McKinsey Solve in 2026: Red Rock, Sustainable Future Lab, and What Actually Improves Your Score

For three years Solve was a two-module assessment built around Ecosystem Building (Red Rock) and Plant Defense. As of late 2025 a third module — Sustainable Future Lab — has rolled out broadly, with the analysis section restructured around it. Candidates report seeing different module combinations depending on office and role, and the cycle is still settling.

What that means in practice: nobody has a clean playbook anymore, and most of the prep guides circulating in 2026 were written for the two-module version. The mechanics have shifted. The signal McKinsey is trying to read has not.

What Solve is actually measuring

Solve is not a problem-solving aptitude test. It is a gamified rehearsal of the working habits a first-year consultant needs on Day 1: read incomplete data, make a decision under a clock, revisit your decision when new information arrives, do not get rattled when the game does something unexpected.

The scoring is opaque on purpose. McKinsey has been clear that Solve measures both process (what you click, in what order, how often you change your mind) and product (whether your answers are right). Process is the half candidates underweight. A correct answer reached via 40 minutes of frantic guessing scores worse than a slightly-wrong answer reached via a clean, ordered exploration.

The three modules, briefly

1. Ecosystem (Red Rock)

The classic module: build a food chain on a fictional reserve under species, terrain, and calorie constraints. The visible challenge is the puzzle; the invisible challenge is order of operations. Strong candidates establish the binding constraint first (calories or terrain, almost never both at once) and let everything else fall out of that.

The most common failure mode is iterating species choices randomly because the interface invites it. The simulator rewards not doing that.

2. Sustainable Future Lab

The newer module pushes you toward a sustainability-themed scenario with multi-step trade-offs across emissions, economics, and stakeholder constraints. Tasks are more narrative than Red Rock — closer to a written case in feel, with branching consequences.

Candidates report it rewarding scenario-thinking over optimization. The bar is whether you can hold multiple objectives in mind simultaneously and read which ones actually move under your control, rather than which ones the interface lets you click.

3. The redistributed analysis section

The data-table reading and quantitative-reasoning content that used to be threaded into Plant Defense now appears as standalone analysis tasks in many sittings. These are the tasks where candidates lose the most points the most avoidably — not because they cannot do the math, but because the tables are dense, the clock is short, and the wrong column is one click away from the right one.

What actually improves a Solve score

Stop optimizing your gameplay; optimize your habits

The single largest score driver is reducing avoidable error rate on the analysis sections. That is a habit problem, not an intelligence problem. The fix is the same kind of deliberate practice that improves case math: short, timed reps with a feedback loop that flags errors as they happen, not after.

Read the constraint structure before you click anything

In Red Rock specifically, candidates who win the first three minutes by mapping calorie deltas, terrain dependencies, and species rules before placing anything almost always outscore candidates who jump in. The interface is built to invite jumping in. Resist it.

Treat narrative modules like a case, not a game

Sustainable Future Lab is the module where business-case muscle transfers most cleanly. Identify the objective, name the binding trade-off, choose a path, and check whether the path is consistent with what the prompt actually rewards. If you train cases with a real interviewer, the same instincts apply here — they just happen on a screen.

Where AI-led practice helps and where it does not

AI interviewers and AI-graded simulators are useful for the analysis section and for the case-like reasoning in Sustainable Future Lab — anywhere the task is read, decide, defend, refine. They are not useful for Red Rock specifically, which is a closed simulator with constraints that no AI tool can faithfully replicate.

The honest split:

• Use AI-led practice to drill data-table reading under time pressure with a rubric that catches sloppy column-picks and unit errors. The feedback loop is the point.
• Use AI-led case practice to build the objective-and-trade-off reflex you need in Sustainable Future Lab. Cases and SFL share the same underlying structure-then-decide muscle.
• Do not use AI for Red Rock itself. Use McKinsey's official sample, the in-product tutorials, and one or two timed solo runs to get the feel.

A two-week plan that respects the new shape

• Days 1–3. Read the McKinsey official materials. Run the Red Rock tutorial once untimed, once timed. Do not grind it.
• Days 4–8. Daily 25-minute analysis drills on dense tables with a feedback loop. Track which kinds of errors (unit, column, rounding, sign) you make most.
• Days 9–12. Three to four live cases — ideally narrative-heavy ones — to rehearse the structure-then-decide loop you will need in SFL.
• Days 13–14. One full Red Rock run, one mock SFL-style narrative case, and one analysis set under full time pressure. Sleep.

What good looks like

A candidate scoring well on the new Solve does not look like someone who has played 40 hours of ecosystem-builder games. They look like someone with steady working habits — reads prompts twice, names the constraint before acting, catches their own column errors, does not double back nervously, and finishes with time on the clock. Those habits do not come from playing the simulator. They come from practice that catches the things the simulator will not.