Can You Cheat McKinsey Solve? What the Rules Actually Say
The most widely recommended Sea Wolf shortcut is a rules breach, and the site recommending it says so on a different page.
There is no answer key to find, because no two candidates get the same parameters. And McKinsey reserves the right to ask you, at random, to sit an additional task in person or to explain the logic you used.
There is no answer key, because there is no shared test
Most assessments can in principle be leaked, because most assessments hand every candidate the same items from a finite pool. Solve does not. McKinsey states that the assessment “uses advanced AI methods to ensure that there is ample variation in each scenario, and it is designed so that no two individuals have the same parameters and combinations of data”. Its own 2020 account of building the thing puts it more plainly: “thousands of permutations of the same game, so no two applicants have the same experience”.
That is not marketing. The peer-reviewed paper describing the original design, written by the Imbellus team with a McKinsey co-author, states it as a design goal: every user receives a unique task instance. Which means the most-traded contraband in the Solve market, a Sea Wolf answer set, describes three sites whose attribute ranges are not your ranges and a microbe pool that is not your pool. Working from it is slower than working from the screen.
The most-recommended shortcut is against the rules, and the rules name the penalty
This is the part worth reading even if you had no intention of cheating, because a candidate can be walked into a breach by advice that sounds like preparation.
A major preparation site's Sea Wolf guide states that “McKinsey allows you to use any tools during the Solve test”, that “you are not monitored through your webcam, and you can use a second computer, pen, paper, or spreadsheets”, and recommends building an Excel solver in advance, claiming it cuts calculation time from over five minutes to under two per site.
McKinsey's own FAQ says candidates are not permitted to “use applications, websites, AI tools, or pre-prepared notes”, and lists the consequences of breaching the rules as “exclusion from consideration, rescinding of an offer, restrictions on future applications, or legal action”. A spreadsheet built in advance is an application, and it is a pre-prepared note.
The same preparation site contradicts itself. Its Red Rock guide warns that the firm “prohibits using outside applications, websites, AI tools, screenshots, or help from others” and that “breaking those rules can cost you the offer entirely”. Two pages of one site, opposite advice, and only one of them agrees with the firm running the assessment.
Why a solver would not save you much anyway
Sea Wolf rewards a judgement that no spreadsheet makes for you. The scored rule is that the mean of each attribute across your three chosen microbes must land inside the site's range, that one of the three must carry the desired trait, and that none may carry the undesired one. Enumerating the combinations is the easy half. The hard half is upstream: choosing which characteristics to reveal, deciding which microbes to keep and which to set aside for a site whose requirements you have not seen yet, and recognising when a pool simply cannot produce a perfect treatment so that 80% is the right answer and the remaining minutes belong to the next site.
Candidates report finishing sites at 60%, 80% and 100% in the same sitting, and coaches describe that spread as unremarkable. A solver optimises the final selection screen, which is the one step that was already reversible and already the least likely to be where the site was lost.
What to do instead
Solve is unusual among the assessments on this site in that its designers explicitly set out to measure how you work rather than only what you submit. The published research behind it describes deriving scores from telemetry and contrasts that with tests that “focus on product scores (i.e., correctness) not the process of how (i.e., strategy) a user got there”. Nobody outside McKinsey knows how much that is still true of the current modules, and anyone who tells you how to perform for the process score is guessing.
Which leaves the advice that is honest because it also produces a better answer. Read the research objective before you collect anything. Reuse the calculator's logged results rather than retyping rounded versions, because the Redrock questions chain and one bad intermediate poisons everything after it. Work from the mean in Sea Wolf rather than filtering microbe by microbe. Change an earlier allocation when the game shows you information that makes it wrong. And in the mini-cases, take the two minute limit seriously: the six cases share one clock, so a perfect answer that costs you the next two questions is a loss dressed as care.