Sea Wolf · sourced guide
McKinsey Solve game: make sense of Sea Wolf.
A good-looking microbe can make a bad treatment. Read the whole site brief, keep the pool rules straight, and check the trio before you submit. The worked example below shows the method.
What is known about Sea Wolf?
CORROBORATED — The fidelity register describes three contaminated ocean sites, one treatment per site and a single 30-minute clock. Reported site rules combine numerical ranges with a desired and an undesired trait. Source and context.
VERIFIED — McKinsey says each task begins with an untimed tutorial. The invitation states the length of your assessment. Read both before relying on any preparation guide’s timings. Source and context.
UNKNOWN — The sources do not establish McKinsey’s internal scoring weights or a public passing threshold. CaseStar’s practice effectiveness is a result under its own model. Source and context.
Keep classification and prospecting separate.
REPORTED — The detailed walkthrough distinguishes classifying organisms from building a treatment pool. Saved classifications are reviewed at the next site; they do not fill that site’s treatment pool. Source and context.
CORROBORATED — The source register records six initial prospects, followed by four rounds in which you retain one of three offers. The resulting pool has ten organisms; the treatment uses three. This is the reported sequence, not access to a proprietary generator. Source and context.
Practice method: before choosing an offer, ask what the pool is missing. An organism that adds a needed trait or balances an extreme value may be more useful than another individually tidy option. Compare only pools you could legally have built: alternatives from the same offer round cannot all be retained.
Check averages first, then traits.
CORROBORATED — Reported treatment rules check each attribute’s average across the selected trio, require the desired trait on at least one organism and exclude the undesired trait from the trio. Source and context.
Original CaseStar exercise
One tidy average is not the whole answer
This simplified exercise has two attributes. A must average 5–7; B must average 4–6. At least one organism must be salt-tolerant; none may be corrosive. These are original practice values and names. They do not describe a live assessment question.
| Organism | A | B | Trait |
|---|---|---|---|
| First | 4 | 2 | Salt-tolerant |
| Second | 6 | 5 | Neutral |
| Third | 8 | 8 | Neutral |
Does the trio meet this brief?
Show the working
For attribute A: (4 + 6 + 8) ÷ 3 = 6. That meets 5–7. For B: (2 + 5 + 8) ÷ 3 = 5. That meets 4–6.
The trio contains the desired trait and no forbidden trait, so it meets all the conditions in this simplified exercise. Individual values outside the ranges do not invalidate an average that is inside them.
Counterexample: give the third organism the forbidden trait. Both averages still pass, but the treatment fails that trait condition. Numerical fit alone is insufficient.
CORROBORATED — Candidate accounts describe imperfect treatments and numerical near-misses. The fidelity register allows that some sites cannot reach full effectiveness. A candidate’s belief that every combination was tried is not an exhaustive proof. Source and context.
Make each practice sitting answer a question.
- Read the brief and restate every condition in your own words.
- Keep a clean record of the pool you actually built.
- Check the entire trio against every condition before submitting.
- In review, separate a missed condition from a constraint the available pool could not satisfy.
Then try a fresh paper and see whether you catch the same mistake without a hint. CaseStar’s examples, clocks and feedback are practice choices. No practice score establishes a McKinsey result.
Common questions
How long is Sea Wolf?
CORROBORATED — the fidelity source register records one 30-minute clock across three sites. That is candidate-reported format evidence. VERIFIED — McKinsey’s September 2026 FAQ says the invitation gives your assessment length; use that invitation for your own sitting.
Should every microbe fall inside every range?
CORROBORATED — the reported numerical rule checks the average of the three selected microbes for each attribute. A microbe outside a range can still belong to a treatment whose average is inside it. CaseStar practice uses this rule.
Do saved classifications enter the treatment pool?
REPORTED — the walkthrough separates classification from prospecting: saved classifications return for review at the next site. They are not extra organisms in that site’s treatment pool. The exact live interaction has not been independently validated.
Does an imperfect site mean I have failed?
UNKNOWN — a site’s effectiveness does not establish a hiring outcome. CORROBORATED — candidate accounts include imperfect sites and discussion of numerical constraints. Do not assume that every offered pool has a perfect treatment. CaseStar feedback evaluates its own authored paper.
Is this the real assessment?
No. CaseStar uses original practice organisms, papers and feedback. UNKNOWN — McKinsey’s internal scoring and generation rules are not established by these sources. The free sample is CaseStar practice, not an assessment question.
Sources and limits
- McKinsey: Solve assessment overview and integrity rules
- McKinsey: Solve FAQ (September 2026)
- January 2025 candidate account: classification and prospects (historical timing is superseded)
- Secondary walkthrough: classification, prospecting and treatment
- March 2025 candidate account: an average outside the target range
- CaseStar: what changed in McKinsey Solve in 2026 (dated change log with confidence labels)
CaseStar (casestar.io) is an independent practice platform for consulting recruiting: timed McKinsey Solve simulators, including Red Rock, and voice case interviews. It is not affiliated with McKinsey & Company.
McKinsey publishes only a broad description of Solve. Details labelled as candidate-reported come from debriefs and can differ between candidates, regions and test versions. Worked examples on this page are original CaseStar exercises, not assessment questions. Your invitation and recruiter take precedence.