McKinsey Solve Score: How It's Judged and What Passing Means

TL;DR: McKinsey does not publish a Solve pass mark, does not send you a score, and the bar is not fixed — it moves by office, cohort and cycle. What is consistent is what gets read: not only your final answers, but the process that produced them — how you explored, how you used your time, how consistent your decisions were. That distinction is the single most useful thing to understand before you sit it.

Why there is no number to aim for

Three things make "what score do I need for McKinsey Solve?" unanswerable in the terms candidates want.

The result is not a simple total. Solve is a simulation, not a quiz. The tasks generate a rich behavioural record — the order you did things in, how long you spent, what you revisited, how stable your approach stayed — alongside the outcomes you produced. Collapsing that into "you got 62%" is not how the assessment is designed to be read.

The bar is relative and it moves. You are judged against the applicant pool that applies where you applied, in the cycle you applied. The same performance can clear the line for one office and one intake and not another. Solve is also one input beside your application and, later, your case interviews — its weight is not uniform either.

Nothing is published, and nothing is disclosed to you. McKinsey does not release cut scores, and candidates receive an advance-or-not outcome rather than a report. Anyone quoting you a precise Solve threshold is repeating a rumour, usually one that started as a guess.

So the useful question is not "what is the pass mark" but "what is being measured, and which parts of it can I improve in the time I have."

Process and outcome are both read

This is the part worth internalising. In a simulation, how you got there is data.

Outcome is the obvious half: did your ecosystem survive, did your defence hold, did your analysis land on defensible conclusions.

Process is the half candidates neglect: whether you gathered information before committing, whether your decisions followed a consistent logic, whether you managed the clock or ran out of it, whether you flailed and reversed yourself under pressure.

Two candidates can reach a similar outcome and be read very differently — one methodically, one by thrashing until something worked. And a candidate whose outcome was mediocre but whose approach was disciplined and coherent is not automatically behind. Consultancies are hiring for structured thinking under uncertainty, and the simulation is built to expose it.

The practical consequence: do not abandon a sensible approach halfway because it feels slow. Erratic strategy-switching reads worse than a steady approach that runs slightly short of a perfect result.

What the tasks demand

The Solve battery has varied over time — the ecosystem and Red Rock tasks are the ones most current candidates describe, with plant defence appearing in earlier versions. The game-by-game detail lives in the McKinsey Solve game guide, the ecosystem building guide, the Red Rock study guide and the plant defense guide.

Across all of them, the same underlying capacities do the work:

• Holding a complex state in mind while you evaluate options — many interacting constraints, none of which you can write down comfortably.
• Reading rules from a system rather than from instructions — inferring how something behaves by observing it.
• Numerical judgement at speed, usually estimation and comparison rather than precise arithmetic.
• Spatial and relational reasoning — routes, dependencies, what connects to what.
• Time discipline under a clock that does not forgive exploration you never converted into a decision.

What you can actually train

You cannot rehearse Solve itself — the scenarios are proprietary and change. You can absolutely train the capacities above, and unlike the pass mark, your progress here is measurable.

• Pathfinder puzzle — route and constraint reasoning, the closest analogue to planning under interacting rules.
• N-back memory test — holding and updating a shifting mental state, which is what a live simulation demands continuously.
• Mental math — fast numerical judgement without a calculator.
• Corsi block visual memory — spatial working memory for layouts and positions.

Each run scores against fixed norms and returns a percentile, so you can find your weakest capacity in about fifteen minutes and spend the rest of your preparation there rather than on whichever task you enjoy most.

How to use the last week

• Baseline everything once, cold. Then rank your percentiles and pick the bottom two.
• Practise finishing. Set a timer and force a decision before it runs out. Running out of clock mid-exploration is one of the most common self-inflicted Solve outcomes.
• Rehearse a default approach, so that under pressure you have a method to fall back on instead of improvising.
• Read the in-assessment tutorials properly. They are the only authoritative description of the version you are taking; every second-hand account, including this one, is a cycle or two behind.
• Do not chase a number. There isn't one. Chase coverage of the skills and a process you can hold steady.

For the wider context of how simulation assessments fit modern screening, see the complete guide to game-based assessments.

FAQ

What is a good McKinsey Solve score? There is no published pass mark and you never receive a score. Your performance is judged relative to the applicant pool for that office and cycle, alongside the rest of your application, so no fixed number exists to aim for.

Does McKinsey tell you your Solve result? No. Candidates receive an advance-or-not outcome rather than a score report or feedback breakdown.

Is Solve scored on the outcome or the process? Both. The simulation records how you worked — exploration, time use, consistency of approach — not only what you produced. A disciplined process with an imperfect result is not automatically behind a good result reached chaotically.

Can you fail Solve but still progress? Solve is one input among several, and its weight varies. It is a genuine screen, so a weak performance hurts, but it is not always the sole decider.

What percentile do I need? Unknowable, and not a useful target. The bar moves by office, cohort and cycle, and none of it is published. Work on the underlying skills and your time discipline instead.

Can you practise for McKinsey Solve? Not the scenarios themselves — they are proprietary and change between cycles. The capacities they load (spatial reasoning, working memory, numerical judgement, decision-making under a clock) do respond to practice, and that is where preparation time pays.

GamePrep is an independent practice platform and is not affiliated with McKinsey & Company. McKinsey does not publish Solve's scoring or cut scores and its assessment changes between cycles — your invitation and the in-assessment tutorials are the only authoritative sources. We provide related skill practice only — no leaked questions or answers.