McKinsey Solve: Firm Facts, Game Map, and Process Score

Most candidates treat McKinsey Solve like a homework syllabus: memorize the games, drill the folklore answer keys, hope the ecology trivia sticks. That is the wrong bet. McKinsey built Solve to score how you work a novel constraint problem, not whether you studied reefs. After this guide you should be able to separate what the firm publishes from what candidates report, run one calorie-constraint puzzle yourself, and map each live game to a timed simulator without re-reading three other encyclopedias.

What McKinsey actually publishes

Start on the firm's own pages. Everything else is secondary.

On its Solve careers page, McKinsey describes Solve as a gamified assessment meant to showcase problem-solving ability. Performance is considered with the rest of your application and any other assessments. The same page bans AI tools, notes, screenshots, and outside help unless McKinsey has approved a disability accommodation in writing.

The firm's interviewing overview repeats the pitch for consulting roles: Solve tests natural problem-solving; McKinsey says there is "no need to prepare." Take that as positioning, not as a reason to walk in cold. It does tell you what they refuse to sell you: a study guide, a practice test, or a list of tonight's modules.

McKinsey's Game-Based Innovation Lab is clearer about purpose. In a 2023 careers blog, the firm says it wanted to measure how people think, as opposed to what they know, after years of a quantitative multiple-choice test. Solve goes to around 300,000 early-career consultant candidates per year. The same piece names Sea Wolf: selecting and evaluating microbes to clean contaminated sites so recruiters can watch analytical thinking in context.

An earlier recruiting piece, Build a reef? Save a species?, describes the original digital assessment as scenario-based problem solving designed with Imbellus and UCLA CRESST — lower stress than a traditional test, less leverage for prior business background. That is the public origin of what candidates later called Ecosystem Building.

What McKinsey does not publish on those pages: a fixed module roster for your invite, pass rates, weightings, or Redrock/SFL walkthroughs. Your invitation email and the untimed tutorial before each task are the only authorities on length and games for your sitting. Older FAQ language circulating publicly puts Solve around a 70-minute window; newer invites often show other totals. Believe the email.

The games map (and which simulators mirror them)

McKinsey names Sea Wolf in public writing. Ecosystem Building is the food-chain / habitat task candidates and early firm posts describe around reefs and species. Redrock Study and Sustainable Futures Lab (SFL / Seawolf-adjacent naming varies) are widely reported by candidates in 2025–2026 invites; McKinsey has not published module manuals for them on the Solve page. Treat module-level detail as candidate-reported unless your tutorial says otherwise.

Sibling CoachNed posts cover Redrock, Sea Wolf, and SFL in depth. This flagship piece maps the lineup and stops duplicating those walkthroughs.

Candidate reports commonly pair Redrock and Sea Wolf on shorter invites, and add SFL when the invitation clock is longer. Ecosystem Building appears less often in recent folklore than it did in 2021–2023 writeups. None of that overrides your email.

Worked Ecosystem-style puzzle: calories and populations

Invented numbers. Reproduce every line. The point is the constraint logic candidates report for Ecosystem Building — not a clone of tonight's McKinsey parameters.

You must seat four species in one habitat. Rules:

1. Every consumer needs enough calories from what it eats.
2. Prey populations cannot be driven to zero: calories taken from a prey species cannot exceed that prey's available calories (population × calories contributed per individual when eaten).
3. Every species must tolerate the habitat's temperature band (here: 12–18°C).

Habitat: temperate marsh, 12–18°C.

Step 1 — Temperature. All four ranges overlap 12–18. Habitat is legal. If any species sat outside (say a 4–9°C fish), you would drop it before touching calories.

Step 2 — Heron demand. 4 herons × 120 cal = 480 cal needed from frogs. Each frog provides 30 cal if eaten, so herons need 480 / 30 = 16 frogs worth of prey. Population is 20, so 16 frogs can be taken and 4 remain. Legal.

Step 3 — Frog demand. Remaining frogs must still feed. All 20 frogs need food before predation: 20 × 40 = 800 cal from snails. Each snail provides 8 cal, so frogs need 800 / 8 = 100 snails. Only 80 snails exist. Illegal. The heron–frog link looked fine in isolation; the snail layer breaks.

Fix (one legal rebuild). Cut herons to 2 (common move: shrink the top predator). Heron need = 2 × 120 = 240 cal → 240 / 30 = 8 frogs taken, 12 frogs remain. Frog need = 20 × 40 = 800 cal still (population is scored before or after predation depending on the rule card — candidates report both; here we feed the starting population). Still need 100 snails. Keep the failure visible: you still need either more snails or fewer frogs.

Change frogs to 15 (and herons to 2). Frog need = 15 × 40 = 600 cal → 600 / 8 = 75 snails. Snail population 80 → 5 snails remain after frog feeding. Herons take 8 frogs from 15 → 7 frogs remain. Snail need from grass: 80 × 5 = 400 cal → 400 / 2 = 200 reedgrass individuals eaten; 400 available → 200 remain. Chain closes.

Ned's takeaway from the room: candidates lose Ecosystem-style tasks by optimizing one trophic level. Build bottom-up or top-down with a full pass, then re-check every link. Pretty heron counts that starve snails score like a case recommendation that ignores capacity.

Ned's rule. Treat Solve like a partner watching your working papers: a correct final roster with thrashing clicks still looks weak, and a near-miss roster reached by disciplined constraint checks still looks hireable. Optimize the process you would defend in a debrief, not the screenshot of the board.

Process score and product score

McKinsey's public language stresses observing how candidates approach tasks, not only whether a multiple-choice bubble is right (PSG recruiting history; GBIL interview). Imbellus, the assessment company McKinsey partnered with and later folded into its game lab, described item scores as covering both product (whether an output was satisfactory) and process (how the test-taker reached it) in a 2019 Education Week interview with its founder. That is as close as the public record gets to "process and product." McKinsey does not publish weights, telemetry formulas, or cutoffs.

Practical consequences anyway:

• Opening the data, sorting, checking averages, and committing once beats frantic re-clicks that land on the same answer.
• Leaving a module unfinished to "save" another is a product miss you cannot hide.
• Interface familiarity helps process more than trivia flashcards help product. That is why simulators matter more than reading another guide.

What to practice (without duplicating the sibling posts)

1. Read the invite. Length and any named tasks beat every blog table.
2. One timed Redrock-style data pass on /mckinsey-red-rock — chart triage and clean arithmetic. Details: Redrock guide.
3. One timed Sea Wolf constraint pass on /mckinsey-sea-wolf — illegal traits first, then averages. Details: Sea Wolf guide.
4. If your invite is long enough that SFL appears in candidate reports for your office, one SFL run on /mckinsey-sfl. Details: SFL guide.
5. If the tutorial is food-chain Ecosystem, redo the calorie table above with new invented numbers until the surplus/gap check is automatic, then use Sea Wolf for timed constraint habits.

Do not prep for Plant Defense or other retired folklore modules unless your materials say they are back. After Solve is stable, move the same structured reasoning into McKinsey case interviews — Solve clears a gate; it does not replace the room.

Run this before your invite window

Block 90 minutes. Confirm your invite length. Run Sea Wolf or Redrock once under the timer — whichever your email implies. Score yourself on two columns: product (did the final answers or teams satisfy the stated rules?) and process (did you gather data before locking, or thrash?). Everything on CoachNed is open for seven days, no card; then $120 for a recruiting season or $49 a month. Start at /start if you have never done a scored case, then hit the Solve paths above.

CoachNed is not affiliated with McKinsey & Company.

Sources

• Solve, McKinsey's assessment game — purpose, integrity rules, results considered with the application. Checked 2026-09-24.
• Interviewing at McKinsey — Solve as a consulting-process step; firm "no need to prepare" language. Checked 2026-09-24.
• From apprenticeship in space to selecting microbes — how-vs-what framing, ~300,000 candidates/year, Sea Wolf description. Checked 2026-09-24.
• Build a reef? Save a species? — early digital recruiting scenario design with Imbellus. Checked 2026-09-24.
• Innovating recruiting through online gaming — observing how candidates approach tasks; permutations so no two experiences match. Checked 2026-09-24.
• Straight Up Conversation: The Woman Who's Trying to Reimagine Testing — Imbellus founder on product vs process item scores. Checked 2026-09-24.