"McKinsey Solve Game: How It Works & the Skills to Train"

McKinsey screens a huge share of its consulting applicants with Solve — a gamified assessment (you may know it as the "Problem Solving Game" or the "ecosystem game") completed online before any case interview. Unlike a classic aptitude test, Solve drops you into scenarios and watches how you think, not just what you answer. As of 2026 the scenario set evolves regularly (ecosystem building, disaster/disease management and other variants have rotated through), so memorising walkthroughs is a losing strategy — training the underlying skills is not.

What Solve actually is

You typically get about 70 minutes across two or more scenarios. The classic ecosystem scenario asks you to build a stable food chain in a mountain or reef environment: pick species whose calorie needs, terrain and eating relationships form a self-sustaining system. Other scenarios involve identifying patterns, managing spreading threats, or allocating scarce resources under uncertainty.

Two layers are scored:

• Product score — did your solution work (did the ecosystem survive)?
• Process score — how you got there: how you explored data, how systematically you tested constraints, how you handled time. McKinsey has been open that process telemetry matters, which is why "someone told me the answer" does not transfer.

There is no published pass mark; your performance feeds into the interview-invite decision alongside your application (the benchmark logic works like every other game-based screen — see can you fail a game assessment).

The trainable core skills

Solve is deliberately unfamiliar, but it leans on a stable set of cognitive muscles:

Constraint-based planning. The ecosystem puzzle is a chain of dependencies you must order correctly before committing — the exact skill the pathfinder puzzle drills: plan the full route first, then execute.

Working memory under load. You juggle calorie numbers, terrain rules and species relationships simultaneously. Train capacity with the n-back test — the closest lab analogue to "hold several changing facts while reasoning" — and Corsi block for the spatial layer.

Fast, accurate arithmetic. Calorie provided vs. calorie needed comparisons are simple maths done many times under a clock. Ten minutes a day on mental math makes those checks automatic, freeing attention for the actual logic.

Calm time allocation. Most failed attempts are time-management failures: perfectionism on the first half, panic on the second. Every timed practice game teaches the same meta-skill — decide, commit, move.

How to prepare (one week)

1. Days 1–2: Read McKinsey's official Solve materials (they publish a candidate guide) so zero minutes are spent decoding the interface. Play the four games above cold; note weak spots.
2. Days 3–5: Daily 20-minute sessions on your two weakest skills. For ecosystem-style logic, practice building dependency chains on paper: pick any system (a supply chain, a food web) and map what depends on what.
3. Day 6: Full timed dress rehearsal of the practice games back-to-back — the point is pacing under fatigue.
4. Day 7: Rest. Solve rewards fresh working memory more than one more cram session.

FAQ

Can I retake Solve? Generally only with a new application after the standard waiting period (historically around two years for reapplication). Treat your attempt as final.

Do I need to finish everything? No — partial but systematic beats complete but sloppy, because process is scored.

Is it beatable by memorising walkthroughs? Scenario details rotate and telemetry watches your process. Skill preparation transfers; script-following does not.

GamePrep is an independent practice platform, not affiliated with McKinsey & Company. Assessment formats change — verify details against your official candidate instructions.