McKinsey Solve Assessment: The Complete Preparation Guide

The McKinsey Solve assessment eliminates roughly 70% of candidates before they ever reach a case interview. Since around 2020, Solve—McKinsey’s gamified digital assessment—has replaced the paper-based Problem Solving Test (PST) as the global default. If you are applying to McKinsey in 2026, this is almost certainly the test you will face.

This guide is the master reference for the Solve assessment: what it is, which mini-games you will encounter, how the algorithm scores you, and how to prepare deliberately. It also includes a short, clearly-labeled section on the legacy PST for the small number of regional offices that still administer the older format.

What Is the McKinsey Solve Assessment?

Solve (sometimes called the “McKinsey Digital Assessment” or, historically, the “Imbellus game” after the company that built it) is a set of ecology-themed strategy games. Instead of asking you to answer multiple-choice questions, it drops you into simulated environments and watches how you make decisions.

Here is the shape of a typical session:

• Two mini-games per session. You are assigned two of McKinsey’s mini-games. Most candidates get Ecosystem Building paired with the Red Rock Study (a data-analysis investigation). A minority encounter other games such as Sea Wolf or Plant Defense.
• Roughly 60–81 minutes total. The two games run back-to-back with a short transition between them. Each game carries its own on-screen timer.
• Algorithmic, process-based scoring. Solve records your decision sequences, click patterns, and time allocation—not just your final answers. Because it measures behavior, random guessing is actively penalized.
• A screening filter. Solve sits before the interview rounds and screens out about 70% of applicants. Roughly 30% advance to the Personal Experience Interview (PEI) and the interviewer-led case.

The takeaway: Solve is not a quiz you can cram trivia for. It is a behavioral assessment of your problem-solving process, and preparation means building genuine, repeatable analytical habits.

Why McKinsey Moved from the PST to Solve

The paper PST had well-known limitations that Solve was designed to fix:

1. Test-prep vulnerability: Candidates could memorize question patterns and pass without genuine problem-solving ability.
2. Cultural and language bias: Reading-heavy questions disadvantaged non-native English speakers.
3. Limited signal: A multiple-choice format captures only the final answer, never the thinking that produced it.
4. Logistics and cheating risk: Supervised paper tests were expensive to run and still vulnerable.

Solve addresses these by measuring the process behind your decisions, not just the conclusion. The gamified format also gives McKinsey far richer behavioral data and a more engaging candidate experience.

The Solve Mini-Games

Solve draws from a small pool of mini-games, and you are assigned two per session. The specific pairing is determined algorithmically and can vary by office and cycle, but the overwhelmingly common experience is Ecosystem Building + Red Rock.

flowchart TD
    A[McKinsey Solve Assessment] --> B{Two Mini-Games Assigned}
    B --> C[Ecosystem Building]
    B --> D[Red Rock Study]
    B -.occasional.-> E[Sea Wolf]
    B -.occasional.-> F[Plant Defense]

    C --> G[Constrained Optimization]
    D --> H[Data Analysis]
    E --> I[Systems Thinking]
    F --> G

    G --> J[Process-Based Score]
    H --> J
    I --> J

    J --> K[Advance to PEI + Case, or Screened Out]

Ecosystem Building (Core Game)

The challenge: You are presented with an environment—a mountain, a coral reef, or a similar habitat—and a list of 30–40 candidate species. Your task is to select a set of species that form a sustainable food chain under the environment’s constraints (altitude, temperature, terrain type, moisture, and similar rules).

What is really being tested: Constrained optimization and systematic hypothesis testing. The algorithm watches whether you:

• Read and understand all constraints before you start selecting
• Approach the problem systematically (for example, starting from producers)
• Adjust your strategy when a selection does not work
• Backtrack efficiently rather than tearing down and rebuilding from scratch

Core strategy:

1. Map the constraints first (2–3 minutes). Before selecting anything, read every species card and note which environmental conditions each requires. Group them mentally by trophic level (producers → primary consumers → secondary consumers → apex predators).
2. Start from one end of the food chain. Working upward from producers (plants/algae) is usually easier because producers have fewer dependencies. Pick producers that match the terrain, then find primary consumers that eat those producers and satisfy the environmental constraints.
3. Use paper. Build a simple grid: species down the left, the environmental constraints across the top. This stops you from rechecking the same information.
4. When stuck, backtrack one level. If you cannot find a valid apex predator, do not restart. Swap one secondary consumer for an alternative and see whether that opens new options above it.
5. Verify the complete chain before submitting. Trace the energy flow from bottom to top. Every species must have at least one food source below it (except producers) and satisfy every environmental constraint.

For a deeper tactical walkthrough of Ecosystem Building, see our Ecosystem and Red Rock strategy guide.

Red Rock Study (Data Analysis)

The challenge: The Red Rock Study (also written “Redrock”) is an investigation game built around a research packet—several pages of ecological data in charts, tables, scatter plots, and text. You work through a case that asks you to extract, combine, and apply the data to answer a sequence of questions, often while the source material becomes harder to reference as you progress.

What is really being tested: Your ability to extract, synthesize, and apply quantitative information under time pressure—the same muscle you use in profitability cases and financial analysis.

Core strategy:

1. Take structured notes while you read. Do not try to memorize. Write down key numbers, trends, and relationships, organized by data source (Chart A, Table B, and so on).
2. Record scales and units explicitly. Many wrong answers come from misreading axes—thousands versus millions, monthly versus annual. Write the units down.
3. Identify relationships. Which variables move together? Which tables connect to which charts? Draw arrows between related data points in your notes.
4. Read each question twice. Pinpoint which data source contains the answer. If your notes are organized, you will not need to search.
5. Separate what is proven from what is merely suggested. The correct answer must be directly supported by the data, not just plausible.

Because Red Rock is fundamentally a data-interpretation task, the drills in our Red Rock data analysis guide transfer directly.

Occasional Games: Sea Wolf and Plant Defense

Most candidates never see these, but it is worth recognizing them.

• Sea Wolf (systems thinking): You manage a population across multiple rounds—allocating resources and adapting as conditions such as weather, food availability, and competitor behavior change. It rewards identifying which variables matter most, predicting second-order effects, and adapting when conditions shift. Avoid over-optimizing a single metric; the game usually tracks several success criteria at once.
• Plant Defense (constrained optimization): A tower-defense-style game where you protect a habitat by placing defensive organisms. Each placement costs resources and threats arrive from specific directions with varying strength. Assess all threats before committing resources, place defenses at chokepoints that cover multiple threats, and reserve resources for the stronger later waves.

Do not over-invest in preparing for these specifically. The underlying skills—systematic analysis, resource allocation, adaptation—are the same ones Ecosystem and Red Rock reward.

How Solve Is Scored

Solve is scored algorithmically and by process, which is what makes it fundamentally different from the old PST. Understanding the scoring dimensions lets you practice with intent.

The Skills McKinsey Measures

Skill	What It Means	How It Shows Up in the Games	How to Demonstrate It
Critical Thinking	Analyzing information logically	Quality of your decision sequence	Make decisions that follow from the available data
Decision-Making	Acting on your analysis	Speed and accuracy of choices	Act once you have sufficient information; don’t over-deliberate
Metacognition	Executing a strategy consistently	Consistency of your approach	Show a clear, repeatable method, not random exploration
Situational Awareness	Anticipating change	Responsiveness to new information	Adjust promptly when conditions shift
Systems Thinking	Understanding cause and effect	Ability to predict outcomes	Consider second-order effects before acting

How the Algorithm Reads Your Behavior

McKinsey does not publish its scoring formula, but the design goals are clear and consistent across candidate reports:

Process over outcome. Two candidates can reach the same answer and score differently. The one who got there through systematic exploration scores higher than the one who stumbled onto it by trial and error.

Time allocation is a signal. Spending proportionally more time on genuinely complex decisions—and less on simple ones—shows good judgment about problem difficulty. This is telemetry: the assessment logs where your time and clicks go.

Consistency is rewarded. A clear, repeatable methodology beats switching approaches mid-game for no reason.

Recovery counts. An error is not fatal. Recognizing it quickly, understanding why it happened, and adjusting systematically is exactly the metacognition McKinsey wants to see.

Random behavior is penalized. Because scoring is behavioral, guessing and rapid, patternless clicking actively hurt you. There is no “guess and hope” strategy the way there was on the PST.

Estimated Scoring Emphasis

McKinsey does not release exact weights. The distribution below reflects aggregated candidate feedback and third-party analysis and is directional, not official:

Scoring Dimension	Estimated Emphasis	What Tanks Your Score
Problem-solving approach	Highest	Random clicking, no clear strategy
Decision quality	High	Ignoring available information
Adaptability	Moderate	Failing to adjust when conditions change
Efficiency	Lower	Excessive time on simple decisions
Completion	Lower	Not finishing the game

Passing the Filter

McKinsey does not publish a specific passing score for Solve, and you should be skeptical of any source that claims a precise cutoff. What we can say with confidence:

• Solve screens out roughly 70% of candidates; about 30% advance to interviews.
• There is no “partial pass”—you either advance to the PEI and case, or you do not.
• Thresholds appear competitive rather than absolute, and can vary by office and applicant pool.

Treat Solve as pass/fail. Your goal is to clear the filter with a demonstrably structured process, not to chase a mythical perfect score.

Your Preparation Roadmap

Solve rewards habits, not cramming, so structured practice over a few weeks beats a last-minute sprint. A proven approach for candidates with full-time jobs is a four-week plan; students with more time can compress it to two or three weeks.

flowchart LR
    W1[Week 1: Foundation] --> W2[Week 2: Skill Building]
    W2 --> W3[Week 3: Simulation]
    W3 --> W4[Week 4: Peak and Rest]

    W1 --- W1D[Understand the format\nDiagnostic self-assessment\nIdentify weak areas]
    W2 --- W2D[Mental math drills\nData interpretation\nStrategy development]
    W3 --- W3D[Full timed simulations\nReview your process\nBuild consistency]
    W4 --- W4D[Light practice\nConsolidate 3-5 rules\nRest before test day]

The four phases, in brief:

1. Foundation and diagnosis. Learn the format, run a diagnostic, and pinpoint your two or three weakest areas.
2. Core skill building. Drill mental math, data interpretation, and systematic problem-solving until they are second nature.
3. Full simulations. Practice under realistic, timed conditions and review not just your answers but your process—where did you hesitate or improvise?
4. Peak and rest. Consolidate your strategy into a handful of rules and arrive rested. Fatigue costs more than a final cramming session can add.

For the complete day-by-day schedule with daily activities and time budgets, follow our four-week McKinsey preparation plan.

Mental Math: The Foundation Skill

Red Rock and the analytical portions of Solve require quick calculation without a calculator. Candidates who invest in mental math consistently see the largest improvements—it is often the difference between advancing and being screened out.

Percentage calculations (the most common):

• 10% = move the decimal one place left: 10% of 840 = 84
• 5% = half of 10%: 5% of 840 = 42
• 15% = 10% + 5%: 15% of 840 = 84 + 42 = 126
• 1% = move the decimal two places: 1% of 840 = 8.4
• Combine: 23% of 840 = 20% + 3% = 168 + 25.2 = 193.2

Division shortcuts:

• Divide by 5 = multiply by 2, divide by 10: 840 / 5 = 1680 / 10 = 168
• Divide by 8 = halve three times: 840 / 8 = 420 / 4 = 210 / 2 = 105
• Rough division: 2,350 / 7 ≈ 2,100 / 7 + 250 / 7 = 300 + 36 ≈ 336

Compound growth estimation:

• Rule of 72: years to double = 72 / growth rate. At 8%, roughly 9 years.
• Five years at 8%: multiply by 1.08^5 ≈ 1.47 (so 100 → about 147).

A daily 15–20 minute routine: ten percentage problems, five division problems, three compound-growth estimates, and two multi-step business math problems (revenue = price × volume, margin calculations). Our mental math guide sequences these drills so speed builds progressively.

Tactical Tips for Each Mini-Game

Ecosystem Building: 10 Tactical Tips

1. Spend the first three minutes reading only—do not click anything.
2. Build a constraint matrix on paper: species down the left, constraints across the top.
3. Count the trophic levels you need before selecting (usually three to four).
4. Start with the producers that match the most terrain constraints.
5. Check predator–prey compatibility before confirming any selection.
6. When stuck, swap one species rather than rebuilding from scratch.
7. Track eliminated species so you do not recheck them.
8. Watch for “keystone” species that connect multiple branches of the food chain.
9. Verify the chain is complete before submitting—every consumer has a food source.
10. If you have time left, check whether an alternative combination is more robust.

Red Rock: 8 Tactical Tips

1. Use the first 30 seconds to count the exhibits and scan the question topics.
2. Build a “data map” noting what information each exhibit holds.
3. Write down exact numbers for any data point that looks important.
4. Mark trend directions with arrows (↑ ↓ →) rather than trying to memorize values.
5. Identify outliers in any dataset—they are frequently tested.
6. Check axis labels obsessively; they are designed to trip you up.
7. For comparison questions, write the values side by side rather than holding them in your head.
8. Budget your time: roughly 60% on reading and notes, 40% on answering.

General Solve Tips

• Stable connection. Use a wired connection if you can; lag can distort your behavioral data.
• Quiet environment. Interruptions force you to re-orient, which reads as inconsistency to the algorithm.
• Full screen, no distractions. The assessment tracks focus patterns.
• Practice the interface if an official demo is available.
• Never refresh the page. It can reset your session or lose progress.

Common Mistakes and How to Avoid Them

Strategic Mistakes

Mistake	Why It Hurts	How to Fix
Starting without reading instructions	You miss critical constraints and look disorganized to the algorithm	Force yourself to read everything first, even when anxious
Random trial and error	The algorithm detects non-systematic behavior	Always have a reason for each action
Restarting from scratch	Wastes time and signals an inability to recover	Backtrack one step at a time
Ignoring time management	Incomplete submissions score worst	Set checkpoints at 25% / 50% / 75% of the clock
Over-optimizing one dimension	The games test multiple criteria	Balance speed, accuracy, and adaptability

Technical Mistakes

Mistake	Why It Hurts	How to Fix
Unstable internet	Lag can corrupt your behavioral data	Use a wired connection; close other tabs
Taking it on mobile	A small screen impairs performance	Use a laptop or desktop with a large display
Interfering browser extensions	They can block game elements	Use incognito / private mode
Noisy environment	Interruptions fragment your process	Book a quiet room; tell your household
Testing while tired	Cognitive performance drops 20–30%	Schedule for your peak alertness

For a fuller catalog of pitfalls, see our guides on common mistakes and time management.

Practice Resources

Free resources:

1. McKinsey official practice. McKinsey occasionally posts sample material on its careers page—check regularly.
2. Ecosystem and food-chain games. Anything that involves building a viable system under constraints trains the right mindset.
3. Mental math apps. Drilling percentages, division, and estimation builds the calculation speed Red Rock demands.
4. Logic and constraint puzzles. Advanced Sudoku-style puzzles train systematic elimination.
5. Real data. Government statistical publications give you authentic charts and tables to interpret.

Structured preparation:

• Our four-week preparation plan gives you a day-by-day schedule.
• Our mental math guide builds calculation speed systematically.
• Practice with real McKinsey-style cases to develop the analytical thinking Solve rewards.

What not to waste time on:

• Expensive “crack the Solve code” courses—McKinsey updates the assessment regularly.
• Memorizing questions from forums—there is no static question bank to memorize.
• Speed-clicking drills—the algorithm measures decision quality, not click speed.

Legacy: The Former PST Format

Historical context. The paper-based Problem Solving Test is McKinsey’s legacy screening test. It has been retired in favor of Solve almost everywhere. Read this section only if your regional office has confirmed it still administers the older PST—most candidates in 2026 will never see it.

The PST was a 60-minute, 26-question, paper-based multiple-choice test of business logic, data interpretation, and mental math. Its passing threshold sat around 70% correct (roughly 18–19 of 26 questions), with about 30–35% of candidates advancing.

Its question types, by rough share of the test:

Question Type	Share	What It Asked
Reading Facts	~38%	Locate a statement directly supported by a passage or chart
Fact-based Conclusion	~14%	Determine which conclusion the combined data proves
Root-cause Reasoning	~13%	Identify the factor that best explains an outcome
Word Problem	~12%	Translate a business scenario into math and solve step by step
Client Interpretation	~8%	Decide which recommendation the data best supports
Formulae	~5%	Apply a given formula, watching for unit conversions

The key difference from Solve: the PST scored only your final answers, so guessing carried no penalty and test-prep against known patterns could help. Solve reverses both of these—it scores your process and penalizes random behavior. The good news is that the underlying analytical skills overlap heavily, so mental math and data-interpretation practice serve you well on either format. If you have confirmed you face the PST, work through our practice guide for timed, format-specific drills.

After Solve: What Comes Next

Clearing Solve moves you into McKinsey’s interview rounds, which have two components:

• The Personal Experience Interview (PEI). Structured behavioral questions about times you showed leadership, drove impact, or overcame conflict. Prepare two or three detailed stories per theme.
• The interviewer-led case. A live case where the interviewer guides you through a business problem. This is where structured, hypothesis-driven thinking and clean quantitative work decide the outcome.

Shift your focus immediately once you advance:

• Practice with real McKinsey-style cases to get used to the interviewer-led format.
• Build hypothesis-driven structures for common case types like profitability and market sizing.
• Rehearse under realistic conditions with our AI Mock Interview.

Frequently Asked Questions

How long is the McKinsey Solve assessment?

A session runs roughly 60–81 minutes and consists of two mini-games back-to-back, each with its own timer and a short transition between them. You complete the games sequentially—you cannot skip ahead or return to a finished game.

Which mini-games will I get?

Most candidates get Ecosystem Building paired with the Red Rock Study. A minority encounter other games such as Sea Wolf or Plant Defense. The specific pairing is assigned algorithmically and can vary by office and cycle, so prepare the underlying skills rather than banking on one game.

How is Solve scored?

Algorithmically and by process. Solve tracks your decision sequences, click patterns, and time allocation, and rewards a systematic, consistent approach. Because scoring is behavioral, random guessing and patternless clicking are penalized—there is no “guess and hope” strategy.

What is the passing rate?

Solve screens out roughly 70% of candidates; about 30% advance to the interview rounds. McKinsey does not publish a specific cutoff score, and thresholds appear to be competitive rather than absolute. Treat it as pass/fail.

Can I retake Solve if I do not pass?

Typically not for 12–24 months, depending on the office’s policy. When you reapply you take the assessment fresh; your previous result is not carried over. Use the waiting period to build genuine analytical skills through case practice and data-heavy work.

Does Solve really track my mouse and clicks?

Yes. Solve uses behavioral analytics—mouse movement, click patterns, hover duration, decision sequences, and time allocation. This telemetry is how it distinguishes candidates who reach answers through systematic analysis from those who arrive by random exploration. Move deliberately and click with purpose.

What happens if I do not finish a game?

Incomplete games score significantly lower—completion is one of the scored dimensions. Even a suboptimal but finished solution beats an abandoned one. If time is running short, keep making quick but logical decisions; the algorithm can still read a positive process signal from rushed-but-structured choices.

Is the PST still relevant?

For almost everyone, no. Solve has replaced the PST as the global default. Only a few regional offices may still run the older paper test, and your recruiter will tell you if yours is one of them. See the legacy PST section above if that applies to you.

Key Takeaways

• Solve is McKinsey’s current gamified assessment and has replaced the paper PST as the global default.
• A session is two mini-games (usually Ecosystem Building + Red Rock) running roughly 60–81 minutes.
• Scoring is algorithmic and process-based; it tracks your decision sequences, clicks, and time, and penalizes random behavior.
• Solve screens out about 70% of candidates; roughly 30% advance to the PEI and interviewer-led case.
• Prepare habits, not trivia—work a structured multi-week plan and build genuine mental math and data-interpretation skills.
• Start any system from one end (top or bottom), never the middle, and document your hypothesis testing on paper.
• Finish every game even if imperfect; incomplete submissions score worst.
• The legacy PST is retired almost everywhere—only prepare for it if your office confirms it.

Ready to Prepare?

The surest way to build the structured, quantitative reasoning Solve rewards is deliberate practice on real problems. CasesCoach’s AI Mock Interview lets you rehearse breaking down cases and defending recommendations under pressure—the same thinking Solve measures through its games, and exactly what the PEI and interviewer-led case demand next. The free tier includes three full cases plus AI Mock; Pro unlocks 835+ real cases and multiple AI mock sessions so you can build the habit until it is automatic.