McKinsey Sustainable Futures Lab: the two versions, and how to prepare for each

Which version you'll get

The Sustainable Futures Lab is the third game, after Red Rock Study and Sea Wolf. Not every candidate gets it, and the ones who do don't all get the same version. The only reliable signal is the total length on your invitation email.

Where sources disagree: one prep site that says it sat the redesigned version in September 2026 describes the 13-question format as retired outright. Two others, working from candidate reports, say both versions are still in use, that the invitation length tells you which, and that the 95-minute version looks like a pilot. A fourth, checked on 24 September, still described only the 20-minute game and said nobody had reported a new game on 95-minute invitations. McKinsey hasn't published anything about either version, so we can't settle it. What we can say: if your email says 85 minutes, prepare for the original; if it says 95, prepare for the redesign; and read your on-screen instructions on the day, because they beat any guide, including this one. The full breakdown of all three lengths is in what your invitation length means.

Be careful with older advice generally. Anything written before 2026 doesn't mention this game at all, and some of it still describes the retired Ecosystem game as current. It isn't; it was dropped from the live test in 2025.

The original format: 13 prompts, one evolving story

This is the version on 85-minute invitations and the one our simulator recreates. You lead an environmental research team through a single continuous scenario in about 20 minutes:

1. One priority-ranking prompt: a drag-and-drop ordering of four early actions, from what you'd do first to last. It sets your baseline.
2. Twelve decisions: each presents a new complication (a team member pushes back, a stakeholder changes their mind, new data contradicts the plan) and two to four options. The options are full paragraphs, usually starting "You suggest..." or "You recommend...", not short labels. Your choice triggers a consequence screen and reshapes what comes next. Two candidates who diverge at decision three are living different stories by decision eight.

There's no calculator, no data exhibits and no scratchpad. It's reading and judgement. The scenario theme rotates between candidates: wetland restoration, air quality, coastal habitat recovery and similar.

Five stakeholders (your team, a funder, a partner organisation, the public and leadership) react to what you do. Their relationships with you shift as the scenario unfolds, and later decisions land differently depending on the goodwill you've built or burned.

How it's scored: consistency, not correctness

McKinsey doesn't publish how this game is scored, so treat every scoring claim, including ours, as a model rather than a fact. What the question design points at is whether your judgement hangs together across five things: prioritisation, deciding under uncertainty, reading messy information, balancing trade-offs, and handling the people involved. Prep sites describe the same five, and they agree on one more point: extreme options (halt everything, ignore it, wait indefinitely for perfect data) read badly, and balanced, proactive ones read well.

Flip-flopping is the failure mode. If you play the cautious consensus-builder for four decisions and then lurch into unilateral risk-taking because an option looked clever, the inconsistency is the signal, not either choice on its own. Our simulator scores exactly that: consistency across the run, plus where the five relationships end up.

What a decision actually looks like

To make "consistency" concrete, picture one branching decision. Your field team reports early, unverified data suggesting a restoration site is failing. You're given four options:

1. Halt the project and reallocate the budget immediately.
2. Commission a two-week independent study before deciding.
3. Push forward on the original plan and monitor.
4. Tell the funder and let them make the call.

None of these is "correct." What matters is that it fits the leader you've already been. If your priority ranking put "evidence quality" at the top and you commissioned studies in two earlier decisions, option 2 is consistent. If you've been the decisive, act-on-partial-signal leader all game, option 1 or 3 fits, and suddenly deferring to the funder in option 4 would read as a break in pattern. The assessment isn't grading the choice; it's watching whether your choices trace one coherent philosophy across all twelve.

How the five stakeholders work

Five relationships run underneath the whole scenario: typically your team, a funder, a partner organization, the public, and leadership. Each starts at a neutral level and moves up or down with your decisions. The mechanic that catches people out: a relationship you've neglected can quietly close off options later. Burn the funder early and a decision that needed their backing may not be available when you reach it. You can't keep all five happy (the game is built so that pleasing one often costs you another), so the goal isn't maximum goodwill everywhere, it's spending goodwill in a pattern that matches your stated priorities.

The redesigned format: three days, four stages (candidate-reported, as of 10 October 2026)

Everything in this section comes from candidate reports and from prep sites. A candidate on a consulting forum first described a new 30-minute version with "a much more complex interface" on 11 September 2026, and candidates in the US and the UK reported it again on 4 to 7 October. Two prep sites that describe sitting it agree on the structure below, so we're confident in the shape. Whether it has replaced the original or is still a pilot, and how it's scored, is not confirmed: McKinsey hasn't said anything about it publicly.

The new version keeps the name and the environmental setting but changes what you do. Instead of answering a sequence of questions, you run a four-person research team for three simulated project days, in about 30 minutes. A tutorial with the clock paused comes first, then a short onboarding task where you order the briefing questions you want answered. Each day then opens with a goal (day one is about what the researchers need, day two about what the workstations need, day three about balancing both) and runs the same four stages:

1. Explore. You get a small number of requests per day, and each buys one question to a researcher or a workstation. Not everyone is available every day. Everything you learn is saved to a notes panel that carries across all three days. The scarcity is the test: ask what the day's goal needs, not what you're curious about.
2. Assign. Drag each researcher onto a workstation. Every move asks you for a reason from a short list, and the reason is recorded along with the move. The board reshuffles each morning, so keeping a placement that was working is an active decision.
3. Support. Researchers escalate problems one at a time: pressure to act on incomplete evidence, how to deliver unwelcome findings, that kind of thing. You pick between paragraph-length responses. When several arrive together, which one you handle first is part of it.
4. Reflect. Rate how each researcher is doing when the day closes. "I don't know" is one of the options, and both sources say it scores better than a confident wrong read.

What carries over from the original: it's still judgement, not arithmetic; it still rewards consistency across the run; and it still watches whether you commit to a course or hand the problem back. What's new: information is scarce and you choose what to spend it on, you have to justify delegation, and you're asked to remember what people told you. One of the two sources also reports a required feedback survey after the third game, including a question about how many hours you spent preparing.

SolveForge's simulator covers the original format only. We haven't built the redesigned version yet. If your invitation says 95 minutes, use the guide above for the judgement habits (pick a posture, read everything, don't swing to extremes) and expect the interface to look nothing like the 13-question version.

The mistakes that quietly tank the score

• Choosing the "smartest-looking" option each time. Optimizing every decision in isolation produces a jagged, inconsistent path, exactly what the scoring penalizes. Pick a posture, then choose the option that fits it, even when a flashier one is on the table.
• Ignoring the consequence text. After each decision the game tells you how stakeholders shifted. That's the board state, not flavor text. Skipping it means walking into later decisions blind.
• Trying to keep every stakeholder happy. Impossible by design. Consistent trade-offs beat universal approval every time.
• Contradicting your own priority ranking. The drag-and-drop that opens the game is a public commitment. If you ranked stakeholder trust first and then repeatedly sacrifice it for speed, that gap is the inconsistency being measured.

How to prepare

• Pick a posture in the first two prompts, then hold it. Decide what kind of leader you're being (transparent, evidence-first, stakeholder-balancing, whatever feels natural) and let that filter every later choice. Your priority ranking in prompt one is a commitment; treat it as one.
• Read the consequence text. Each decision's outcome tells you how stakeholders moved. That's not flavor. It's the state of the board.
• Don't optimize for being liked. Keeping every stakeholder happy at all times isn't possible by design. Trade-offs are the point; consistent trade-offs are the score.
• Run it more than once. The mechanics (branching, stakeholder tracking, pacing across 12 decisions) stop costing you attention by the second or third run, which frees you to actually think. That's the entire case for practicing it at all.
• If you're on the 95-minute version, prepare the habits, not the screens. Decide in advance how you weigh incomplete information and when you act versus wait. Practise attaching a reason to every delegation. Read slowly: the options are paragraphs and the differences between them are the whole test.

Keep going

Read the full McKinsey Solve guide, or go deeper on the other two games: the Red Rock Study walkthrough and the Sea Wolf strategy guide. When you're ready, play the Futures Lab.