McKinsey Solve is the online problem-solving assessment McKinsey & Company sends to many applicants before case interviews. Instead of a multiple-choice test, you work through simulated scenarios set in nature: analysing a wildlife study, building microbe treatments for polluted ocean sites and, on longer invitations, steering a sustainability project. In 2026 prep sources report invitations of about 65, 85 or 95 minutes with two or three games. McKinsey says no business knowledge is needed. Below: the current line-up, what changed, reported scoring and original practice items for each game.
Key facts
- Vendor
- McKinsey, built with Imbellus; replaced the paper PST
- Games
- Red Rock Study, Sea Wolf; Sustainable Futures Lab on some invitations
- Time
- stated in the invitation; about 65, 85 or 95 min reported
- Answer format
- drag data, number entry, report blanks, charts, ranking, single choice
- Device
- PC or Mac only; no phone or tablet; sound not needed
- Languages
- English, Spanish, Portuguese and Japanese reported
- Adaptive
- not stated; each candidate gets a generated scenario variant
- Scoring
- not published; considered alongside the rest of the application
What McKinsey Solve is (and is not)
Solve was built with Imbellus, whose scenarios generate many permutations of the same task, and replaced McKinsey's pen-and-paper test. The official FAQ says no preparation is expected and no business knowledge or gaming experience is needed. The skills are still learnable: reading dense data fast, slip-free percentage maths, filtering options against several constraints, pacing. Unlike pymetrics-style trait games, Solve has better and worse answers (see the game-based assessment overview).
The 2026 line-up: what changed and where sources disagree
McKinsey does not publish which games a given invitation contains. The picture below combines prep-site trackers from 2026. They agree on the core and disagree on dates.
| Game | Status in 2026 | Reported time |
|---|---|---|
| Red Rock Study | Core, in every reported invitation | about 35 min |
| Sea Wolf (ocean clean-up) | Core, in every reported invitation | about 30 min |
| Sustainable Futures Lab | Added on longer invitations from around March 2026; a longer version reported since late August 2026 | about 20 min (standard), about 30 min (pilot) |
| Ecosystem Building | Retired | — |
| Plant Defense | Retired | — |
Disagreements worth knowing:
- When Ecosystem Building disappeared. CaseStar (September 2026): phased out from about July 2025. RoadToOffer (October 2026): Ecosystem and Plant Defense left the standard rotation in 2023–2024 in most offices, with its own sources disagreeing. MyConsultingCoach (June 2026) still calls Ecosystem "legacy and less common". Guides quoting a 70 or 110-minute Ecosystem-based test are outdated (McKinsey's older FAQ said 70 minutes; the later one says the length is in your invitation).
- The name. "Sustainable Future Lab" and "Sustainable Futures Lab" both appear; McKinsey's public Solve page does not name it (CaseStar).
- The 95-minute invitation. CaseStar reports it since August 2026 with no fourth game; RoadToOffer attributes it to a longer Lab pilot.
- Same everywhere? No source reports office-specific game sets, but McKinsey publishes no rule on who gets which length, or who is invited at all.
Read your invitation e-mail: it states the length, and that tells you whether to expect the Lab.
How each game works and how to approach it
Red Rock Study (about 35 minutes)
Reported structure: one Study in three phases, then six short cases, all on one clock.
- Investigation. A research objective, a briefing and several exhibits (text, tables, charts). You drag the values the objective needs into a research journal. You can move back and forth here.
- Analysis. Two to four calculation questions answered from your journal: percentage change, percentage points, weighted averages, splits, ranking by growth rate. An in-game calculator is reported in some sittings.
- Report. Fill the blanks in a pre-written memo and choose and build the chart that fits the message. Entering the Report is reported to be a one-way gate: you cannot go back to the earlier phases.
- Six cases. Standalone questions with their own data: evidence selection, a percentage share, mean versus median, simple probability, a ratio after reallocation. The journal resets between cases.
Strategy.
- Read the objective first and collect only what it needs, with units noted.
- Budget the clock: a coaching example cited by RoadToOffer uses about 23 minutes for the Study and 12 for the cases.
- The Study reportedly asks for two-decimal answers while cases may not state a precision (CaseStar). Keep full precision until the last step.
- Chart choice follows the message: categories → bar; time → line; parts of a whole → pie or stacked bar.
More chart practice: numerical reasoning tests.
Sea Wolf (about 30 minutes)
Reported structure: three contaminated ocean sites, one shared clock. For each site you build a treatment of three microbes. Each microbe has numeric attributes (reported as three, on a 1–10 scale) and a trait. Each site sets a target range for each attribute's team average, one desired trait that must be present and one forbidden trait that must be absent. Difficulty rises from site to site. MyConsultingCoach describes several steps per site (sorting microbes into site pools or rejecting them, building a prospect pool, choosing the final three) and a reported split of 20 % per attribute in range plus 20 % per trait rule. CaseStar reports an effectiveness score after each site and says some sites cannot reach 100 % with any team, so 80 % is not necessarily a failure. None of this is official.
Strategy.
- Convert ranges to sums: an average of 5.0–6.0 over three microbes is a sum of 15–18.
- Filter hard rules first: strike forbidden-trait microbes, then anchor the team on a desired-trait microbe.
- Use extremes as balancers, but recheck every attribute after each swap.
- Pace three sites: about 10 minutes each, less on the first, more on the last.
Sustainable Futures Lab (about 20 or 30 minutes)
Reported standard format (about 20 minutes): one storyline about a sustainability project; a drag-to-rank task of four priorities, then twelve single-choice decisions with long answer options, each followed by an outcome page. It is text-based, with little or no calculation, and you cannot go back. Earlier choices are reported to shape later situations. The newer pilot version (about 30 minutes, reported from August 2026) has you lead a small research team over several simulated days: requesting information, assigning people, resolving a team member's dilemma and rating the team.
Strategy.
- Read the goal and constraints on the intro screens closely; later decisions are judged against them.
- Prefer options that gather missing information, protect the main objective and keep stakeholders on board over dramatic, unilateral or scope-expanding ones.
- Stay consistent with your own ranking, and do not read outcome pages as right/wrong signals.
Retired games, briefly
- Ecosystem Building: pick species so a food chain stays stable, meeting each species' food and location rules. The skill (filtering many options against several constraints) transfers to Sea Wolf.
- Plant Defense: protect a plant from waves of invaders with limited resources, adjusting placements each round.
Original practice items
All items are original, with invented data, mimicking reported structures, not real Solve scenarios.
Red Rock Study: meadow butterfly survey
Objective: "Assess how butterfly numbers changed across four alpine meadows."
| Meadow | Count, year 1 | Count, year 2 |
|---|---|---|
| North | 1,200 | 1,500 |
| East | 800 | 720 |
| South | 2,000 | 2,300 |
| West | 500 | 480 |
A side note says wing-span samples were taken in two meadows: North, 40 butterflies, average 31 mm; South, 60 butterflies, average 36 mm.
Q1. Which meadow had the highest percentage growth?
Answer: North (+25 %). - North: 300 / 1,200 = +25 %. South: 300 / 2,000 = +15 %. East: −80 / 800 = −10 %. West: −20 / 500 = −4 %. - Trap: North and South grew by the same absolute number; the smaller base wins on percentage.
Q2. By how many percentage points did North's share of the total count change? Round to two decimals.
Answer: +3.33 percentage points. - Totals: year 1 = 4,500; year 2 = 5,000. - North's share: 1,200 / 4,500 = 26.67 % → 1,500 / 5,000 = 30.00 %. - 30.00 − 26.67 = 3.33 points.
Q3. What is the average wing span across both samples combined?
Answer: 34.0 mm. - (40 × 31 + 60 × 36) / 100 = (1,240 + 2,160) / 100 = 34.0 mm. - Trap: the simple average of 31 and 36 (33.5 mm) ignores sample sizes.
Q4. The memo says: "Year-2 numbers rose in two meadows and fell in two." Which chart best supports it?
Answer: a clustered bar chart, one pair of bars per meadow; a pie chart cannot show change.
Red Rock cases: two standalone items
Case A. Seven orders took 2, 3, 3, 4, 5, 6 and 26 days to deliver. Which figure better describes a typical order, the mean or the median, and what are they?
Answer: the median, 4 days. The mean is 49 / 7 = 7 days, pulled up by one 26-day outlier; only one order out of seven took longer than 6 days.
Case B. In a survey, 30 % of visitors bought a map, 20 % bought a guidebook and 8 % bought both. What share bought at least one?
Answer: 42 %. P(map or guide) = 30 + 20 − 8 = 42 %. Trap: adding 30 + 20 = 50 % counts the 8 % twice.
Sea Wolf: build one treatment
Site requirements (team averages, inclusive): Salinity tolerance 5.0–6.0; Growth rate 6.0–7.0; Toxin uptake 7.0–8.0. Desired trait: Biofilm. Forbidden trait: Spore release.
| Microbe | Salinity | Growth | Uptake | Trait |
|---|---|---|---|---|
| M1 | 5 | 8 | 7 | Biofilm |
| M2 | 3 | 6 | 9 | Nitrogen fixer |
| M3 | 6 | 4 | 6 | Spore release |
| M4 | 7 | 5 | 4 | Night-active |
| M5 | 4 | 6 | 8 | Biofilm |
| M6 | 2 | 8 | 4 | Night-active |
| M7 | 6 | 7 | 6 | Cold-hardy |
| M8 | 9 | 3 | 9 | Biofilm |
Question: which three microbes meet every requirement?
Answer: M1, M5 and M7 (the only valid team). - Convert to sums for three microbes: Salinity 15–18, Growth 18–21, Uptake 21–24. - Strike M3 (forbidden trait). Every team needs one of M1, M5 or M8 (Biofilm). - M1 + M5 + M7: Salinity 5 + 4 + 6 = 15; Growth 8 + 6 + 7 = 21; Uptake 7 + 8 + 6 = 21. All in range, Biofilm present, no Spore release. - Why M8 fails: its Growth of 3 means the other two need at least 15 Growth between them. Only M1 + M7 (15), M1 + M6 (16) and M6 + M7 (15) reach that. M8 + M1 + M7 has Salinity 20 (too high); M8 + M1 + M6 has Uptake 20 and M8 + M6 + M7 has Uptake 19 (both too low). - The trap: M1 + M3 + M5 also fits every number (15, 18, 21) but carries the forbidden trait. Filter traits first.
Sustainable Futures Lab: a judgement item
Setup: you lead a coastal wetland restoration with a fixed budget and a two-year timeline. Rank four priorities: biodiversity outcome, local community support, staying on budget, finishing on time.
Decision: six months in, local fishers say the planned access closures will cut their income and threaten to campaign against the project. Which do you do?
- A. Proceed as planned; the closures are needed for the biodiversity goal.
- B. Pause all work until the fishers agree.
- C. Meet the fishers' representatives, ask what access they need, and test whether closures can be phased by season while protecting breeding areas.
- D. Double the budget for public relations.
A reasoned choice: C (McKinsey publishes no key; this illustrates the pattern). C gathers information, protects the objective and keeps a key stakeholder on board. A risks support, B gives up the timeline, D spends budget without fixing the cause.
Traps and how to avoid them
- Losing the tutorial-to-clock handover. Each tutorial is untimed, but the timed task is reported to start abruptly. Read the last tutorial screen before clicking on.
- Entering the Red Rock Report too early. Once inside, you cannot fetch a missing value. Check your journal against the objective first.
- Percent versus percentage points, and simple versus weighted averages (see Q2 and Q3 above).
- Using outside tools. McKinsey's integrity rules prohibit outside aids and recording any part of the assessment, and it may ask for a follow-up task to confirm your work.
How to prepare: a 5-day plan
| Day | Focus | Time |
|---|---|---|
| 1 | Read your invitation (length, deadline); run the system check; read this guide | 45 min |
| 2 | Red Rock skills: 20 chart and table questions on percentage change, percentage points and weighted averages, untimed then timed | 60 min |
| 3 | Sea Wolf skills: build constraint puzzles like the one above (8 options, 3 attributes, 2 trait rules); aim for under 4 minutes each | 45 min |
| 4 | A full timed Red Rock-style run: collect, calculate, report, six short cases in 35 minutes | 60 min |
| 5 | Two Lab-style judgement scenarios, then rest; sit Solve early in the deadline window | 30 min |
Setup. PC or Mac, stable connection, notifications off, and a slot about 20 minutes longer than the stated time. If something breaks, contact McKinsey support at once: the FAQ says it cannot make exceptions for a reset you did not report. Request accommodations (such as extra time) through your recruiter beforehand.
After Solve, the next stage is usually case interviews: see the case interview guide.
Sources
- McKinsey, "Solve, McKinsey's assessment game" (careers page)
- McKinsey, Solve FAQ 2024 (device, tutorials, no business knowledge needed)
- McKinsey, Solve FAQ August 2025 (length in invitation, support and resets, accommodations)
- McKinsey, older Problem Solving Game FAQ (70 minutes)
- McKinsey careers blog, "Innovating recruiting through online gaming" (Imbellus, permutations)
- CaseStar, "What changed in McKinsey Solve in 2026" (updated October 2026; line-up, times, dates, Sea Wolf effectiveness, Red Rock gates, languages, validity): https://www.casestar.io/guides/mckinsey-solve/what-changed-2026
- RoadToOffer, McKinsey Solve guide (updated October 2026; timeline, Red Rock phases and pacing, Lab pilot, calculator, retakes): https://www.roadtooffer.com/blog/mckinsey-solve-guide
- RoadToOffer, "Is McKinsey Solve proctored?" and "Does everyone get McKinsey Solve?" (October 2026; integrity rules, follow-up task, candidate volume): https://www.roadtooffer.com/blog/is-mckinsey-solve-proctored, https://www.roadtooffer.com/blog/does-everyone-get-mckinsey-solve
- MyConsultingCoach, McKinsey Solve guide (updated June 2026; Sea Wolf mechanics and reported scoring split, Lab format, product and process): https://www.myconsultingcoach.com/mckinsey-problem-solving-game
MSc em Psicologia, 7 anos a fazer testes.
Última atualização: Oct 1, 2026