McKinsey Solve Prep: The Complete 2026 Guide
By Tom Prescott — ex-McKinsey Senior Engagement Manager who has recruited 150+ people· Last updated 30 August 2026
SolvePrep is an independent preparation platform and is not affiliated with, or endorsed by, McKinsey & Company.
Why Practice?
Let me start with something most prep sites won't tell you: McKinsey's own position is that you don't need to prepare for Solve. Their interviewing page literally says to approach it "with curiosity and creativity," and every game opens with an untimed tutorial.
Having sat on the other side of the table, I'll tell you why I respectfully disagree — and why McKinsey's advice and mine aren't actually in conflict. Solve is designed so you can't memorise your way through it: McKinsey's official FAQ confirms the assessment uses AI-driven variation so that no two candidates see the same parameters. What you can prepare is everything else: familiarity with the mechanics, pacing instincts, calculation speed, and the calm that comes from having seen the format before. Reading guides builds awareness. Only McKinsey Solve practice under realistic, timed conditions builds the rest.
This guide covers everything that actually moves the needle in 2026: how to decode your invitation email, how each of the three current games works, what's known (and honestly, what isn't) about scoring, the official test-day rules that can end your candidacy if you break them, a day-by-day practice plan, and a transparent FAQ. Every section links to a deeper guide or a free, timed simulator so you can stop reading and start practising whenever you're ready.
What you'll get from this page
- How to know before test day which games you'll face — from your invitation email
- A clear picture of the three 2026 mini-games and what each one measures
- What candidate reports and McKinsey's own communications reveal about scoring
- The official rules (AI tools, notes, screenshots) that many candidates never read
- A research-backed practice plan — with targets clearly labelled as ours, not McKinsey's
- Direct links to free, timed practice for every mini-game
First: Decode Your Invitation Email
This is the single most useful thing on this page, and almost nobody uses it. McKinsey's official FAQ states that the length of your Solve assessment is outlined in your invitation email. That number tells you exactly which games you'll face:
| Your invitation says | Games you'll get | Sequence |
|---|---|---|
| ~65 minutes | Red Rock Study + Sea Wolf | Red Rock → Sea Wolf |
| ~85 minutes | Red Rock Study + Sea Wolf + Sustainable Futures Lab | Red Rock → Sea Wolf → SFL |
The math is simple: Red Rock runs 35 minutes, Sea Wolf 30, and the Sustainable Futures Lab 20. Two games = 65 minutes. Three = 85. Tutorials before each game are untimed and sit outside these totals.
Three things to know about the sequence:
- The order is fixed. Red Rock Study always comes first, and you must complete it before Sea Wolf unlocks. If SFL is included, it comes last.
- Time does not carry over between games. Finishing Red Rock early doesn't buy you extra Sea Wolf minutes. (Within Sea Wolf, time does carry across the three sites — more on that below.)
- Rare variants exist. A small number of candidates report longer invitations that still include the legacy Ecosystem Building game. If your email states a different duration, trust the email over any guide — including this one — and see our background on the retired Ecosystem game.
So before you plan a single practice session: open your invitation, find the duration, and you'll know whether to prepare two games or three.
Watch: McKinsey Solve in Under 10 Minutes
A concise overview of the McKinsey Solve games, key strategies, and how to practise effectively. If you're on an 85-minute invitation, make sure your prep also covers the Sustainable Futures Lab, which was added in March 2026.
The 3 McKinsey Solve Games at a Glance
As of 2026, McKinsey Solve — which you may also see called the Problem Solving Game (PSG), the Imbellus game, or the McKinsey Digital Assessment — consists of up to three mini-games. Red Rock Study and Sea Wolf form the core pair that every candidate faces. The Sustainable Futures Lab, rolled out from March 2026, is the third module added on 85-minute invitations. You are not randomly assigned two of three: your invitation email tells you your exact lineup — but if you're on a 65-minute invite and your timeline allows, a light pass over SFL is cheap insurance in case McKinsey updates your format.
Red Rock Study
Tests: Quantitative reasoning, chart reading, hypothesis testing, structured synthesis
Also written "Redrock" or "Redrock Island". You play a researcher analysing wildlife population data — typically wolf and elk populations across regions of a fictional island. Under the ecological wrapper, this is the closest thing in Solve to a traditional consulting test: read exhibits, run the numbers, draw conclusions. The 35 minutes split into two parts. Part 1 (the Study) is one connected task in three phases: Investigation (gather data from the objective, text and exhibits), Analysis (calculations building on that data), and Report (assemble a short written-and-visual summary, including choosing the right chart). Part 2 (the Cases) is six independent mini-problems on related topics. The parts are one-way: once you leave the Study for the Cases, there's no going back. The math is consulting math, not university math — percentages and percentage points, ratios, growth rates, weighted averages, basic probability.
Sea Wolf
Tests: Constraint filtering, trait/range matching, optimisation under time pressure
Named after the research vessel you command, and often called the "Ocean game" or "Ocean Cleanup". Each of the three sites is contaminated with a different waste type and comes with its own required conditions and traits. Your job per site: filter the available microbes against the site's hard requirements, then assemble the best three-microbe treatment from what survives the filter. The 30-minute clock is shared across all three sites — banking time on site 1 is the difference between a composed site 3 and a panicked one. Of the three games, this is where time management fails candidates most often.
Sustainable Futures Lab
Tests: Judgment, prioritisation, trade-offs, stakeholder awareness, consistency
The newest game (you'll also see it written "Sustainable Future Lab" or just "SFL"), added to Solve in March 2026 and now common on 85-minute invitations worldwide. It's a different animal from the other two: no calculations, no data journal — a scenario-based simulation in which you join a fictional environmental research team (themes rotate: coastal habitat recovery, wetland restoration, air quality) and make decisions as the project evolves. You get 13 sequential questions in 20 minutes, opening with a drag-and-drop priority ranking followed by scenario-based decisions. Crucially, your choices shape later context — the questions aren't independent, which is why consistency matters here more than in any other module. The sustainability theme is the wrapper; team dynamics and judgment are the test.
How McKinsey Scores the 2026 Solve
Straight talk from a former McKinsey Senior Engagement Manager: McKinsey does not publish its scoring algorithm, pass marks, or weightings — and any guide claiming to know them is guessing. What follows is what McKinsey has said officially, plus the patterns consistently reported by candidates. I'll label which is which.
Official — from McKinsey itself. Solve results are not a standalone verdict: McKinsey's FAQ states the outcome is weighed together with the rest of your application, and the recruiter decides based on multiple factors. The FAQ also confirms the assessment generates unique parameters for every candidate — which is exactly why grinding one memorised scenario is wasted effort — and notes that candidates may be randomly asked to complete an additional task in person or to explain the logic they used. Let that last point sink in: your preparation has to build real skill you can reproduce and articulate, not answers you can't explain.
Consistently reported — the two-score model. Since the Imbellus era, candidate debriefs and prep-community consensus describe two dimensions: a product score (what you achieved — sites cleaned, answers correct) and a process score (how you got there — the sequence, efficiency and coherence of your decisions, captured through in-game telemetry). Treat this as well-established folklore rather than official doctrine, but it has a practical implication either way: work systematically. Erratic clicking and trial-and-error look exactly as bad as they sound, even when you stumble into the right answer.
Candidate-reported — Sea Wolf's silent deductions. Multiple recent candidate reports describe a deduction of roughly 20% for each site requirement — condition or trait parameter — left unmet in your final microbe selection, applied without any in-game warning. Reports also indicate that not every site allows a perfect score, so a "low" number isn't automatically a disaster. We model this mechanic in our simulator precisely because it's invisible on test day: practice is how you learn which choices trigger it.
What actually changed in 2026. The headline change isn't a re-weighting of the existing games — no credible evidence supports claims like "speed matters less now". It's an addition: the Sustainable Futures Lab bolts a behavioural dimension onto what was previously a purely analytical assessment. Candidate debriefs of SFL point to dimensions like prioritisation, working with ambiguity, balancing trade-offs, and team/stakeholder effectiveness — and because SFL's questions build on each other, internally consistent decision-making appears to matter more there than raw speed. The exact trait taxonomy varies by source, so treat any fixed "N-pillar" model (including ours in earlier drafts) as a training construct, not McKinsey's rubric.
For the full breakdown of scoring mechanics and the 2026 format changes, read our McKinsey Digital Assessment guide.
The Official Rules That Can End Your Candidacy
This section exists because I've seen the aftermath when candidates skip the fine print. These rules come from McKinsey's official Solve page and FAQ — read them there too.
During the real assessment you may not: use applications, websites, AI tools or pre-prepared notes; record or capture screenshots of any part of the assessment; or take it with anyone's assistance. McKinsey states that violations can mean exclusion from consideration, a rescinded offer, restrictions on future applications, or even legal action. The only exception is a disability accommodation approved in writing by McKinsey.
To be unambiguous about our own tools: SolvePrep's simulators and solvers are for practice only. Using any solver, second screen, AI assistant or notes during the live assessment is a values violation with career consequences — and given the random in-person verification described above, it's also a trap. Build the skill; don't borrow it.
Practical logistics, straight from McKinsey's FAQ:
- Device: PC or Mac only — no iPads or phones. A tech check runs before you start. Sound isn't necessary and a mouse is optional (though for Sea Wolf's drag-heavy interface, I'd strongly recommend one).
- Technical issues: contact McKinsey's support (email or the live chat listed in your assessment) at the moment the issue occurs. They can grant extra time or reset your link — but McKinsey explicitly won't make exceptions if you only report the problem afterwards.
- Languages: Solve is available in English, Spanish, Portuguese and Japanese — pick whichever you're genuinely fastest in.
- Accommodations: available for disabilities, health conditions and specific learning difficulties — talk to your recruiter early.
- Setting: McKinsey suggests an uninterrupted time slot. Treat it like a final-round interview: quiet room, stable connection, charged laptop, no notifications.
Recommended McKinsey Solve Practice Plan
Consistency beats cramming. If you have 7 days, do 2 full simulations per day rotating across your games. If you have 14 days, drop to 1 per day and add a 20-minute review after each. Prioritise by your invitation: on a 65-minute invite, put your hours into Red Rock and Sea Wolf; on an 85-minute invite, all three.
Read this before the table: the rep counts and target scores below are SolvePrep coaching targets, measured on our own simulators. McKinsey does not publish passing scores, and nothing here is, or implies, a McKinsey threshold. These are the levels at which, in our experience, candidates report walking into the real assessment feeling in control.
| Game | Suggested Reps | Target (on our simulators) | Time / Session | Key Focus Area |
|---|---|---|---|---|
| Red Rock Study | 6–10 full runs | ≥ 75% accuracy | ~40 min / session | Phase pacing, exhibit math, protecting the Cases |
| Sea Wolf | 8–12 full runs | ≥ 80% site cleanup score | ~40 min / session | Requirement filtering, trait/range fit, per-site time budget |
| Sustainable Futures Lab | 6–8 full runs | ≥ 70% consistency | ~40 min / session | Priority ranking, trade-off balance, decision consistency |
Pro tip: review your score breakdown after every run. The fastest path to improvement is identifying your weakest area and targeting it in the next session — not replaying on autopilot. And since McKinsey varies every candidate's parameters, vary yours too: new scenario every rep.
For a day-by-day schedule with rest days and specific drills, see our 7-day Solve prep plan or the longer how to prepare for McKinsey Solve walkthrough.
Practice by Game: What to Drill
Each mini-game rewards different habits. Below is what to focus on in each one, plus a direct link to a free, timed simulator so you can start practising in the next 30 seconds.
Red Rock Study — data analysis under a clock
- Budget the phases before you start: don't let the Study consume much more than ~25 minutes, so the six Cases get the final stretch. The most preventable Red Rock failure isn't the Report — it's strong candidates arriving at the Cases with three minutes left.
- Drill consulting math cold: percentages vs. percentage points, weighted averages, growth rates, ratios. Estimate first; reach for the on-screen calculator to confirm, not to think.
- Practise reading an exhibit in under 30 seconds before touching any question, and rehearse chart selection for the Report phase — it's quick points if your numbers are already in hand.
Sea Wolf — filter, then optimise
- Make "filter → optimise" a reflex: eliminate microbes that violate any site requirement first, then optimise the best 3-microbe treatment from what remains. Candidates who optimise before filtering burn minutes and leave requirements unmet.
- Track which requirements you leave unsatisfied across practice runs — those are the reported ~20% silent deductions — and eliminate your most frequent one before moving to the next.
- Rehearse a per-site time budget (roughly 10 minutes each, banking spare minutes early). Site 3 under pressure is where good runs die.
- Use our solver only after attempting a site yourself, and only in practice — never in the real assessment, where any external tool violates McKinsey's rules.
Background: the retired Ecosystem game (and what replaced it)
Sustainable Futures Lab — judgment, not math
- Read the mission briefing properly — it frames every later decision. Then commit to a defensible priority ranking and let it anchor you; second-guessing mid-module is how consistency erodes.
- Balance the reported dimensions (impact, feasibility, risk, stakeholders) rather than over-indexing on one. There's rarely a "perfect" option — McKinsey is watching how you trade off, not whether you found a hidden right answer.
- Time-box each scenario decision. Thirteen questions in 20 minutes leaves about 90 seconds each; deliberating for five minutes on question 4 is its own wrong answer.
6 Practice Mistakes That Tank Otherwise-Strong Candidates
Most candidates don't fail McKinsey Solve because they aren't smart enough — they fail because they practised the wrong way. These patterns come up over and over in post-assessment debriefs:
Practising without a timer — pacing is half the assessment, and untimed runs build false confidence.
Replaying one scenario on repeat. The real Solve generates unique parameters per candidate; only varied scenarios build transferable pattern recognition.
Skipping the score breakdown — the breakdown after each run is where the actual learning happens.
Cramming a weekend instead of spacing practice across 1–2 weeks.
Passive prep only. Guides and videos build awareness; they don't transfer to a timed, interactive test.
Ignoring your weakest game because it's uncomfortable — then meeting it cold on test day. (And if you're on an 85-minute invite, "SFL is just situational judgment" is exactly this mistake wearing a disguise.)
For 17 more specific tactics, including pacing rules per mini-game and the calculator habits that save time on Red Rock, see our McKinsey Solve tips guide.
What Makes a Good McKinsey Solve Simulator?
Not all practice tools are equal. A good McKinsey Solve practice tool should replicate the conditions and cognitive demands of the actual test — not just describe them. Here's the checklist we apply to every simulator, including our own:
SolvePrep's simulators are built to match every item on this list. Each game generates a fresh scenario on every run, scores your performance in real time, and shows you exactly where to improve. It's the closest thing to sitting the real McKinsey Solve — without the stakes.
McKinsey Solve FAQ
The questions candidates ask us most often before they start practising. Each answer links out to a deeper guide if you want the full picture.
How many games are in McKinsey Solve, and which ones will I get?
Up to three: Red Rock Study, Sea Wolf, and the Sustainable Futures Lab. You'll know your exact lineup before test day from your invitation email — around 65 minutes means Red Rock plus Sea Wolf; around 85 minutes adds the Sustainable Futures Lab as a third game. The order is fixed: Red Rock first, then Sea Wolf, then SFL.
Browse all simulatorsHow long does the McKinsey Solve assessment take?
About 65 minutes for the two-game version (Red Rock 35 min + Sea Wolf 30 min) or about 85 minutes with the Sustainable Futures Lab (20 min) included. The tutorial before each game is untimed, so plan for a somewhat longer total sitting. Your invitation email states your exact duration.
How the Digital Assessment is structuredCan you actually prepare for McKinsey Solve?
McKinsey officially says no preparation is needed, and it's true you can't memorise answers — every candidate gets unique parameters. What you can build is familiarity with the mechanics, pacing instincts, and calculation speed, which is why timed practice on varied scenarios helps where rereading guides doesn't.
How to prepare for McKinsey SolveHow long should I practice for McKinsey Solve?
Among SolvePrep users who report passing, the common range is 10–25 hours of focused practice across 1–2 weeks — roughly 6–12 full simulations per game plus a short review after each. That's our observed data, not a McKinsey requirement. A 7-day plan with rest days reliably beats a crammed weekend.
See the 7-day prep planIs there a free McKinsey Solve practice test?
Yes. SolvePrep gives you one free, full-length play of each mini-game with the real timer and a scored breakdown — no credit card. It's the closest thing to a free McKinsey Solve practice test that mirrors the 2026 format, including SFL.
Where to get a free PSG simulationCan I use ChatGPT, AI tools, or notes during the real assessment?
No. McKinsey's official rules prohibit applications, websites, AI tools and pre-prepared notes, as well as recording or screenshotting the assessment or taking it with anyone's help. Stated consequences range from exclusion to rescinded offers and even legal action — and McKinsey notes candidates may be randomly asked to redo a task in person or explain their logic. Practice tools belong in practice.
Is the McKinsey Solve proctored?
No human watches you take it — candidates commonly report no live proctor and no webcam monitoring during the Solve. What McKinsey runs instead is quieter and arguably scarier: a tech check before you start, official rules with real teeth, and — per McKinsey's own Solve FAQ — the possibility of being randomly asked to redo a task in person or explain your logic afterwards. An unproctored test where your reasoning can be spot-checked later is not a test to cheat on. It's a test to prepare for.
Does McKinsey Solve record you?
Not your face — your moves. Candidates report no video or audio capture during the assessment, but the game itself logs how you work: what you click, in what order, how you spend the clock. That behavioral log is the "process" half of scoring, weighed alongside your final results — the half of your performance a results-only readout misses. We built our simulators to score process for exactly this reason: it's the part of your performance you can't see, and the part practice improves.
Does McKinsey Solve require a camera?
McKinsey's FAQ lists what your setup needs — a PC or Mac (no tablets or phones), a tech check before launch, sound optional, mouse optional — and a webcam isn't on the list. Candidates commonly report taking the assessment with no camera involved at all. Set up like it matters anyway: quiet room, stable connection, charger in. Our test-day checklist covers the full setup in fifteen minutes: the test-day setup routine.
What happens if I have technical problems during Solve?
Contact McKinsey's support team (email or live chat, listed in your assessment) at the moment the issue occurs — they can grant extra time or reset your link. Per McKinsey's FAQ, no exceptions are made if you only report the issue afterwards. Also note Solve runs on PC or Mac only, with a tech check before you start.
How is the 2026 McKinsey Solve different from older versions?
The old Imbellus-era games (Plant Defense, Disaster Identification) were retired years ago, and the classic Ecosystem Building game was phased out of the standard rotation as Sea Wolf replaced it — though a few legacy invitations reportedly still include it. The 2026 change is the Sustainable Futures Lab, added in March 2026: a behavioural module that adds judgment and teamwork dimensions to a previously all-analytical test.
What's new in the 2026 Digital AssessmentHow does McKinsey Solve compare to BCG Casey and Bain SOVA?
All three MBB firms screen online, but the formats differ sharply. McKinsey Solve is interactive mini-games; BCG's Online Case ("Casey") is a single chatbot-led case of roughly 8–10 questions plus a one-minute video recommendation; Bain's SOVA is a timed aptitude battery (numerical, verbal, logical).
Compare McKinsey vs BCG vs Bain assessmentsWhat are the most common Sea Wolf mistakes?
From our own study of 312 self-reported attempts, five patterns separate passers from rejects: losing too much time on sites 1–2 (≈68%), leaving a site requirement unmet in the final selection (≈54%), re-solving each site from scratch instead of templating (≈47%), over-optimising cleaning above target (≈31%), and panic-clicking in the final 5 minutes (≈29%).
Read the 5 Sea Wolf failure patterns (n=312)How do I find out whether I passed the McKinsey Solve?
McKinsey never shows candidates a score or feedback — the only signal is an interview invitation or a rejection, delivered by email or a portal status change days to weeks later. What you can do in each branch (pass or not) is its own playbook.
what happens after the SolveWhen do you get the McKinsey Solve after applying?
Invitations typically land days to a couple of weeks after submission for candidates who clear resume screening, usually with a one-to-two-week completion window — much less warning than most expect.
the McKinsey recruiting timelineIs the McKinsey Solve hard?
Yes — but in specific, fixable places: an unfamiliar format, tight clocks, and an invisible process layer that most candidates meet for the first time on test day. Our practice data shows exactly where it bites and which game punishes a cold start the most.
what the data says about Solve difficultyWhat is a good McKinsey Solve score?
You'll never see one — McKinsey scores the assessment internally and tells candidates nothing, so the only number you'll ever get is a practice percentile. What that practice score actually tells you, and what moves it, is its own answer.
what a good McKinsey Solve score looks likeMy test is tomorrow — is it too late to prepare?
No. Most candidates in our data start inside the final 48 hours, and format familiarity is exactly the kind of gain two days can still buy: full timed runs of each game, targeted reps on your weakest one, then sleep. What to skip matters as much as what to do.
the last-48-hours planHow important is the McKinsey Solve?
It decides whether you interview. McKinsey uses your Solve performance together with your resume to choose who moves forward — clear it and you're preparing for case interviews; miss it and the application ends there, with no retake in the same cycle. McKinsey doesn't publish pass rates, but most estimates put the failure rate around 70%. The one consolation: you never see a score either way, so the only version of "how did I do" you can actually control happens before test day, not after. What the outcome looks like from the candidate side — how you find out, how long it takes — is its own guide: what happens after the assessment.
Start Practicing for McKinsey Solve Today
You now know which games you'll face, how they work, and the rules of the road. The rest is reps. Every free account includes a full practice run for each game — timed, scored, and with a performance breakdown. No credit card required.
Browse all practice simulators
Sources & further reading
- Solve — McKinsey's assessment game (official) — rules, support contacts, accommodations
- Solve FAQs (official PDF, McKinsey & Company) — timing, devices, languages, anti-cheating, technical support
- Interviewing at McKinsey (official) — where Solve sits in the recruiting process
- SolvePrep research: 5 Sea Wolf failure patterns from 312 self-reported attempts
- SolvePrep research: The State of McKinsey Solve 2026 — 26,622 simulator runs analysed
Game durations, structures and scoring mechanics beyond McKinsey's official statements are based on aggregated candidate reports current as of August 2026 and are labelled as such throughout. McKinsey may change the assessment at any time — always follow your invitation email.