The McKinsey Sea Wolf game is a 30-minute decision-making task inside the McKinsey Solve assessment. You work through three contaminated ocean sites and build one microbe treatment for each site. The core task is simple to state: average three microbes into the target attribute ranges, include the desired trait, and keep the undesired trait out of the treatment.
This guide explains the four candidate-facing steps, what the averaging math looks like, how the 20% penalty model works, and how a full Sea Wolf site feels in practice. The walkthrough follows a Density / Energy / Size site from the profiling choices through the final treatment.
Key takeaways
- The Sea Wolf game is a 30-minute decision-making task inside the McKinsey Solve assessment. It has four candidate-facing steps: choose two profiling characteristics, categorize microbes, build the prospect pool, and select the final treatment.
- You select microbes by averaging their attribute values to fit target ranges, while making sure at least one microbe carries a desired trait and none carry an undesired trait.
- Sea Wolf is also called "the microbe game," and it replaced the older Ocean Cleanup (sometimes called "ocean treatment") game during the Solve rollout.
- The visible treatment rule is a 20% effectiveness penalty for each missed condition, so your practical goal is to minimize total penalties across the three sites.
- The most useful practice is a full timed run through all four steps, supported by solver work so you can test microbe combinations quickly and learn which averages, traits, or prospect choices change the result.
What is the McKinsey Sea Wolf game?
The Sea Wolf game, sometimes called the McKinsey microbe game or just "the sea wolf," is the optimization task inside McKinsey Solve. McKinsey describes Solve on its official digital assessment page as a game-based exercise used during the screening stage of the application.
In Sea Wolf, you are given contaminated ocean sites and a set of microbes with numerical attributes and traits. For each site, you need to choose a treatment of three microbes. The treatment works best when the average Density, Energy, and Size values land inside the target ranges, at least one selected microbe has the desired trait, and the undesired trait stays out.
Sea Wolf is the successor to McKinsey's earlier Ocean Cleanup game (also referred to in candidate forums as "ocean treatment"). The scenario changed names, but the core reasoning stayed similar: read the constraints, compare microbes quickly, and make a constrained treatment decision under time pressure.
The numbers and microbe names rotate between runs. The structure stays consistent: three sites, four candidate-facing steps per site, one 30-minute timer for the Sea Wolf game.
Who gets the Sea Wolf game, and when?
Current candidate reports show Sea Wolf appearing in active McKinsey Solve formats. The McKinsey Solve FAQ updated August 2025 says the total length of your Solve assessment is outlined in your invitation email, so treat that email as the source for your exact game mix and timing.
- 65-minute Solve. You receive Redrock and Sea Wolf only.
- 85-minute Solve. You receive Redrock, Sea Wolf, and the Sustainable Future Lab.
The Sea Wolf timer itself is the same in current simulations and candidate reports: 30 minutes for Sea Wolf alone. The other games run on their own clocks, so finishing Sea Wolf early will not add time to Redrock or Sustainable Future Lab.
If your invitation is for the 65-minute version, this article and the Redrock Island guide cover the two games you need to prepare. If you have the 85-minute invitation, add the Sustainable Future Lab guide as the third game.
Each game has its own timer. You cannot carry minutes between Redrock, Sea Wolf, or Sustainable Future Lab. Finishing Sea Wolf with extra time is good, but the spare minutes will not be added to the next game.
How the McKinsey Sea Wolf game is structured
Each of the three sites runs through the same four candidate-facing steps. The timing column below is a practical training range, so use it as a starting target and adjust once you have taken a few full runs.
| Step | What you do | Suggested time per site | Why it matters |
|---|---|---|---|
| 1. Choose profiling characteristics | Read the current site's ranges and traits, then pick exactly two characteristics to profile | 1-2 min | Shows which constraints you notice first and may influence the prospect pool you receive later |
| 2. Categorize microbes | Sort 10 microbes as keep, save for next, or discard | 2-3 min | Tests fast constraint reading and, where shown, next-site planning |
| 3. Select prospects | Start with 6 prospect-pool microbes, then pick 1 of 3 across 4 selection rounds | 4-5 min | Builds the 10-microbe pool available for the final treatment |
| 4. Select the treatment | Choose 3 microbes from the prospect pool | 1-2 min | Determines the final treatment averages and trait penalties |
Those four steps sit inside two larger blocks:
- Profiling block: choose two profiling characteristics, then categorize the 10 microbes. The microbes you categorize here do not become the prospect pool. This block is about how you read the site and how you handle keep / save / discard decisions.
- Treatment block: build the prospect pool, then choose the final three-microbe treatment. This is where most of the heavy math and final penalty risk appears.
A few interface details matter in practice:
- The first screen gives the current site information: three target attribute ranges, one desired trait, and one undesired trait.
- The profiling-characteristics choice can be two attributes, two traits, or a mix. The 10 microbes you categorize next stay the same; the profile choice might influence the initial prospect pool you receive later. It also gives McKinsey signal on how you prioritize constraints.
- The categorization screen shows 10 microbes with their three attribute values and one trait. Sites 1 and 2 also show a next-site trait or attribute preview at this point, which helps you decide whether a borderline microbe should be saved for the next site.
- The prospect pool starts with 6 microbes and grows to 10 as you choose one microbe from each of four groups of three.
- The treatment is exactly 3 microbes drawn from the 10-microbe prospect pool. Their attributes are averaged, and the resulting averages need to land inside the three target ranges.
McKinsey gives you one 30-minute Sea Wolf timer covering all three sites. A practical rhythm is roughly 8-10 minutes per site, with a small buffer for reading and submission. If Site 1 takes 15 minutes, Sites 2 and 3 will feel compressed. Practice full 30-minute runs until you can keep the site rhythm without rushing the prospect and treatment steps.
What math is on the Sea Wolf game?
The arithmetic on Sea Wolf is light. The hard part is using it quickly while you compare several possible three-microbe treatments.
| Skill | Where it shows up | What to practice |
|---|---|---|
| Three-number averages | Treatment selection | Add the three values for Density, Energy, and Size, then divide each total by 3 |
| Range checks | Treatment selection | Compare each average with the site's minimum and maximum for that attribute |
| Edge-range bounds | Extreme targets like 1-2 or 8-10 | Spot when one value is so far away that the other two microbes cannot rescue the average |
| Compensation logic | Prospect selection and treatment selection | Judge microbes as part of a trio, because a value outside the range can be useful when it balances the other two values |
| Penalty counting | Final treatment review | Count each missed attribute range and each trait miss as a separate treatment penalty |
During practice, the Sea Wolf Excel solver is built for this exact calculation. Enter the available microbes and target ranges, then test three-microbe combinations quickly to see which trio gives the best range and trait fit.
Edge ranges decide many penalties. A target range of 1–2 only allows averages between 1.0 and 2.0. To hit 2.0 with three microbes, the highest single value you can include is 4, because . A value of 5 would require the other two microbes to average below 0.5, which cannot happen on a 1–10 scale.
How should you think about Sea Wolf scoring?
The practical scoring rule shown in the Sea Wolf experience is treatment effectiveness: each unmet condition reduces the treatment's effectiveness by 20%. For practice, use that rule directly. Build the treatment that leaves you with the fewest total misses.
A treatment loses 20% effectiveness for each of the following:
- An attribute average that falls outside its target range (this can happen on any of the three averaged attributes).
- The undesired trait appearing on at least one of the three selected microbes.
- The desired trait being absent from all three selected microbes.
Some sites give you a constrained pool where a penalty-free treatment is unavailable. When that happens, choose the trio with the smallest available penalty count and move on. Chasing a perfect treatment after the pool is set will burn time you need for the next site.
What this means in practice:
- Count total penalties. One missed range and one missed trait both cost 20%, so the better treatment is the one with fewer total misses.
- Treat prospect selection and treatment selection as the heavy steps. The profiling block matters, but the four prospect rounds and final three-microbe treatment usually decide the biggest score swings because that is where the math and final trait checks come together.
McKinsey reviews Solve results alongside the rest of your application. The Solve FAQ states the result is one input into the screening decision, not a standalone pass mark. A weaker Sea Wolf run is reviewed in the context of the full application.
A walkthrough from a real Sea Wolf simulation
The fastest way to understand Sea Wolf is to see one site played in full. The example below uses Site 1 from a PSG Cracked Sea Wolf simulation, built to match the structure and decision flow candidates report from the official assessment. The numbers, microbe names, and traits rotate between runs, but the decision pattern is the same.
Site information: what you read before choosing profiles
Before you choose the two profiling characteristics, Site 1 gives you these target ranges and traits.
| Attribute / Trait | Target |
|---|---|
| Density | 6 – 8 |
| Energy | 2 – 4 |
| Size | 8 – 10 |
| Desired trait | Light Sensitive |
| Undesired trait | Heat Resistant |
Two of the three ranges live near the edges of the 1–10 scale: Energy at the low end (2–4) and Size at the high end (8–10). That is the first thing to register. Edge ranges are unforgiving on averages, so the prospect pool needs to give you microbes that can actually combine to those edges.
The desired trait is Light Sensitive. The undesired trait is Heat Resistant. In practice, that means the final three-microbe treatment must include at least one Light Sensitive microbe and zero Heat Resistant microbes. Any breach is a 20% penalty.
Profiling: which two characteristics to "program"
Before you see the 10 microbes, you pick two profiling characteristics. For Site 1, Size (8–10) is the first characteristic to consider because it is the most extreme range. A Size value that falls too low is hard to rescue with averaging.
For the second profile, Energy (2–4) and Light Sensitive are both reasonable. Energy is another non-central range, so profiling it helps you protect the low side of the treatment. Light Sensitive is also strong because the final treatment needs at least one microbe with that trait. The useful habit is to profile the constraints that will be hardest to recover later, rather than picking the easiest-looking labels.
Profiling: keep, save, or discard the 10 microbes
The simulation now shows 10 microbes with three attribute values and one trait each. This is also where the next-site preview appears. In this example, the preview points to Density for Site 2, so high-Density microbes can matter even when they are weak for Site 1.
| Next site | Attribute or Trait | Attribute range | Site 1 implications |
|---|---|---|---|
| Site 2 | Density | 8-10 | A microbe with Density 8, 9, or 10 may be worth saving for Site 2, even if another attribute makes it weak for Site 1. |
The job is to sort the 10 microbes in seconds. These 10 categorized microbes do not become the prospect pool for your final treatment. Treat this as a profiling task: show that you can identify current-site fits, useful next-site saves, and microbes that are too costly to carry.
The three calls worth getting right are:
- Keep microbes that fit the current site cleanly. Cado Vibrio is a good example: its Density 8 / Energy 3 / Size 8 values sit inside the Site 1 ranges.
- Send forward microbes that are poor for Site 1 but useful for the previewed Site 2 need. The next-site preview points at Density 8-10, so high-Density microbes can still matter.
- Reject microbes whose numbers cannot be averaged back into range. A desired trait helps, but the values still need a workable path to the current-site averages.
Worked through Site 1's microbe list, three calls illustrate the pattern:
- Cado Vibrio (Density 8, Energy 3, Size 8, Hydrophilic): keep. Its three attributes already fit Site 1's ranges.
- Lano Fungor (Density 10, Energy 10, Size 5, Light Sensitive): send to Site 2. It is weak for Site 1 because Energy is too high, but Density 10 fits the Site 2 preview.
- Saro Proto (Density 7, Energy 2, Size 1, Light Sensitive): reject. The desired trait is useful, but Size 1 makes the final Site 1 average too hard to rescue.
- Lior Alga (Density 8, Energy 4, Size 8, Heat Resistant): send to Site 2. It carries Site 1's undesired trait, so it is a poor current-site keep, but Density 8 fits the Site 2 preview.
The keep/save/discard step matters because it tests fast profiling judgment and next-site planning. The final treatment will come from a separate prospect pool, so remember that a microbe you keep here is still outside the three-microbe treatment decision.
Think in three-microbe averages. One weak value can sometimes be offset by two stronger values. Check the edge math before you decide: for a Size 8–10 target, a Size = 1 microbe makes the final average impossible to rescue. The trait label alone is never enough.
Prospect Pool: pick 1 of 3, four times
The prospect pool starts with 6 pre-loaded microbes selected by the system. You then build it up to 10 by choosing one microbe from each of four groups of three. This is the first step that directly controls the treatment options you will have at the end.
Look at the prospect-pool screenshot below before reading the four rounds. The six pre-loaded microbes are already part of your final pool, so each new pick should complement what is missing there. In this Site 1 example, the starting pool already contains useful high-Size options, but it also carries awkward traits and several high-Energy values. That is why the round choices below prioritize low Energy, high Size, and at least one usable Light Sensitive option.
In the computed Site 1 solution, one optimal prospect path selects Beryx Virus, Vornis Agaric, Glyx Volvox, and Kyra Alga across the four rounds. Several useful prospects will stay unused at final treatment. Your job is to keep enough low-Energy, high-Size options with clean trait coverage so the final trio can avoid every penalty.
Use the table below as a quick read. Values are Density / Energy / Size.
| Round | Options | Pick | Why |
|---|---|---|---|
| 1 | Thero Mucor 10 / 2 / 2 Beryx Virus 8 / 3 / 8 Omra Bacil 9 / 7 / 1 |
Beryx Virus | It fits all three Site 1 ranges directly. The other two create a Size or Energy problem. |
| 2 | Kyra Phage 9 / 5 / 1 Dano Ciliate 1 / 5 / 1 Vornis Agaric 8 / 2 / 10 |
Vornis Agaric | It gives you exactly what this site needs: good Density, low Energy, and high Size. |
| 3 | Glyx Volvox 4 / 6 / 3 Myxa Plasma 4 / 9 / 4 Moro Loxa 5 / 5 / 3 |
Glyx Volvox | None of the three is a clean final-treatment microbe. Glyx is the least costly pool addition because it avoids the undesired trait and keeps Energy lower than Myxa. |
| 4 | Prax Diatom 10 / 9 / 8 Kyra Alga 4 / 5 / 8 Lior Diatom 1 / 1 / 1 |
Kyra Alga | It supplies Light Sensitive and keeps Size high. Prax carries Heat Resistant, and Lior pulls every average toward 1. |
The pattern is the same across all four rounds: judge the candidate as part of a future trio. Sometimes a standalone value outside the range is still useful if it balances the rest of your pool. Sometimes an in-range value is a trap because the other attributes or trait make the final average worse.
Reason from the final trio. For a target range like 8–10, no combination of three microbes can hit the average if any one value is below 4. But a value just below the target range can still be useful when the other two microbes are high enough to pull the average back in.
Final Treatment: pick 3 microbes that beat every penalty
By the time you reach the final-treatment screen, the math should be straightforward. You have 10 microbes in the pool. You need three whose averages fit the three target ranges, with at least one Light Sensitive microbe and no Heat Resistant microbe.
For Site 1, the computed clean treatment uses:
| Microbe | Density | Energy | Size | Trait |
|---|---|---|---|---|
| Beryx Virus | 8 | 3 | 8 | Aerobic |
| Vornis Agaric | 8 | 2 | 10 | Aerobic |
| Kyra Alga | 4 | 5 | 8 | Light Sensitive |
Here is the averaging math:
- Density average → In range (target 6–8)
- Energy average → In range (target 2–4)
- Size average → In range (target 8–10)
The trait check also clears: Kyra Alga supplies Light Sensitive, and none of the three microbes is Heat Resistant. That gives Site 1 a 100% treatment effectiveness score in the simulation.
A nearby wrong choice shows why the earlier filtering matters. If you swap in Saro Proto because it has Light Sensitive, the trio of Beryx Virus, Kyra Alga, and Saro Proto gives you:
- Density average → In range (target 6–8)
- Energy average → In range (target 2–4)
- Size average → Out of range (target 8–10)
Density and Energy fit, but Size costs a 20% penalty. The desired trait is useful, but the failed average still counts.
That is the heart of Sea Wolf: the trio of three-microbe averages, the desired trait, and the undesired trait must all clear at the same time. Strong runs come from the prospect-pool choices and final-treatment math working together.
How Sea Wolf differs from Redrock and Sustainable Future Lab
| Game | Timer | What it tests | What practice should target |
|---|---|---|---|
| Redrock Study | 35 min | Arithmetic, chart selection, report communication | Fast exhibit reading, percentage math, ratios, chart choice |
| Sea Wolf | 30 min | Constraint-based optimization with averaging rules | Keep/save/discard speed, pool selection, three-microbe averaging |
| Sustainable Future Lab | 20 min | Scenario judgment across team and stakeholder cues | Reading scenarios, weighing trade-offs, consistent decisions |
Extra time does not carry between games, so practice each game on its own clock. If your invitation includes all three, use a practice plan that covers Redrock, Sea Wolf, and Sustainable Future Lab separately.
Common Sea Wolf mistakes
These mistakes usually hurt more than the averaging itself.
| Mistake | Why it happens | Better move |
|---|---|---|
| Letting the undesired trait distract you | The microbe looks numerically strong | In categorization, save it only when the next-site preview gives a clear reason. In treatment selection, leave it out of the final trio. |
| Profiling weak characteristics | The easiest label feels safest | Profile the hardest constraint to recover later: usually an edge range or a required trait. |
| Missing the next-site preview | The categorization screen feels self-contained | Use the preview during microbe categorization to decide between "save for next" and "discard" on borderline microbes. |
| Judging prospects one microbe at a time | A standalone value looks in range | Think as if you are already choosing a three-microbe treatment. A high or low value can be exactly right when the rest of the pool needs it for averaging. |
How to prepare for the Sea Wolf game
Most candidates start preparing two to four weeks before the test. The plan below assumes that window. If you have less time, keep Phase 1 short and spend most of your time on full timed practice plus solver work.
Phase 1: Learn the format
Before any timed practice, get the structure into long-term memory:
- Three contaminated sites; four candidate-facing steps per site.
- One 30-minute timer covering all three sites.
- Two profiling characteristics, 10 microbes to categorize, 6 starting prospects plus 4 prospect-round picks, and 3 microbes in the final treatment.
- Each missed condition (range, desired trait, undesired trait) costs 20% of treatment effectiveness.
- Use the tutorial, Help panel, and Key/Legend once before timed practice so the icons and instructions do not cost you time during the live run.
Phase 2: Practice with realistic simulations and the solver
This is the highest-value practice block. Run realistic Sea Wolf-style simulations, then use the solver to test the available microbe combinations and understand why one treatment beats another. The detailed feedback helps you identify the mistake pattern: weak profile choices, slow keep/save/discard calls, poor prospect-pool decisions, or missed three-microbe averages.
Start with the free Sea Wolf simulation for an easy first rep. For deeper practice, the Sea Wolf simulation pack includes multiple full Sea Wolf runs across different attribute and trait combinations so you can build pattern recognition across sites. Use the Sea Wolf Excel solver alongside those reps to search treatment combinations, compare borderline trios, and build speed on the averaging logic.
Per session:
- Run a full three-site simulation under the 30-minute clock.
- Note the site where you lost the most time and the phase that caused it.
- Replay only that phase before moving to the next simulation.
Phase 3: Rehearse to time
In the last few days, switch from learning to consolidation:
- Do one full timed run every day or two.
- Keep solution review short; the final stretch is for timing and consistency.
- If your invitation includes Redrock and Sustainable Future Lab, give each its own timed rep on the same day to simulate the real session.
Frequently asked questions
What is the McKinsey Sea Wolf game? A 30-minute decision-making task inside the McKinsey Solve assessment. You clean up three contaminated ocean sites by selecting microbes whose attribute averages fit each site's target ranges, with one desired trait present and one undesired trait absent.
Is Sea Wolf the same as the Microbe game, Ocean Cleanup, or Ocean Treatment? Yes, in practice. Candidates often use "the microbe game" and "ocean treatment" as informal names for Sea Wolf. McKinsey's earlier Ocean Cleanup game was the predecessor and was retired during the Solve rollout. The mechanics carried over; the naming and interface changed.
How long is the Sea Wolf game? Sea Wolf itself is treated as a 30-minute task in current simulations and candidate reports. For the full Solve assessment length, use your invitation email; McKinsey's public FAQ says that is where your total timing is outlined.
How is the Sea Wolf game scored? The visible treatment rule is a 20% effectiveness reduction for each missed condition in the final treatment: an out-of-range averaged attribute, the undesired trait being present, or the desired trait being absent.
Are there leaked Sea Wolf answers? No. Each run uses different microbes, attribute values, and trait combinations, so memorized answers do not transfer. Practice runs train the decision pattern, which is why simulation volume helps and answer keys do not.
What math do I need for Sea Wolf? Three-microbe averaging on a 1–10 scale, applied across three attributes per site. The arithmetic is light; the difficulty is doing it cleanly under time pressure across many candidate trios.
Can I use a real calculator or Excel during the game? Use the Sea Wolf Excel solver as an active practice tool. Load the target ranges and available microbes, test treatment combinations quickly, and compare which three-microbe trio gives the strongest fit across attributes and traits. The goal is to build the combination logic until the timed game feels much faster.
Is Sea Wolf harder than Redrock? Different. Redrock rewards arithmetic speed and chart literacy; Sea Wolf rewards constraint-based pattern recognition. Most candidates find one harder than the other based on which type of work feels more natural. From PSG Cracked simulator runs, Sea Wolf tends to be the game candidates retry more often.
What to do next
Start with the free Sea Wolf simulation and run it under the 30-minute clock, even if the first attempt feels messy. Read the feedback, then write down the one decision type that cost you the most: profiling, categorization, prospect selection, or final treatment.
For the next rep, use the Sea Wolf Excel solver to test the treatment combinations you missed, then move to the Sea Wolf simulation pack for full 30-minute reps across varied attribute and trait combinations.
If your invitation is for the 85-minute Solve, use the all-in-one McKinsey Solve bundle so Redrock, Sea Wolf, and Sustainable Future Lab practice sit in one place. The full set of options is in the McKinsey Solve collection.
