make sure the food chain is continuous – that is, the food sources listed fit with the description of each species.

• Check if each species in the food chain is supplied with enough calories and not eaten into extinction using the given eating rules.

• Adjust the food chain if any of the three checks are not met.

Watch more: McKinsey PSG Explained

Breaking down the test – Plant-Defense

*June 2023 Update: Though McKinsey is gradually phasing out this test, we are still receiving sporadic reports of it being used for candidates (about 10-20% in total). So for the sake of information sharing, this section will still remain on our article, and will be updated as changes happen.

Mini-game overview & description

The second mini-game of the McKinsey Solve Game – Plant-Defense – is a turn-based tower-defense game. The candidate is charged with defending a plant at the center of a grid-based map from invading pests, using obstacles and predators, for as long as possible, until the defenses are overwhelmed and the plant is destroyed.

Figure 11: Screenshot of Plant Defense minigame

Here’s a detailed description of the gameplay:

• The visual map is divided by a square grid (size from 10×10 to 12×12), with natural obstacles (called Terrain, or Terrain Transformations) are scattered across the map.

• The game has a recommended time allocation of 12 minutes per stage – which makes 36 minutes in total.

• The game is divided into three maps, each with 2 phases – “planning phase” and “fast-forward phase”.

• The “planning phase” is divided into 3 waves of 5 turns each. Every 3-5 turns, a new stack of Invader appears at the border of the map and starts travelling towards the center base – you have lay out defensive plans to tackle them. The phase lasts until you eliminated all the Invaders / you survive at the end of the 15th turn / your plant is destroyed.

• The “fast-forward phase” comes after the 15th turn of the planning phase. All the remaining Invaders from the planning phase will continue attacking. Your defensive scheme remains unchanged, and unchangeable. Invaders will continuously spawn and attack until the base is destroyed.

• After you’ve finished the game, the number of turns your plant survived will be used as the basis for the product scores.

Game aspect 1: Resources

At the beginning of each wave, you are allowed to choose and place 5 resources – divided into Defenders (such as Coyote, Snake, Falcon etc. which kill the Invaders) and Terrains (comprised of Cliff, Forest, and Rocky, which slow down or block the invaders). Each will be assigned to one turn of the current wave.

After each turn, the Defender/Terrain of that turn will be activated and locked – meaning you cannot change or remove its placement. The rest can be altered to adapt with the circumstances. The only exception is the Cliff, which activates right after its placement. 

Each Defender has a range/territory – once an invader steps into that range/territory, the Defender will damage them, reducing their population. The range vary between each Defender type – but in general the more powerful they are, the smaller their range is.

Each Terrain is effective towards different types of Invaders and in different ways, with some blocking the Invaders while others slowing them down.

Each Terrain and Defender will occupy one square. You cannot place Defender on top of an existing Defender, and if a Terrain is placed on top of an existing Terrain, it will replace the existing Terrain.

Defenders and Terrains form mutually compatible pairs which can exist on one same square. 

 Game aspect 2: Invaders

Invaders will appear from the map borders every 3-5 turns, in stacks of 100-200 population each, and move one step closer to your plant by each turn. The population of the stacks increase gradually.

Each Invader stack is accompanied by a path indicator – a long yellow arrow showing the direction it will take. The invader will always take this path unless blocked by Cliff.

Each Invader is countered by certain types of Terrain/Defender.

Figure 12: Description of Plant Defense minigame's interface

Cracking the mini-game

As the Plant Defense mini-game of the McKinsey Solve Game is essentially a tower-defense game, the basic tactics of that game genre can be applied – namely inside-out building and kill-zones. However, as the mini-game locks you from changing placement after a number of turns, contingency planning is also necessary.

I will elaborate each of those tactics:

INSIDE-OUT, MULTI-LAYERED DEFENSE

In this tactic, you build multiple layers of defenders outwards from the base, assisted by terrain.

Place your resources close to the plant first. As the inner rings of the map are smaller in circumference, and paths usually converge as you advance towards the center, this helps you maximize the coverage of each resource around the plant early on.

In the example below, the “inside-out” approach only takes 8 resources to protect the plant from all directions, while the “outside in” approach takes 24. With this approach, place your most powerful resources closest to the plant, and expand with the less powerful, but longer-range ones.

Figure 13: Visualization of the Inside-out, multi-layered defense tactic

BIG-PICTURE MINDSET

This isn’t so much of a “tactic”, but a reminder – after 15 turns, you won’t be able to change or place more resources, so try to identify the pattern of the invaders, and quickly adapt your strategy accordingly. It will take a few initial turns to experiment which works best for each type of invader.

Use your resources prudently, create an all-round protection for the plant – lopsided defenses (i.e heavy in one direction, but weak in others) won’t last long – and lasting long is the objective of this mini-game.

Alternative mini-games

In June 2023, we have received reports that these alternative mini-games have disappeared completely. When McKinsey decided that these games can’t accurately assess a candidate’s skills, they removed these tests. But in the future, as the McKinsey Solve evolves, there’s a chance they will re-adopt these games or develop new ones based on them. Thus, this section of the article exists only to provide a record, you can skip right to the next part.

Alternative 1: Disaster Management

In the Disaster Management mini-game of the Solve Game, the candidate is required to identify the type of natural disaster that has happened to an ecosystem, using limited given information and relocate that ecosystem to ensure/maximize its survivability.

With the two main objectives in mind, here’s how to deal with them:

• Identify the disaster: this is a problem-diagnosis situation – the most effective approach would be to draw an issue tree with each in-game disaster as a branch, skim through data in a bottom-up manner to form a hypothesis, then test that hypothesis by mining all possible data in game (such as wind speed, temperature, etc.)

• Relocate the ecosystem: this is a more complicated version of the location-selection step in the Ecosystem-Building mini-game, with the caveat that you will first have to rule out the locations with specs similar to the ongoing disaster. The rest can be done using a spreadsheet listing the terrain requirements of the species.

Like the Ecosystem Building mini-game, you will solve this mini-game only once, unlike the Plant Defense and the next Disease Management mini-games with multiple maps.

Alternative 2: Disease Management

In the Disease Management mini-game of the Solve Game, the candidate is required to identify the infection patterns of a disease within an ecosystem and predict the next individual to be infected.

The game gives you 3-5 factors for the species (increasing as the game progresses), including name, age, weight, and 3 snapshots of the disease spread (Time 1, Time 2, Time 3) to help you solve the problem.

There is one main objective here only: identify the rules of infection (the second is pretty much straightforward after you know the rules) – this is another problem-diagnosis situation. The issue tree for this mini-game should have specific factors as branches. Skim through the 3 snapshots to test each branch – once you’re sure which factor underlies and how it correlates with infection, simply choose the predicted individual.

Figure 14: Screenshot of Disease Management minigame

Alternative 3: Migration Management

The Migration Management mini-game is a turn-based puzzle game. The candidate is required to direct the migration of 50 animals. This group carries a certain amount of resources (such as water, food, etc.), often 4-5 resources, each with an amount of 10-30. Every turn, 5 animals die and 5 of each resource is consumed.

It takes 3-5 turns from start to finish for each stage Migration mini-game, and the candidates must place 15 stages in 37 minutes. The candidate must choose among different routes to drive the animals. In each stage, there are points where candidates can collect 3 additional animals or resources (1-3 for each type), and choose to multiply some of the collected resources (1x, 3x and 6x); the game tells the candidate in advance which resources/animals they will get at each point, but not the amount.

The objective is to help the animals arrive at the destination with minimal animal losses, and with specific amounts of resources.

With all of these limited insights in mind, here’s what I recommend for the strategy:

• Nearly every necessary detail is given in advance, so use a scratch paper to draw a table, with the columns being the resources/animals, and the rows being the routes. Quickly calculate the possible ending amount for each resources, assuming you get 2 at every collection point (good mental math will come in handy).

• Choose the route with the highest number of animals, and “just enough” resources to meet requirements.

Watch this video below for a detailed, visualized explanation of all frequently encountered McKinsey Solve games:

Test-taking tips for the McKinsey Solve 

Besides the usual test-taking tips of “eat, sleep and rest properly before the test”, “tell your friends and family to avoid disturbing”, etc. there are five tips specifically applicable to the McKinsey Solve Game I’ve compiled and derived from the reports of test takers:

Tip 1: Don’t think too much about criteria and telemetry measurements

You can’t know for sure which of your actions they are measuring, so don’t try so much to appear “good” before the software that it hurts your performance. One of our interviewers reported that he refrained from double-checking the species information in the Ecosystem Building mini-game for fear of appearing unsure and unplanned.

My advice is to train for yourself a methodical, analytic approach to every problem, so when you do come in for the test, you will naturally appear as such to the software. Once you’ve achieved that, you can forget about the measurements, and focus completely on problem-solving.

Tip 2: Don’t be erratic with in-game actions

While you don’t want to spend half your brain-power trying to “look good” to the software, do avoid erratic behaviors such as randomly selecting between the info panels, or swinging the mouse cursor around when brainstorming (yes, people do that – my Project Manager does the same thing when we do monthly planning for the website).

This kind of behavior might lead the software into thinking that you have unstable or unreliable qualities (again, we can never know for sure, but it’s best to try). One tip to minimize such “bad judgment” is to take your brainstorming outside of the game window, by using a paper, or a spreadsheet. 

Tip 3: Always strive for a better solution (Ecosystem Building)

Some of the interviewed test-takers seem to be under a wrong impression that “the end results do not matter as much as the process” – however, for the McKinsey Solve, you need good end results too. This is especially true in the Ecosystem Building, where a “right” answer with no species dying can be easily found with the right strategy.

Consulting culture is highly result-oriented, and this game/test has product scores to reflect that. Having a methodical and analytical approach is not enough – it’s no use being as such if you cannot produce good results (or, “exceptional” results, according to MBB work standards).

Tip 4: Showcase fundamental skills for a McKinsey consultant (Redrock Study)

McKinsey is always looking for candidates with the exact skill set for a model consultant: structured, logical, and professional. The McKinsey Solve Test is designed to do just that: to look for the right set of skills (with a lot of tracking and algorithms).

Through all parts of the Redrock Study Task, you must exhibit that you are a model McKinsey prospect. Here are a few things that they will value during the Redrock Study:

• Strong mental math skills: A consultant MUST quickly pitch insights and calculations to clients and CEOs (elevator pitch). You’ll have to quickly choose a logical math formula and deliver results (not necessarily accurate). That’s why in all stages of the test involving math and a calculator, always do your calculations step-by-step on screen (if there’s an on-screen tool). 

• Structured, top-down thinking: A candidate has to demonstrate that they are  a hypothesis-driven, structured problem solver. In other parts of the interview process (like the case interview), it is shown through a MECE, top-down issue tree. In the Redrock Study Task, you can show off this skill via organizing data points in the Research Journal, which we discussed above.

• Choosing the right charts: A McKinsey consultant will chart like a McKinsey consultant. Each type of data must go with a corresponding type of chart. We have included a guide on consulting charts in our product shop. So check it out 

We have also linked to relevant preparation resources below, to help you master these skills more easily. So be sure to check them out.

Tip 5: Prepare your hardware and Internet properly before the test

While the McKinsey Solve Test does not require powerful hardware, the system requirements are indeed more demanding than the usual recruitment games or tests. A decent computer is highly-advised – the smoother the experience, the more you can focus on problem-solving.

On the other hand, a fast Internet connection is a must – in fact, the faster, the better. You don’t want to be disconnected in the middle of the test – so tell other users on your network to avoid using at the same time as the test, and go somewhere with a fast and stable connection if it’s not available at your home.

How to practice for the McKinsey Solve Test

Hypothesis-driven problem-solving approach

See this article: Issue Tree, MECE

You may have noticed a lot of the solutions for the mini-game involve an “issue tree” – the centerpiece of the hypothesis-driven problem-solving approach that real consultants use in real projects.

This problem-solving approach is a must for every candidate wishing to apply for consulting – so learn and try to master it by applying it into everyday problems and cases you read on business publications. Practicing case interviews also helps with the McKinsey Solve as well.

You can see the above articles for the important concepts of consulting problem-solving.

Mental math and fast reading skills

See this article: Consulting Math, Fast Reading

The McKinsey Solve Test – especially the 3 ecosystem-related mini-games – require good numerical and verbal aptitude to quickly absorb and analyze the huge amounts of data. Additionally, such skills are also vital to case interviews and real consulting work.

That means a crucial part of practice must include math and reading practice – see the above articles for more details on how to calculate and read 300% faster.

Practice with video games

*June 2023 update: As many games in the previous PSG have been eliminated, playing video games as part of practice has become less effective. But, we still recommend playing similar games to the Ecosystem Building (mainly) and Plant Defense mini-games.

Test-takers who regularly play video games, especially strategy games, report a significant advantage from their gaming experience. This is likely due to three main factors:

• The McKinsey Solve Test’s games are in fact similar in logic and gameplay to a few popular video game genres. The more similar a game is to the McKinsey Solve, the better it is for practice.

• Video games with data processing and system management also improve the necessary skills to pass the Solve.

• Playing video games helps candidates understand how the interface as well as the objective system of a game works – improving their “game sense”.

I am not a fan of video games – in fact, after leaving McKinsey I founded an entertainment startup with the mission to fight the increasing popularity of video games. Yet now I have to tell you to spend a few hours each week playing them to get into McKinsey.

The question is, which games to play? Here’s a list of the games and game genres my team have found to possess many similarities with the McKinsey Solve Test:

City-building games

• SimCity series

• Caesar series (Zeus and Poseidon, Caesar III, Emperor ROTK)

• Anno series (Anno 1404, Anno 2070, etc.)

• Cities Skylines

These are very similar in logic to the Ecosystem Building mini-game – you need to balance the production and consumption of buildings and communities, which usually have specific requirements for their locations.

The difference between these and the PSG is that most games are real-time and continuous, meaning you have the opportunity to watch your city develop and correct the mistakes – in the Solve you need to nail it from the start! With that said, the amount of data you need to process in these games will make the McKinsey Solve a walk in the park; the learning curve is not too high either, making these games good practice grounds.

Figure 15: Screenshot from Cities Skylines

 Tower defense games

• Kingdom Rush series

• Plants vs Zombies series

Tower-defense games such as Kingdom Rush are near-perfect practices for the Plant Defense mini-game of the McKinsey PSG. Our basic “kill-zone” tactic in fact comes from these games.

Again, there is a caveat when practicing with games – both Plants vs Zombies and Kingdom Rush allow you to correct your mistakes by having the invaders attack the base multiple times before you lose. Both games also feature fixed and predictable paths of invasion. In the PSG, the path of the invaders changes with your actions, and if they reach your base, you’ll lose immediately.

Figure 16: Screenshot from Kingdom Rush

Grand strategy and 4X games

• Civilization series

• Europa Universalis series

• Crusader Kings series 

Grand strategy and 4X games combine the logic of system-building and tower-defense games (with Civilization being the best example), making them good practice for both games of the Test . They also require players to manage the largest amount of data among popular game genres (sometimes multiple windows with dozens of stats each).

However, they are also the game with the steepest learning curves – so if you are not one for video games, and/or you don’t have much time before the Test, these games are not for you. They are also less similar to the PSG on the surface, compared to the two genres above.

Figure 17: Screenshot from Civilization VI

New release: Redrock Expansion (early access), an update of McKinsey PSG simulation

In 2023, we released a new product – Redrock Expansion to feature a new game of McKinsey. The Redrock Simulation can be purchased standalone or in Mckinsey Solve Simulation (All-in-one package).

As the official game is still in Beta, we are constantly updating the product. The simulation is now providing a 90% accurate reconstruction of the Part 1 case. Part 2 will come later in a future update.