A McKinsey mock interview with AI should rehearse its two interview types: the Personal Experience Interview (PEI), where one story about you is explored in depth, and the problem-solving interview, a case that candidates commonly describe as interviewer-led, with the interviewer moving you through set questions on structure, an exhibit, math and a recommendation. A chatbot can play both interviewers if you give it strict rules, and it is good at relentless PEI follow-ups and case math drills; it is weaker at realistic exhibits and pressure. McKinsey welcomes AI for practicing interview questions but not during assessments or interviews, so all of this is preparation.
What McKinsey's interviews test
McKinsey's Interviewing at McKinsey page, as read for our McKinsey interview questions guide on November 17, 2026, describes personal experience interviews and problem-solving interviews for most client-facing roles, sometimes with expertise interviews, and Solve, its gamified assessment, for most consulting roles.
| Part | What McKinsey says | What to mock |
|---|---|---|
| PEI | Four themes: Connection, Drive, Leadership and Growth. Come with two examples for each, focus on your own role, lead with your reasoning rather than your actions, and treat it as a conversation rather than a rehearsed story | One story explored for 10 to 15 minutes, the length candidates commonly report |
| Problem-solving interview | There is no single logical solution; listen carefully and tailor your answer rather than reaching for standard frameworks. Five practice cases are published: Beautify, Diconsa, Electro-Light, Transforming a national education system and Talbot Trucks | An interviewer-led sequence with an exhibit and math |
| Solve | A game-based assessment taken before interviews; applications, websites, AI tools and notes are not permitted | Practice the game types separately and take it alone (McKinsey Solve assessment) |
| AI rules | AI is welcome for practicing interview questions and refining a résumé, but not for real-time answers in interviews or in assessments unless permitted. Separately, Fortune reported in January 2026 that some final rounds have piloted an exercise with McKinsey's own tool, Lilli | Prepare with AI; interview alone (does McKinsey allow AI in interviews) |
What to rehearse
PEI: one story per theme, then the follow-ups
McKinsey's sample prompts, paraphrased in our earlier guide: Connection, working with someone who held an opposing opinion; Drive, working hard to achieve excellence in tough circumstances; Leadership, working effectively with people of different backgrounds; Growth, learning something new quickly to handle a challenge. Expect follow-ups like these, which candidates commonly report:
- What exactly did you say to them?
- What were you thinking at that moment, and why?
- What did they say or do in response?
- What was the hardest moment, and why didn't you stop?
- What would you do differently now?
Case: the interviewer-led sequence
One case we wrote, broken into the stages candidates describe. A national chain of 400 pharmacies asks whether it should open small health clinics inside its stores.
- Structure: “What would you want to know to decide?”
- Exhibit: a table of patient visits and costs from a 20-store pilot. “What does this tell you?”
- Math: “What is each clinic worth per year?” (worked below)
- Brainstorm: “What could stop customers from using the clinics?”
- Synthesis: “The CEO walks in. What do you recommend?”
A 55-minute session script, with prompts
| Minutes | What happens | What you practice |
|---|---|---|
| 0–3 | A short introduction | Thirty seconds, not your whole résumé |
| 3–23 | PEI: the AI asks for one story on one theme and stays on it | Reasoning first, your own words and thoughts, consistency under follow-ups |
| 23–26 | Case prompt and your clarifying questions | Objective, scope, time frame |
| 26–31 | Structure | Buckets that fit this client, said out loud |
| 31–36 | Exhibit | The so-what in one sentence |
| 36–42 | Math | Every step aloud, then a sanity check |
| 42–46 | Brainstorm | Ideas grouped, not listed |
| 46–50 | Synthesis | Answer first, then reasons, risk and next step |
| 50–55 | Feedback | One fix for the PEI and one for the case |
Prompt: a PEI interviewer who won't let go
You are a McKinsey interviewer running a Personal Experience Interview on the theme [Connection / Drive / Leadership / Growth]. Ask me for one specific experience. Then spend about 15 minutes on that single story, asking one short follow-up at a time: what exactly I said, what I was thinking and why, what the other person said or did, and what I'd do differently. If I describe what 'we' did, ask what I did. If I describe actions before reasons, ask why I chose that. Never accept a summary. At the end, tell me where the story got vague and whether I led with my reasoning.
Why it works: it recreates the depth candidates describe and enforces McKinsey's own advice to lead with reasoning.
Prompt: an interviewer-led case
Run an interviewer-led case in the style of McKinsey's problem-solving interview. First write a private fact sheet for a [industry] client with one clear question, and keep every number consistent with it. Then lead me through it in this order, one question at a time: 1) the prompt and my clarifying questions; 2) 'What would you want to know to decide?'; 3) a data table I must interpret; 4) one calculation; 5) a brainstorm question; 6) 'The client walks in now: what do you recommend?' Don't do my math, but tell me if it's wrong. Afterwards, score structure, exhibit reading, math, creativity and synthesis from 1 to 5, with one fix each.
Why it works: the private fact sheet keeps numbers consistent, and the fixed order mirrors an interviewer-led case. Our guide to case interview practice with AI has a candidate-led version and math drills.
Case math, said out loud: a worked example
The pharmacy case above, with numbers we invented for practice. Say each line as you write it.
- “The pilot shows about 30 patients a day per clinic, open 300 days a year: 9,000 visits per clinic per year.”
- “At an average of 60 euros per visit, that's 540,000 euros of revenue per clinic.”
- “Running costs are 400,000 euros a year, so each clinic contributes 140,000 euros before any effect on the pharmacy.”
- “If 40% of patients fill a prescription in the store, with 15 euros of margin each, that's 3,600 prescriptions and 54,000 euros more: about 194,000 euros per clinic.”
- “Only stores with enough foot traffic should get one. If that's half the chain, 200 clinics add roughly 39 million euros a year.”
- Sanity check: “30 patients a day is one every 16 to 20 minutes in an eight-to-ten-hour day, which seems plausible for a pilot, but novelty may inflate it.”
Then the synthesis, answer first: “I'd roll out in phases to the 200 busiest stores, which adds about 39 million euros a year if pilot volumes hold. The main risk is that pilot volumes fall once the novelty wears off, so I'd open a second wave of 20 and track visits for six months before committing to the rest.”
Chatbots can slip on arithmetic, so redo any number they give you. Round the numbers, keep units on every line, and finish with an implication rather than a figure.
Score yourself
| Criterion | Strong | Weak |
|---|---|---|
| PEI: reasoning first | Says why before what, at each step | A list of actions |
| PEI: your role | “I” decisions, the words you actually used | “We” throughout |
| PEI: depth | New, consistent detail at every follow-up | Repeats the summary |
| Case: structure | Tailored to this client's question; buckets don't overlap | A memorized framework recited |
| Case: exhibit | The so-what in one sentence, linked to the question | Describes the table |
| Case: math | Clean steps aloud, units kept, a sanity check | Silent working, unchecked result |
| Case: synthesis | Answer first, two or three reasons, a risk and a next step | A recap of the case |
This rubric is ours, built on McKinsey's published advice; McKinsey doesn't publish its scoring.
A three-week plan
| Week | PEI | Case |
|---|---|---|
| 1 | Write eight stories, two per theme; run the PEI prompt on one a day | Learn the case stages; read our case interview explainer; daily math drills |
| 2 | Rerun your two weakest stories with a person asking the follow-ups | Work McKinsey's five published practice cases, then one AI case a day |
| 3 | Choose which story fits each theme on the day; keep spares | Live cases with a partner; one full 55-minute mock; taper |
What an AI mock can't simulate
- Real exhibits. A text table is easier than a dense chart. Use McKinsey's published practice cases for exhibit practice.
- A case partner's pressure. An AI won't sigh, interrupt or look at the clock. Do several live cases with a person.
- A reliable calculator. Chatbots can change a number mid-case. The fact-sheet rule helps; checking yourself helps more.
- The Lilli exercise. If your invitation includes it, it uses McKinsey's own tool, set up by McKinsey. Practicing critique of a general chatbot's case answer is the closest preparation.
Where CoPilot Interview fits
CoPilot Interview is an AI interview assistant whose free AI mock interview builds PEI-style questions from your résumé and the McKinsey posting, reads them aloud and scores each spoken answer out of 10, which suits rehearsing your eight stories until reasoning comes first without thinking. It runs question-by-question rounds rather than a full interactive case, so keep the case prompt above for cases. The free plan gives 15 minutes a day of live help in sessions of up to 15 minutes, with AI mock interviews included. McKinsey prohibits AI in assessments and real-time answers in interviews, so follow your employer's rules: practice before the day, and answer alone on it.
Disclosure: CoPilot Interview is our product, so weigh what we say about it accordingly.
Frequently asked questions
Is the McKinsey case interview interviewer-led?
It is commonly described that way: the interviewer moves you through set questions, usually a structure, an exhibit, a calculation, a brainstorm and a recommendation. McKinsey itself says there's no single logical solution and asks you to tailor your answer to the question.
How many PEI stories should I prepare?
McKinsey suggests two personal examples for each of its four themes, Connection, Drive, Leadership and Growth: eight in total.
Can I use AI during a McKinsey interview?
No. McKinsey welcomes AI for practicing questions and refining your résumé, but not for real-time answers in interviews or in assessments unless permitted. The reported exception is a piloted final-round exercise using McKinsey's own tool, Lilli.
Can AI help with McKinsey Solve?
Only before it. Solve doesn't permit AI tools, websites or notes, so practice the game types beforehand and take it alone.
Can ChatGPT run a McKinsey case?
Yes, with strict rules: a private fact sheet, data only when you ask, no doing your math, and a score at the end. Check its arithmetic, and add live cases with a person.
Rehearse it out loud before the real one
The AI mock interview builds questions from your resume and the job description, reads them aloud and scores each spoken answer out of 10. Included on the free plan.
Start a mock interview