Most candidates who lose points on McKinsey's judgment module do not lack climate literacy. They ignore the ranking rule the screen already printed, then invent a better-sounding order from instinct.
SFL — the name candidates use for Sustainable Futures Lab — is a prioritization and situational-judgment task inside Solve. McKinsey has not published a module-by-module breakdown. Treat every format detail below as candidate-reported unless labeled official. After this page you should separate firm fact from folklore, run one original ranking item to completion, and know which two habits recover most avoidable misses. Practising with a simulator is covered in a sibling post; this one is the component itself.
What McKinsey publishes, and what it does not
Start with the firm's own pages. On the Solve careers page, McKinsey describes Solve as a gamified assessment of problem-solving ability, says there is no need to prepare, and states that Solve results are considered with the rest of the application. The same page bans AI tools, pre-prepared notes, recording or screenshots, and outside assistance, except for written disability accommodations. That integrity language is official. Module names are not.
McKinsey's Solve FAQ PDF (2024) adds operational facts that still matter in 2026: the assessment runs on PC or Mac only; length is in your invitation email; each task starts with an untimed tutorial; no prior business or gaming knowledge is required; scenario parameters vary by candidate. The FAQ does not name Redrock, Sea Wolf, or SFL, and it does not publish scoring weights or a pass mark.
For Sea Wolf specifically, McKinsey's game-based innovation lab has been more open. A GBIL blog post describes Sea Wolf as selecting and evaluating microbes to clean contaminated sites, and notes Solve is given to roughly 300,000 early-career consultant candidates per year. That is the level of detail the firm will put in print. SFL does not get a comparable official write-up.
| Claim | Status as of 2026-09-24 |
|---|---|
| Solve is a gamified problem-solving assessment | Official (careers page) |
| No need to prepare; results join the rest of the file | Official |
| No AI, notes, recording, or assistance during the live sitting | Official |
| Length and tasks come from the invitation plus tutorials | Official (FAQ) |
| Module mix and per-module clocks (e.g. ~20 min SFL) | Candidate-reported |
| Ranking plus sequential scenario questions inside SFL | Candidate-reported |
| Exact item counts, pass thresholds, scoring weights | Folklore — discount |
CoachNed is independent practice material and is not affiliated with McKinsey.
Format candidates report seeing
Invitation length is the only early signal most candidates get. Reports cluster around two sitting lengths: about 65 minutes for two modules (commonly Redrock and Sea Wolf) and about 85 minutes when a third judgment module is added. McKinsey does not confirm that arithmetic. Your invitation and the on-screen tutorial are the authorities.
Inside the third module, candidate reports describe a short sustainability-framed scenario — coastal restoration, air quality, flood risk, or similar — then a sequence of decisions. A pattern that shows up often enough to prepare for, without treating it as a guarantee:
| Piece | Candidate-reported shape | What it is testing |
|---|---|---|
| Opening ranking | Drag-and-drop or ordered list of initiatives against stated criteria | Whether you apply the rule before instinct |
| Follow-on MCQs | ~10–12 situational items as new information arrives | Trade-offs, dependencies, stakeholder handling |
| Clock | Often cited near 20 minutes for the module alone | Pacing under light pressure, not a climate exam |
| Computation load | Light relative to Redrock | Judgment and sequencing, not exhibit math |
None of that is in McKinsey's public FAQ. If your screen shows a different structure — including a longer team-allocation variant some candidates have described in 2026 — follow the screen. Format drifts; the skills do not.
What is scored, as far as anyone can tell
McKinsey does not publish an SFL scorecard. What you can defend from adjacent evidence:
- Solve is scored as a process-aware assessment. The firm built Solve with Imbellus to measure how people think, not what they already know. That design intent shows up in the FAQ's emphasis on unique parameter sets and in the GBIL description of analytical thinking in context. Expect process signals (how you navigate, whether you use the given criteria) to matter alongside the final click.
- Ranking items with explicit criteria have a defensible order. If the screen ranks criteria — impact per dollar, then reversibility, then stakeholder readiness — a ranking that ignores that order is wrong even when it feels wise. This is the recoverable part of the module.
- Hard constraints beat soft ratios. When initiative B cannot start until A finishes, putting B first is a process failure, not a creative trade-off.
- Stakeholder pushback without new data is a loyalty-to-criteria test. Reordering to soothe someone who brings no new numbers rewards lobbying. A calm restatement of the criteria plus a concrete next step is the consulting habit the item is probing.
Anything that claims "pass at 70 percent" or "13 questions weighted X/Y" without a McKinsey citation is folklore. Ignore it.
Ned's rule. If the screen states a ranking rule, compute it on paper before you touch the options. The ninety seconds you spend on the arithmetic are cheaper than the points you lose when your gut overrides the criteria you just read.
Worked example: HarborCrest estuary program
Invented company, invented figures — enough to reproduce the habit, not a leaked item.
Brief. HarborCrest runs a five-initiative estuary program with a year-one capital budget of $12m. The screen states the ranking rule in this order: (1) lives protected per million dollars in year one, (2) reversibility (higher is better), (3) dependency (must not start before its blocker finishes).
| Initiative | Year-1 cost ($m) | Lives protected (year 1) | Reversibility (1–5) | Dependency |
|---|---|---|---|---|
| Early warning network | 1.5 | 420 | 5 | None |
| Pump station upgrades | 3.0 | 540 | 4 | None |
| Temporary seawall patch | 2.5 | 300 | 3 | None |
| Zoning buyouts (phase 1) | 4.0 | 280 | 2 | Needs early warning live |
| Wetlands restore (pilot) | 5.0 | 90 | 2 | None |
Step 1 — compute the primary ratio. Lives per $m: Early warning 420/1.5 = 280; Pumps 540/3.0 = 180; Seawall 300/2.5 = 120; Buyouts 280/4.0 = 70; Wetlands 90/5.0 = 18.
Step 2 — draft order by ratio. Early warning, pumps, seawall, buyouts, wetlands.
Step 3 — apply the dependency. Buyouts cannot precede early warning. In this draft they already do not. If your gut had put buyouts second "because they feel structural," you would have failed the constraint even with a pretty story.
Step 4 — use reversibility only on ties. There are no ties on the primary ratio. Do not reshuffle wetlands above seawall because wetlands sound greener. The rule already ranked impact per dollar first.
Pass ranking. 1 Early warning, 2 Pump upgrades, 3 Seawall patch, 4 Zoning buyouts, 5 Wetlands restore.
Instinct trap. Ranking wetlands first "because it is the sustainable futures lab" is the exact failure mode. The framing is a wrapper. The scoring rule is the table.
Stakeholder follow-on (shape, not a script). The wetlands lead objects and offers no new data. Acknowledge the concern, show the three ratios and the stated criteria order, keep the ranking, and propose a concrete next step (for example, a six-month wetlands pilot if year-one flood metrics clear a named threshold). Capitulating without new evidence, or stalling until a later meeting with no decision, are the weak answers.
Common errors that look like judgment
Ranking from values, not from the printed rule. Sustainability framing invites identity. The item rewards mechanical application of stated criteria. Write the criteria down before you look at the options.
Treating dependencies as soft preferences. A blocked initiative in position one is wrong regardless of its ratio. Scan for "cannot begin until" before you submit.
Reordering after lobby pressure with no new facts. That is the same failure as changing a case recommendation because the interviewer frowned. Hold the process; invite new data.
Burning time on domain knowledge you do not need. McKinsey's FAQ says no prior business knowledge is required. Climate vocabulary on the screen is flavor. Every number you need is on the page.
Preparing only Redrock math or only Sea Wolf filters. Those modules train different muscles. SFL is light on computation and heavy on criteria loyalty. Switching costs in an 85-minute sitting are real; practise the switch, not just one game.
Memorizing folklore item counts. If your invitation is 65 minutes, you may not see SFL at all. Prep the skill; confirm the roster from the email.
How SFL sits beside the other Solve pieces
Sea Wolf is constraint filtering then optimization — McKinsey has described the microbe premise publicly. Redrock is exhibit reading and calculation across investigation, analysis, and report. SFL, when present, is judgment under a stated rule set. The umbrella process sits before interviewer-led cases; see the McKinsey case interview guide for what Solve is a gate into.
If your invite includes the optimization module, the Sea Wolf guide covers filter-then-optimize. For the data module, use the Redrock study guide. Do not let SFL prep displace those if your sitting is two modules only.
Drill this before your invitation window
One concrete assignment, forty-five minutes:
- Copy the HarborCrest table onto paper. Cover the pass ranking. Re-rank from the criteria alone under an eight-minute clock.
- Invent one stakeholder objection with no new data and write a three-sentence response: acknowledge, show criteria, next step.
- Run CoachNed's SFL-style practice at /mckinsey-sfl once. It is original prioritization and stakeholder practice, not a replica of McKinsey's item bank. Everything on CoachNed is open for seven days with no card; then $120 for a recruiting season or $49 a month.
For the simulator-focused walkthrough and how to use those rounds, read the sibling SFL simulator post when it is live. For timed criteria discipline without a scenario wrapper, /drills/structure and /drills/brainstorming keep the same habit warm.
Frequently asked questions
What does SFL stand for in McKinsey Solve?
Candidates use SFL for Sustainable Futures Lab, a judgment module reported inside some longer Solve sittings. McKinsey's public Solve pages do not name that module. Treat the acronym as candidate vocabulary, not an official product label.
How long is the McKinsey SFL module?
McKinsey does not publish a per-module clock. Candidate reports often place SFL near twenty minutes inside an ~85-minute three-module invitation, versus ~65 minutes when only two modules appear. Confirm duration from your invitation email and the on-screen tutorial.
Do all McKinsey candidates get SFL?
No. Module mix varies by office, role, and cycle. Many candidates report Redrock and Sea Wolf only. Absence of SFL on your screen is normal, not a signal about your file.
Is SFL a personality test or are there right answers?
Candidate reports describe prioritization items with defensible correct orderings driven by stated criteria, plus situational items that reward process-consistent choices. It is not a values quiz. Practising mechanical application of criteria helps; rehearsing a personality does not.
Does SFL require sustainability or climate expertise?
No. McKinsey's FAQ states Solve does not require prior business knowledge. The environmental framing is a scenario wrapper. Numbers, criteria, and constraints appear on screen.
Sources
- Solve, McKinsey's assessment game — official Solve description, integrity rules, and how results are used. Checked 2026-09-24.
- Solve FAQ PDF (2024) — invitation length, untimed tutorials, PC/Mac, no prior knowledge required, unique parameters. Checked 2026-09-24.
- From apprenticeship in space to selecting microbes — GBIL / Sea Wolf premise and Solve volume context. Checked 2026-09-24.
- Interviewing at McKinsey — where Solve sits relative to PEI and the problem-solving interview. Checked 2026-09-24.