What to take away
- The case is generally interviewer-led, which means the interviewer chooses the sequence and you answer the question in front of you
- The behavioural half is a separate assessment with its own criteria, and it goes deep on a couple of stories rather than wide across many
- Exhibits tend to arrive unprompted, so the skill being tested is reading a chart cold and saying what it means for the client
- When a number is handed to you, the sense-check sentence afterwards is worth more than the arithmetic itself
- Format details vary by office, role and round, so your recruiter's instructions outrank any guide including this one
- The highest-return rehearsal is answering exactly what was asked, then stopping
What does a McKinsey interview usually look like end to end?
The shape most candidates encounter is consistent enough to prepare for. You meet consultants across more than one round, and each interview usually holds two distinct assessments: a case, and a conversation about your own past experience. Both are marked. Candidates who treat the second as a warm-up chat lose points they never see.
Within a slot, the same consultant usually runs both parts. So whoever judges your structure also judges whether your story survived three follow-ups. One impression, from two very different kinds of evidence.
With that caveat carrying full weight, here is the generally observed pattern.
- A screening stage on your written application, before anyone speaks to you
- One or more interview rounds, each usually several separate interviews with different consultants
- Within each interview, a case plus a personal experience conversation, run by the same interviewer
- Later rounds with more senior interviewers, the case content often looser and the questioning sharper
- For some roles and geographies, a digital problem-solving assessment somewhere in the process
On that last point, keep expectations loose. A digital problem-solving assessment has featured at some stages for some roles, and what it involves has changed over time. We deliberately do not describe its mechanics, because anything specific could be wrong for your process. If it applies to you, your recruiter will say so and point you at better materials than any second-hand account.
One disclosure, plainly: MBB Ready is not affiliated with, endorsed by or connected to McKinsey & Company. What follows comes from the experience of having interviewed candidates at the firm, described in general terms, plus what is publicly discussed about the format. It is not an internal document and does not speak for the firm.
What does interviewer-led actually mean when you are in the room?
The thing to understand about this format is who holds the map. In a candidate-led case you are handed a problem and expected to drive: propose a structure, choose a branch, ask for data, decide when to move on. In an interviewer-led case the interviewer already has a sequence and works through it with you. You are not steering. You are answering.
That changes the texture of the conversation. Exhibits arrive because the interviewer decided it was time, not because you asked. Topics change on their cue. A question can land with no obvious link to the branch you were exploring, and that is not a trap, it is the next item on their list. The pace is brisker than candidates expect, because there is a defined set of questions and a finite slot.
What good looks like
Answer the question asked, completely, then stop talking. That is the whole discipline, and it is harder than it sounds. A strong answer here is short, structured and decisive, aimed at the thing in front of you, at a pace that respects the questions still coming.
Concise is not the same as thin. Asked to structure, still take your beat and build something that fits. Asked for a number, give the number. Asked what you would do, commit. Each lands as a self-contained answer, then hands the conversation back.
The trap: driving anyway
Most candidates prepare on candidate-led material, because that is what most published practice cases look like. Then they arrive, get asked a question, answer it, and keep going: "and building on that, the next thing I'd want is the cost side, so if I could see data on...". It feels proactive. It reads as not listening.
The cost lands in two places on the scorecard. Precision drops, because you answered the question asked and then several nobody asked. Listening drops, because the interviewer has to reclaim the conversation, and one who interrupts you twice will note it. Neither is a judgement about your intelligence. Both are judgements about whether you are easy to work with under time pressure.
“That is my structure, three branches. I would start with the second, because that is where the client's own data suggests the problem sits. Where would you like to take it?”
“That is the recommendation I would make on what we have seen. Is there a part of it you want me to push on?”
What is the personal experience interview really marking?
The behavioural half is not a personality check and not filler. It assesses a different dimension, with its own criteria, and in our rubric it carries a quarter of the overall score. Track record and drive is marked from your stories, which is why a case-only session cannot touch it. Five things are marked, and knowing them tells you what to put in the foreground.
| What is marked | The question behind it |
|---|---|
| Self-direction | What did you start that nobody asked you to start, and what did it cost you |
| Collaboration | How do you get things done through people who do not report to you |
| Depth | Does the story hold up under three more questions, or thin out into generalities |
| Growth mindset | What did you change after hard feedback, not what did you learn in the abstract |
| Results delivered | Did anything measurably change, and can you say what and by how much |
Depth is the one candidates systematically under-prepare, and the one this format punishes hardest. These interviews tend to probe one story for several minutes rather than collecting a portfolio. You will be asked who exactly objected, what you said, what happened when that failed, and what you would do differently. A story built to be told once cleanly collapses under that. A lived one does not. So build three, built to survive interrogation rather than to sound good on first telling.
Set the situation in two sentences
Enough context that the stakes are legible, and no more. Who, what was at risk, why it was hard. Candidates routinely spend two minutes on background and thirty seconds on their own contribution, inverting what is assessed.
Say what you did, not what the team did
The commonest failure here is the drift into "we". We aligned the stakeholders, we redesigned the process. Nobody can score "we". Name the move that was yours: the conversation you asked for, the analysis you ran overnight, the decision nobody else would make. Collaboration is still marked, through your actions inside a group.
Land a measurable result
Something changed and you can say by how much: a percentage, a time saved, a decision that went one way instead of the other. Where the number is genuinely unavailable, say so and give the concrete consequence. A vague result is unscoreable.
Prepare the follow-up chain, three deep
Write the three hardest follow-ups for each story and answer them aloud. Who disagreed and what was their argument. What did you get wrong. What happened after the part where the story usually ends. If an answer is thin, do not lead with that story.
Have the honest thing you would do differently
Growth mindset is marked on change, not humility. "I would have communicated more" is not an answer. "I should have taken the finance lead's objection seriously in week two rather than week six, because by then the budget was committed" is. Prepare it, because it is almost always asked and improvised.
How do the maths and exhibit questions work?
The quantitative questions have a particular shape, and it follows from the interviewer holding the map. Rather than deciding what to calculate, you are handed the inputs and asked for a figure. A market sizing from two or three stated assumptions. A breakeven on numbers just read out. A margin from a table you have seen for forty seconds.
Exhibits work the same way. A chart appears, you get a short moment with it, and you are asked what you see. Sometimes the question is genuinely open, which is harder, because an open question about a chart is really a question about commercial judgement: not what the bars are doing, but what they mean for this client and this decision.
Why the sense-check sentence matters so much
When you generate your own numbers you have some feel for whether the answer is plausible, because you chose the inputs. When a number is handed to you that feel is missing, and this is where candidates confidently announce a figure off by a factor of a thousand. The interviewer already knows the answer. What they are learning is whether you would have noticed being wrong.
So the sentence after the arithmetic is not politeness, it is the substance of the mark. Compare the answer to something: another number in the case, a rough industry intuition, the client's own scale. Say why it is plausible, or why it is surprising. A candidate who lands on a number, tests it, finds it strange and says so outscores one who happens to be right and moves straight on.
State units every time, and state what the number is not. Millions or thousands. Revenue or profit. Per year, per month, per store. Many wrong answers we mark are correct arithmetic on the wrong unit, all avoidable by saying the unit aloud while writing it down.
“Let me confirm the inputs: the site runs 300 days a year, 4,000 units a day, 12 euros a unit. Is that right?”
“So annual volume is 300 times 4,000, which is 1.2 million units.”
“At 12 euros a unit, that is 1.2 million times 12, so 14.4 million euros. Revenue, not profit, and for a full year.”
“Sense-check: ten sites and group revenue around 150 million means an average site is about 15 million. At 14.4 million this one sits right at the average, which fits it being a typical plant.”
“So 14.4 million euros of annual revenue. Shall I take that into the cost side?”
What is the interviewer writing down while you talk?
Interviewers do not leave with an overall impression and then reverse-engineer reasons for it. They leave with a filled-in grid: named skills, a score against each, a line of evidence for why. The debrief afterwards is a conversation about those cells.
Our rubric has three dimensions and fourteen skills: analytical thinking at 50%, presence and communication at 25%, track record and drive at 25%. Each skill is marked 1 to 5, where 3 means you met the first-round standard rather than average. Firms differ in vocabulary and weighting at the edges, but the shape is stable, so map each moment onto the cell it feeds.
| The moment | The skill it feeds | What earns the mark |
|---|---|---|
| Your opening structure | Framing | A structure built for this problem, non-overlapping, with a named place to start |
| The exhibit you are handed | Commercial judgement | What the chart means for the client and the decision, not what the chart shows |
| The calculation | Rigor | An audible chain, assumptions stated, units named, sense-checked against something |
| The brainstorm question | Creativity | Ideas from a named cut rather than free-associated, with trade-offs attached |
| Your closing recommendation | Synthesis | Answer first, then reasons with numbers, then risk, then next step, inside a minute |
| How you handle each turn | Precision and listening | Answering exactly what was asked, catching the nudges, stopping when done |
The last row matters most in practice, and nobody drills it. In a candidate-led case you can be verbose and still score well, because verbosity reads as initiative. Here, every unnecessary sentence costs a little in two cells at once, repeated across every question.
How should you prepare for this specific format?
Generic case practice gets you most of the way, because the analytical skills are the same everywhere. What it will not give you is the habit of being led, and that is worth deliberate rehearsal, because it cuts against what candidate-led practice trains in.
Rehearse answering exactly what was asked
Take any list of case questions, answer each aloud, and stop. Then check: did you answer the question, or answer it and add two more things? Highest-return drill for this format, and it needs no partner. Record yourself, because the extra sentences are invisible from the inside.
Practise taking the beat
Ask for your moment, take it in silence, then speak. Thirty to ninety seconds of quiet is normal and expected. What costs you is narrating your thinking while you assemble it. Say "could I take thirty seconds", then be quiet.
Keep a running structure you never present
The interviewer chooses the sequence, so you will build a structure you use one branch of. Keep the whole thing on paper and update it as information arrives. It is what makes your close coherent, and what lets you take an out-of-nowhere question without losing the thread.
Drill handed-over calculations, not self-designed ones
Have someone read you three numbers and ask for a fourth, or write inputs on cards and pull them at random. Confirm, work audibly, state the unit, sense-check. Not arithmetic speed: the aim is making that wrapper automatic under pressure.
Say the behavioural stories out loud, with the follow-up chain
Reading a story you wrote is not rehearsal. Say it, time it, then have someone ask your three hardest follow-ups back to back without letting you reset. Alone, ask and answer them aloud anyway. Hunt for where you start speaking in generalities.
Do one full run under real conditions
Voice, clock running, no looking things up, someone else's questions. The gap between how well you think you handle interruption and how well you do only shows up live.
Both formats are worth running, because the habits differ: an interviewer-led case rehearses the discipline above, a candidate-led one exercises the structuring muscle. Preparing for this firm, weight your reps towards interviewer-led.
Our default session is a 30-minute case-only run, which cannot assess track record and drive because it has no behavioural part. The 60-minute version at /full includes one, and that is the one to use once your case fundamentals hold up. The first case is free. Our coaches are AI personas, designed by ex-MBB interviewers, and we say so everywhere: you are not talking to a person, you are talking to a rubric with a voice around it.
Common questions
Is the McKinsey case interview interviewer-led or candidate-led?
It is generally described as interviewer-led, meaning the interviewer works through a prepared sequence of questions, hands you exhibits when they choose to, and controls the transitions. You still build structures and commit to recommendations, but you are not deciding what gets explored next. Individual interviewers vary in how tightly they hold that sequence, and later rounds are often looser than first rounds.
What is the McKinsey personal experience interview?
It is the behavioural half of the interview, usually run by the same consultant who runs your case, in which you are asked about your own past experience in depth. Expect one or two stories probed for several minutes each rather than a rapid tour of many. It assesses a separate dimension from the case: self-direction, collaboration, depth, growth mindset and results delivered.
How many rounds should I expect, and how long is each interview?
Most candidates go through more than one round, with several separate interviews inside each. Beyond that, we will not give you numbers, because they genuinely vary by office, role and year, and a confident figure from a prep site is worse than no figure. Your recruiter will tell you the actual shape of your process, and that is the version to plan around.
Is there a test before the interviews?
For some roles and in some geographies, a digital problem-solving assessment has featured somewhere in the process, and what it involves has changed over time. We deliberately do not describe its mechanics, because anything specific could be out of date for your process. If it applies to you, your recruiter will say so and will point you at the practice materials worth using.
Do I need to prepare differently for McKinsey than for BCG or Bain?
The underlying skills are the same, so most of your preparation transfers. What differs is format habit. BCG and Bain lean candidate-led, so you drive; McKinsey leans interviewer-led, so you answer precisely and hand back. Spend a few reps rehearsing that difference rather than rebuilding your fundamentals, and prepare the behavioural stories to a genuine three-follow-ups depth.
What is the fastest way to improve at an interviewer-led case?
Practise answering the exact question asked and then stopping. Record yourself doing it, because the extra sentences you add are invisible while you are speaking and obvious on playback. That single habit moves precision and listening, which is where prepared candidates most often lose ground in this format, and it costs nothing to drill on your own.