How is McKinsey Solve scored? Product score, process score, and why finishing matters

Almost every question we get about McKinsey Solve is really a question about scoring. Candidates want to know what a good score looks like, whether a wrong answer is fatal, and whether the assessment is watching them work. The honest answer to that last one is yes, and understanding exactly what it watches is the difference between preparing usefully and preparing at random.

Here is how the scoring actually works, what the interface records, and what that means for the way you should play.

Solve produces two scores, not one

Unlike the old Problem Solving Test, Solve does not simply mark your answers. It evaluates both the outputs you produce and the way you produced them.^[1] Every module feeds both.

The consequence is uncomfortable but useful. Two candidates can submit identical answers and receive different scores, because one worked deliberately and the other flailed and happened to land in the right place.^[2] Conversely, a candidate who misses a calculation but works cleanly throughout is not automatically finished.

What the interface actually logs

This is the part most write-ups gloss over. The assessment is instrumented far more heavily than a normal online test. Reported behavioral signals include mouse clicks, scroll behavior, time spent per section, the sequence in which you considered information, and the moments you changed your mind before submitting.^[1][3]

Those raw signals roll up into judgments about a small set of traits. The dimensions commonly cited are critical thinking, systems thinking, decision making, metacognition and situational awareness.^[4] In practice, the behaviors that map onto them look like this.

None of this rewards cleverness. It rewards a consistent operating rhythm, which is precisely the thing you can only build by working inside the real interfaces repeatedly.

Why nobody can tell you the score you need

Three things make a target number impossible to quote honestly.

First, McKinsey does not publish the weighting between product and process. Coaches who work with large candidate volumes report that the two carry roughly comparable weight, but that is an inference from outcomes, not a disclosed figure.^[5]

Second, your combined result is benchmarked against your candidate pool rather than measured against a fixed cutoff, and is commonly reported as a decile rank.^[4][6] The bar moves with the strength of the cohort you happened to be assessed alongside. A performance that clears in one office and cycle may not clear in another.

Third, the score is not read in isolation. McKinsey states on its own assessment page that your Solve performance is considered alongside the rest of your application and any other assessments you are asked to take.^[10] In practice that means a mid-range result attached to a strong profile can still advance while a better result attached to a weaker one does not.^[4]

Overall pass rates for Solve are generally reported in the range of twenty to thirty percent, meaning most candidates who sit it do not advance.^[1] Treat any site quoting you a precise passing score as guessing.

You will never see your own result

McKinsey does not release component scores or feedback. You receive an email telling you whether you have progressed to case interviews, typically within one to fourteen days depending on the office and candidate volume.^[5] There is no breakdown, no diagnostic, and no way to work out afterwards which part let you down. That asymmetry is exactly why the preparation has to happen before you sit it.

Why finishing matters more than perfecting

This is the single most actionable thing to take from how the scoring works. An unfinished module damages you twice over. Every unanswered item is a product score you did not earn, and the pattern that produced it, over-investing early and running out of clock, is precisely the pacing failure the process score is built to detect.

Which clock you face depends on your invitation. The sixty-five minute version is Redrock and Sea Wolf. The eighty-five minute version adds Sustainable Futures Lab, which McKinsey began rolling out in phases from March 2026, so not every candidate receives it.^[8]

The eighty-five minute version is usually reported as thirty-five minutes for Redrock, thirty for Sea Wolf and twenty for Sustainable Futures Lab, with the clock allotted per module rather than shared across them.^[5][7] You cannot bank minutes you saved on one game and spend them on another.

Time per module, eighty-five minute Solve format

Minutes available. Time is allocated per module and cannot be carried between games.

Source: reported module timings, StrategyCase and MyConsultingCoach, 2026. Chart: case-prep.com

Given that structure, the arithmetic is simple. In Redrock, a mini-case answered at seventy percent confidence is worth more than a perfect answer you never reached. In Sea Wolf, a site left unattempted is a guaranteed zero on that portion. Budget your minutes per task before you start, and treat the budget as binding.

What a strong process looks like in practice

• Orient first, briefly. Read the objective and the constraints before touching anything. Sixty seconds spent here is not wasted time, it is a recorded signal.
• Decide your data before you open it. Know what you need for the task in front of you rather than sweeping every available panel.
• Commit, then move. Make the call, log it, move on. Revisit only if new information genuinely changes the picture.
• Keep the clock visible. Set a per-task allowance and abandon a task that overruns it, even unfinished.
• Use the tools provided. The in-game calculator and notes area exist and their use is observable.
• Never leave a field blank. An estimate scores above a gap on every module.

What the score does and does not decide

Solve is a screen, not a ranking that follows you. Once you reach the case interview stage, the assessment result stops driving decisions.^[3] Its whole function is to reduce the pool before McKinsey spends interviewer hours on it.

It is also effectively a single attempt. You get one sitting per application, and the waiting period before reapplying to the firm is commonly cited as twelve to twenty-four months.^[2][5] There is no retake inside a cycle to fall back on.

One note on preparation and integrity. McKinsey's own position is that there is no need to prepare, and candidates the firm has interviewed about the experience have said they simply read the official guidance and rested beforehand.^[9][10] The same assessment page prohibits using applications, websites, AI tools or pre-prepared notes during the assessment, prohibits recording or capturing screenshots of any part of it, and requires you to sit it alone without assistance, absent a written disability accommodation.^[10] Practicing the formats beforehand on a separate simulator is ordinary interview preparation and sits outside those rules. Bringing anything into the live assessment does not, and the stated consequences run as far as rescinded offers and restrictions on future applications.

What this means for how you prepare

If only the product score existed, the right preparation would be arithmetic drilling. Because the process score exists, the right preparation is repetition inside the real interfaces until your sequence of actions becomes automatic. The candidates who fail are frequently not the weakest thinkers. They are the ones meeting an unfamiliar interface under a hard clock for the first time, and the telemetry captures every hesitation while they work it out.

Interface familiarity is not a comfort blanket in this assessment. It is a scoring advantage.

Frequently asked questions