McKinsey Sea Wolf Practice Simulation

The complete McKinsey Sea Wolf guide

McKinsey publicly describes Solve as a gamified assessment built from a library of tasks and variations, but it does not document every task in detail. Candidates commonly use Sea Wolf to describe a microbe-matching task where you choose a small set of microbes for multiple sites under time pressure.

The useful mental model is simple: Sea Wolf is a constraint-matching game, not a biology test. You are matching numeric ranges and trait requirements while staying organized enough to finish the task cleanly.

Across candidate write-ups the framing varies a little. Some people describe choosing microbes, others describe placing species into a small ecosystem or set of sites. The labels are not the point. The decision is the same: pick a set that satisfies each site's constraints and keeps the overall system stable. If your invitation or tutorial uses different wording than a guide you read online, trust the tutorial in front of you.

Who gets Sea Wolf, and when?

Sea Wolf shows up inside McKinsey Solve, which most applicants complete early in the process, typically after the application and before first-round interviews. The most commonly reported 2026 format pairs Sea Wolf with Redrock Study in a roughly 65-minute sitting, but McKinsey varies the module mix by role, level, and region and does not publish a fixed line-up. Your invitation email and the in-test instructions screen are the only reliable statement of which games and time limit you will get. The McKinsey Solve guide covers the full assessment flow, the variants, and how Solve feeds into the interview rounds.

What are the steps in the Sea Wolf game flow?

The Sea Wolf flow is a short, repeating loop: read the tutorial, read each site's constraints, build a valid set, confirm stability, then move to the next site under a running timer. Candidates do not all see an identical task, so treat the sequence below as the recurring shape rather than a fixed script.

1. Read the tutorial once, carefully. McKinsey opens the module with an interactive walkthrough. This is the only place the exact rules for your version are guaranteed to be correct, so do not skim it to save thirty seconds you will lose later.
2. Open the first site and read its constraints. Each site gives you target attribute ranges and required or forbidden traits. Write these down before you touch any options.
3. Translate ranges into a target you can check fast. If you are selecting a fixed number of options, convert each attribute range into a target sum (see the worked example below) so you can validate a candidate set with quick arithmetic instead of eyeballing.
4. Filter out clear conflicts first. Remove options that obviously break a forbidden-trait rule or that cannot fit the range no matter what you pair them with. This shrinks the search space immediately.
5. Build a set that satisfies every constraint at once. Aim for a group that fits the numeric ranges and includes the required trait. A single odd-looking option can still be correct if it balances the group.
6. Confirm stability before you commit. Many candidate reports describe a check or feedback step where an unbalanced or invalid set is flagged. Verify your set holds together rather than submitting the first thing that looks close.
7. Move to the next site and repeat the loop. Reuse the same process. Consistency across sites matters more than perfecting any single one.
8. Watch the clock the whole time. Candidates commonly report roughly a 30-minute task, but the only timer that matters is the one in your assessment. Budget time per site and move on when a set is clearly workable.

What ecosystem and attribute mechanics do you track?

You track three things at once: each option's numeric attributes, its traits, and how the chosen set holds together as a whole. Sea Wolf is consistently described as an ecosystem or system-balancing task, so individual picks only matter in the context of the full set they create.

The mechanic candidates most often underweight is the set-level view. It is tempting to grade each option on its own and keep only the "good" ones. That is the slow, error-prone path. The faster path is to ask what the full set needs to average out to, then assemble options that hit that target together.

How does Sea Wolf scoring work per site?

McKinsey does not publish a public Sea Wolf score formula, so the honest answer is: nobody outside McKinsey knows the exact per-site weights. What is defensible from McKinsey's own statements and consistent candidate reports is the mechanism, not a percentage.

• Valid, stable sets pass; invalid or unstable ones do not. A set that satisfies the range and trait constraints and keeps the ecosystem balanced is treated as a success for that site. Breaking a constraint or producing an unstable configuration is what costs you.
• Wrong moves deduct rather than simply "not adding." Because the task is constraint-based, the realistic model is that violations work against your site result. The safest assumption is that a clearly invalid choice hurts more than a conservative valid one, so do not gamble on a borderline set when a clean one is available.
• Your process is measured, not only your final answer. McKinsey states publicly that Solve evaluates how you reason, not just the end result. Erratic, undo-heavy, or scattered behavior can read as weaker process even when you eventually land a valid set. Working in a calm, structured way is part of the signal.
• Do not chase a "perfect" site. Once a set clearly satisfies every constraint, the marginal value of polishing it further is low and the time cost is high. Across the whole module, finishing every site with valid sets beats over-optimizing one.

Because the exact internal weighting is private, treat the takeaway as behavioral rather than numeric: get each site to a clean, valid, stable set efficiently, avoid constraint violations, and keep your process orderly.

Worked example

Suppose a site needs:

• Energy average 3-5
• Adhesion average 6-8
• Speed average 2-4
• at least one Heat-Resistant trait

Three microbes:

Now check the averages:

• Energy = (4 + 3 + 5) / 3 = 4
• Adhesion = (7 + 6 + 7) / 3 = 6.67
• Speed = (3 + 4 + 2) / 3 = 3

That set works because the group fits the ranges, even though the page is not asking you to find three individually perfect microbes. That is the practical point of Sea Wolf prep.

The four-step method that survives any version

Use a four-step loop on every site: write the rules, eliminate hard failures, compare survivors, validate the set.

1. Write every condition before sorting. Treat the site rules as a checklist, not background copy.
2. Eliminate hard failures first. One failed mandatory condition is enough to reject an option. Do not average a failure away because the other attributes look strong.
3. Compare every survivor consistently. Read attributes in the same order each time. Switching order creates avoidable omissions.
4. Validate the final set as a set. A microbe can pass individually and still be the wrong treatment partner. Recheck the full combination before submitting.

This method is intentionally less glamorous than a solver. That is the point: it remains usable when the labels, values, or interface change.

How to prepare for Sea Wolf

The best prep is boring in a good way:

• Run one untimed attempt on the simulation above to learn the interaction and vocabulary.
• Do quick average and range-to-sum drills — timed case interview math practice, data interpretation reps, and scored mental math drills make the arithmetic reflexive.
• Practice filtering trait conflicts without hesitation — structure practice builds the same explicit constraint handling, the discipline covered in structured thinking for case interviews.
• Repeat under a clock and keep the same checklist.
• Review by error type, not only by score: missed rule, wrong comparison, or weak combination.

A timed self-simulation also works on paper: invent a few sites with attribute ranges and a required or forbidden trait, then assemble valid sets against a clock. This drills the exact muscle (range-to-sum, conflict filtering, set-level checking) the real module tests.

If your invite does not list module names, do not panic about memorizing labels. Focus on clean arithmetic, calm reasoning, and reading the tutorial carefully once the task opens.

How to review a Sea Wolf practice attempt

Do not review only the final treatment. Reconstruct the first point where your reasoning departed from the rules. A wrong final set can begin with a misread range, an incorrect reject decision, or an individually valid option that combined badly with its partners.

Then replay the logic without changing the rule order. Improvement comes from making the checking routine dependable, not from discovering a new trick after every attempt.

How Sea Wolf fits into the rest of McKinsey prep

Sea Wolf concentrates on multi-constraint selection. Redrock concentrates on evidence retention, exhibit interpretation, and calculation — the Redrock study guide goes phase by phase and the Red Rock practice study runs the loop free. SFL concentrates on prioritization and stakeholder judgment — the SFL practice simulation covers it. All three reward structured problem solving, but practicing only one does not reproduce the switching cost of a longer Solve invitation. Bain runs a comparable digital screen you can rehearse in the Bain SOVA simulator, and the MBB assessment simulator roundup maps every free option in one place.

Solve is only one gate. After that, you still need to handle live case interviews and the Personal Experience Interview.

Sources (checked June 17, 2026)

• McKinsey Solve page: https://www.mckinsey.com/careers/mckinsey-digital-assessment
• McKinsey Problem Solving Game FAQ PDF: https://www.mckinsey.com/~/media/McKinsey/Careers%20REDESIGN/Interviewing/Main/McKinsey-Problem-Solving-Game-FAQ-v2.pdf
• McKinsey careers blog on Solve development: https://www.mckinsey.com/careers/meet-our-people/careers-blog/mck-problem-solving-game-team
• IGotAnOffer Solve guide: https://igotanoffer.com/blogs/mckinsey-case-interview-blog/mckinsey-problem-solving-game

McKinsey and Solve are trademarks of McKinsey & Company. Road to Offer is independent and is not endorsed by McKinsey.