Author: Taylor Warfield, Former Bain Manager and interviewer

Last Updated: May 20, 2026

 

To simulate a McKinsey case with AI, prompt a tool like ChatGPT, Claude, or Google Gemini to act as a strict McKinsey interviewer who drives an interviewer-led case, presents exhibits, and grades you on structure, math, and synthesis after every response. The key is using detailed, McKinsey-specific prompts that force the AI to push back instead of defaulting to generic praise.

 

By the end of this article, you'll know which prompts to use, which AI tools work best, and how to avoid the most common mistakes candidates make.

 

But first, a quick heads up:

 

McKinsey, BCG, Bain, and other top firms accept less than 1% of applicants every year. If you want to triple your chances of landing interviews and 8x your chances of passing them, watch my free 40-minute training.

 

What does it mean to simulate a McKinsey case with AI?

 

Simulating a McKinsey case interview with AI means using a chatbot like ChatGPT, Claude, or Google Gemini to play the role of a McKinsey interviewer. You give the AI a detailed prompt that defines the case, the interview style, and the grading rules. The AI then drives you through the case, asks follow-up questions, presents data, and gives feedback on your performance.

 

Done right, this gives you unlimited free practice in the exact format McKinsey uses. Done wrong, it gives you a chatbot that praises every answer and trains you to fail.

 

The difference is in the prompt. AI tools are trained to be agreeable. Without strong direction, they will call your framework "great" even when it's a memorized template that would fail in a real interview.

 

Why use AI to practice McKinsey case interviews?

 

AI case simulation has become a standard prep tool among serious candidates because it solves the five biggest problems with traditional case practice. It's available 24/7, free, unlimited in volume, McKinsey-specific in format, and gives you instant structured feedback.

 

There are five specific reasons AI case practice works:

 

• Available 24/7. You can run a case at 11 PM or 6 AM. No scheduling, no time zones, no waiting on a partner.

 

• Free or nearly free. ChatGPT, Claude, and Google Gemini all have free tiers that handle case practice well.

 

• Unlimited volume. Most candidates need 30 to 50 cases to be interview-ready. Finding 30 case partners is hard. AI can deliver that volume in one weekend.

 

• McKinsey-specific format. With the right prompt, the AI will drive the case the way a real McKinsey Associate or Engagement Manager does.

 

• Instant structured feedback. You see what worked and what didn't right after each response, while it's still fresh.

 

The catch is that AI alone is not enough. You still need real human practice for the parts AI can't replicate, like reading body language, handling true pressure, and getting nuanced feedback on tone and communication.

 

What you need to know about McKinsey cases before you start?

 

To get realistic practice, your AI prompt has to match how McKinsey actually runs cases. There are four things that make McKinsey cases different from BCG, Bain, or any other firm.

 

1. McKinsey uses an interviewer-led format

 

McKinsey is the only major firm that uses an interviewer-led case interview format. The interviewer drives the case by asking specific questions one at a time. You answer each one and move on.

 

This is different from BCG and Bain, where you present a framework and then drive the case yourself. Your AI prompt has to bake in this McKinsey-specific style or you'll practice the wrong format.

 

2. McKinsey cases are exhibit-heavy

 

According to McKinsey's careers website, interviewers regularly present charts, tables, and graphs that you need to interpret on the spot. McKinsey cases typically include 2 to 3 exhibits.

 

Strong candidates state the takeaway, connect it to a hypothesis, and quantify the impact. If your AI case has no exhibits, you're practicing the wrong skill.

 

3. McKinsey expects CEO-level synthesis

 

At the end of the case, you need to deliver a clear recommendation in 30 to 60 seconds. The structure is answer first, three supporting points, and one risk or next step.

 

Rambling synthesis is one of the fastest ways to get rejected. Your AI prompt should always include a graded synthesis at the end.

 

4. McKinsey evaluates four dimensions

 

McKinsey evaluates candidates on four core dimensions: problem solving, personal impact, entrepreneurial drive, and leadership. The case tests the first dimension. The McKinsey PEI tests the other three. Both carry roughly equal weight in the final hiring decision.

 

This is why your AI prep should include both case simulations and PEI practice. Skipping the PEI means prepping for half the interview.

 

How do you simulate a McKinsey case with AI in 7 steps?

 

To simulate a McKinsey case with AI, choose your tool, paste a McKinsey-specific prompt, set a 30-minute timer, work through the case out loud, ask for exhibits, deliver a tight synthesis, and ask the AI for strict feedback. The full process takes 35 to 45 minutes per case.

 

Here are the seven steps to run a realistic McKinsey case simulation:

 

1. Choose your AI tool (ChatGPT, Claude, or Google Gemini).
   
   
2. Open a new chat and paste a McKinsey-specific case prompt.
   
   
3. Set a 30-minute timer to mimic real interview length.
   
   
4. Work through the case out loud, typing your responses.
   
   
5. Ask for exhibits, data, and follow-up questions throughout.
   
   
6. Deliver your final synthesis in under 60 seconds.
   
   
7. Ask the AI to grade you and identify your three biggest weaknesses.

 

Step 1: Choose your AI tool

 

The three tools that work best for case simulation are ChatGPT, Claude, and Google Gemini. All three have free tiers and can run a credible McKinsey-style case if you prompt them correctly.

 

ChatGPT is the most flexible and has the best voice mode for spoken practice. Claude is more accurate on long math chains and stays focused over a 30-minute case. Google Gemini works well as a backup and handles charts well.

 

You don't need a paid plan. Free tiers are enough for daily practice.

 

Step 2: Open a new chat and paste a McKinsey-specific prompt

 

The prompt is the most important part of the entire simulation. A weak prompt produces a fake case where the AI agrees with everything you say. A strong prompt produces a case that feels close to the real thing.

 

I'll give you the exact prompts to use in the next section. Copy them, swap the industry to match what you want to practice, and paste them into the AI.

 

Start a fresh chat for each case. Old chats build up context that drifts away from the prompt.

 

Step 3: Set a 30-minute timer

 

Real McKinsey cases run 25 to 30 minutes. According to McKinsey's careers website, the full interview is 45 to 60 minutes, with roughly half on the case and half on the PEI.

 

Practicing under time pressure is what separates candidates who succeed from candidates who freeze in the real interview. Use your phone timer or any browser timer.

 

Step 4: Work through the case out loud

 

McKinsey cases are spoken, not typed. To get realistic practice, say your answer out loud, then type a shorter version into the AI. This builds the verbal communication skills you need on the real day.

 

If you're using ChatGPT or Gemini voice mode, you can skip the typing and just speak directly. The AI will respond out loud too.

 

Step 5: Ask for exhibits and data

 

McKinsey cases include data. If the AI is not giving you any, ask for it. Say something like, "Can you give me a chart that shows the revenue breakdown by segment over the last 3 years?"

 

The AI will produce a text-based table or describe the chart in detail. This is one of the most underused parts of AI case practice. Strong candidates demand the data they would get in a real interview.

 

Step 6: Deliver your final synthesis

 

When you've gathered enough information, ask the AI for a final question or move directly to synthesis. Your synthesis should follow this structure: answer first, three supporting points, one risk or next step. Aim for under 60 seconds.

 

A good synthesis sounds like this: "I recommend the client enters the European EV market. First, the market is projected to grow 18% annually. Second, the client's existing battery technology gives them a 30% cost advantage. Third, competitive intensity is low with only two major players. The key risk is regulatory shifts in EU subsidy programs, which we should monitor over the next 6 months."

 

Step 7: Ask the AI to grade you

 

This is where the AI earns its keep. Ask it to grade your performance on five dimensions: structure, hypothesis quality, math accuracy, exhibit interpretation, and synthesis.

 

A good grading prompt looks like this: "Grade my performance from 1 to 5 on structure, hypothesis quality, math, exhibit interpretation, and synthesis. Be strict. Identify the three biggest weaknesses I should fix before my next case."

 

This is the most valuable part of the simulation. Run the feedback, identify your weakest area, and target that in your next session.

 

What are the best AI prompts to simulate a McKinsey case?

 

The best AI prompts to simulate a McKinsey case force the AI into a strict interviewer role, specify McKinsey's interviewer-led format, demand exhibits and pushback, and require strict grading with specific weaknesses. Here are 6 copy-paste prompts that turn any AI tool into a McKinsey-style practice partner.

 

Prompt 1: Full McKinsey interviewer-led case

 

Act as a McKinsey Engagement Manager running an interviewer-led case interview. The client is a regional health insurance company in the US that has seen profits decline 20% over the last 2 years. Drive the case the way a real McKinsey interviewer would. Ask me one structured question at a time, push back on weak hypotheses, and present a data exhibit at some point during the case. Do not let me build a framework alone. After each of my answers, give a one-sentence grade on hypothesis quality, MECE structure, and communication. Be strict. Do not say "great job" unless the answer would actually land at McKinsey. Begin by reading the prompt aloud and asking me for clarifying questions.

 

This is your default prompt. Swap the industry and the problem to match the case type you want to practice. Try variations with airlines, retail banks, pharma companies, or tech firms.

 

Prompt 2: Market sizing drill

 

Generate a McKinsey-style market sizing question. The format should follow what real McKinsey interviewers use: a top-down approach with population, segmentation, behavior assumptions, and a final number. After I solve it, grade me on four dimensions: did I use round numbers, did I segment the population logically, did I sanity check the answer, and did I translate the final number into a business statement? Time me at 4 minutes. Be strict on whether my final answer makes business sense.

 

This prompt is perfect for daily drilling. Market sizing appears in roughly 30% of McKinsey cases, and AI is excellent for high-volume repetition.

 

Prompt 3: Exhibit interpretation

 

Create a fake McKinsey case exhibit. Describe a chart in detail: title, axes, 4 to 5 data points, and any patterns. Examples could be revenue growth by segment, market share over time, or cost breakdown by category. After I read it, ask me three questions: what is the chart telling me, what hypothesis does this support or contradict, and what would I want to investigate next. Grade my answers on accuracy, business judgment, and how clearly I connect the data to a recommendation.

 

Exhibit interpretation is the #1 area where candidates lose points in McKinsey interviews. Run this prompt 2 to 3 times per week.

 

Prompt 4: McKinsey-level math drill

 

Drill me on McKinsey-level case math. Give me 10 problems mixing percentages, growth rates, breakeven analysis, and sensitivity calculations. Use realistic messy numbers like 12.4% growth or $437 million in revenue. Time me at 90 seconds per problem. Grade each answer on four things: did I write the formula first, did I structure the math in clean stages, did I keep my units consistent, and did I translate the final number into a business takeaway. Be strict.

 

McKinsey expects fast, accurate mental math under pressure. If your case interview math is rusty, run this drill daily until your speed and accuracy both improve.

 

Prompt 5: McKinsey PEI practice

 

Act as a McKinsey partner running the Personal Experience Interview. Ask me to share a story about personal impact, leadership, or entrepreneurial drive. After I tell the story, ask three probing follow-ups that a real McKinsey partner would ask: what specifically did I do, what resistance did I face, and what was the measurable outcome. Then grade the story on three criteria: specificity of my role, evidence of impact under resistance, and clarity of the measurable outcome. Tell me which criterion was weakest.

 

The PEI carries equal weight to the case in McKinsey's final hiring decision. Most candidates underprep for it, which is why a graded PEI drill is so valuable.

 

Prompt 6: Synthesis drill

 

Give me a case prompt and 3 data points. After I read them, ask me to synthesize a recommendation in 60 seconds. My synthesis should follow this structure: state the recommendation first, give three supporting points, and identify one key risk. Grade my synthesis on five dimensions: did I lead with the answer, did my three supports actually back the answer, did I quantify the impact, was the risk substantive, and would a CEO act on this recommendation.

 

Synthesis is the most common reason strong candidates fail final round McKinsey interviews. Most candidates do not practice synthesis as a standalone skill. This drill takes 5 minutes per rep and pays off enormously.

 

If you want a faster way to learn McKinsey case strategies, my case interview course walks you through proven techniques in as little as 7 days while saving you 100+ hours of self-prep.

 

Which AI tool is best for simulating McKinsey cases?

 

The three best AI tools for simulating McKinsey cases are ChatGPT, Claude, and Google Gemini. Each has different strengths. The best approach for most candidates is to use ChatGPT for voice practice, Claude for math-heavy cases, and Google Gemini as a backup.

 

Here's how the three tools compare on the dimensions that matter most for case prep:

 

Dimension	ChatGPT	Claude	Google Gemini
Voice mode	Excellent (free tier)	Limited	Good
Math accuracy	Good	Excellent	Good
Long case focus	Drifts after 20 min	Stays on track	Mixed
Sycophancy	High (needs strict prompts)	Lower than ChatGPT	Moderate
Free tier limits	Generous	Tight on long cases	Generous
Best use	Voice case practice	Math + long cases	Backup + variety

 

For most candidates, the right approach is a stack of all three. Use ChatGPT for daily voice cases, Claude when you need precise math or longer multi-part cases, and Google Gemini as a backup or for variety.

 

What are the common mistakes when simulating McKinsey cases with AI?

 

The six common mistakes that ruin AI case practice are weak prompts, accepting agreeable feedback, skipping exhibits, no timer, text-only practice, and ignoring the PEI. Avoid these and your prep will be significantly more effective.

 

Mistake #1: Using weak prompts

 

The biggest mistake is asking the AI to "give me a McKinsey case" with no other direction. The AI will produce a generic case that has nothing in common with a real McKinsey interview.

 

Always use a detailed prompt that specifies the industry, the case type, the interviewer style, the grading rules, and the requirement to push back on weak answers.

 

Mistake #2: Accepting agreeable feedback

 

AI is trained to be helpful and positive. It will tell you your answer was "great" even when it was weak.

 

Always force the AI to identify three specific weaknesses, not just give you a score. Say, "Tell me the three things I did worst in this case."

 

Mistake #3: Skipping the exhibits

 

Most candidates do AI cases without asking for any data. Real McKinsey cases include 2 to 3 exhibits. If your AI case has no exhibits, you're practicing the wrong skill.

 

Demand exhibits. Ask the AI to describe a chart, give you a table, or present a data point you need to interpret.

 

Mistake #4: Not using a timer

 

Without a timer, you'll take 60 minutes on a case that should take 30. The time pressure is half the point of the interview.

 

Set a 30-minute timer at the start of every case and respect it. Once the timer runs out, deliver your synthesis whether you're ready or not.

 

Mistake #5: Practicing only with text

 

Real McKinsey interviews are spoken. Type your answers if you must, but say them out loud first.

 

Better yet, use voice mode in ChatGPT or Gemini. Speaking through a case builds different muscles than typing through one.

 

Mistake #6: Skipping the PEI

 

McKinsey pairs every case with a Personal Experience Interview. If you only practice cases with AI, you're prepping for half the interview.

 

If you want to focus on PEI and behavioral prep specifically, my fit interview course covers 98% of consulting fit interview questions in a few hours.

 

How do you combine AI practice with other prep methods?

 

AI case simulation is one tool, not the entire toolkit. Candidates who get McKinsey offers combine AI practice with three other methods: real partner practice, case interview coaching, and McKinsey's own free practice cases.

 

Here's how the four methods work together:

 

• Real partner practice. AI can't replicate body language, true interpersonal pressure, or back-and-forth dialogue. You still need 10 to 20 cases with a real partner before your interview.

 

• Case interview coaching. After 15 to 20 self-practiced cases, most candidates plateau. A case interview coach who has interviewed at McKinsey can spot patterns AI misses and accelerate your improvement.

 

• McKinsey's own practice cases. McKinsey publishes 7 free practice cases on their careers website. These are the closest thing to a real McKinsey case you'll find anywhere. Do all 7.

 

• AI simulation. Use AI for daily volume, math drills, market sizing, exhibit interpretation, PEI practice, and instant feedback.

 

A solid weekly prep plan covers all four:

 

• 3 to 4 AI-simulated cases per week

 

• 1 to 2 cases with a real partner per week

 

• 1 McKinsey official case per week

 

• 1 coaching session every 2 to 3 weeks in the final stretch

 

What tips help you get the most out of AI case simulation?

 

The 7 tips below will help you get the most realistic, useful practice out of every AI case simulation. Apply them consistently and your AI prep will start producing real interview-day improvement.

 

Tip #1: Use a fresh chat for every case

 

Old chat history makes the AI drift away from your prompt. Start every case in a new chat. Fresh chat, fresh case.

 

Tip #2: Be specific about the industry

 

Generic "company" cases are too easy. Real McKinsey cases are in specific industries with specific dynamics.

 

Always specify the industry: airline, retail bank, pharma, consumer goods, energy, retail, tech, or whatever sector you want to practice.

 

Tip #3: Force the AI to stay in role

 

If the AI breaks character and starts being too helpful, push back. Say, "Stop coaching me. Run the case as a McKinsey interviewer would. Push back on my answers and grade strictly."

 

Tip #4: Record yourself

 

Use your phone to record your voice during the case. Listen back afterward. You'll catch filler words, slow math, and unclear synthesis you couldn't catch in the moment.

 

Tip #5: Save the feedback

 

After each case, copy the AI's feedback into a notes document. Track patterns. If "weak synthesis" shows up in 4 of 5 cases, you know what to drill next.

 

Tip #6: Mix case types

 

McKinsey can give you a profitability case, market entry, M&A, growth strategy, pricing, or new product case. Don't run 10 profitability cases in a row. Rotate the case types.

 

Tip #7: Practice at realistic times

 

If your interview is at 9 AM, practice cases at 9 AM. Your brain performs differently in the morning vs. late at night. Train for the time of day you'll actually interview.

 

Frequently Asked Questions

 

Can AI replace human mock interviews for McKinsey prep?

 

No. AI is a strong supplement, not a replacement. Real human practice is critical for reading body language, handling pressure, and getting nuanced feedback on communication. The right approach is to use AI for daily volume and 10 to 20 real partner cases in the final 2 to 4 weeks before your interview.

 

Which AI tool is best for McKinsey case practice?

 

ChatGPT, Claude, and Google Gemini all work well, and each has different strengths. ChatGPT has the best free voice mode for spoken practice. Claude is more accurate on long math chains and stays focused over a 30-minute case. Google Gemini is a strong backup. All three are free at the volume needed for case prep.

 

How many AI cases should I do before my McKinsey interview?

 

Most candidates need 30 to 50 total cases to be interview-ready. Of those, 15 to 25 can be AI-simulated cases. The rest should be a mix of partner practice, McKinsey's official cases, and coaching sessions. Quality matters more than quantity, so review feedback after every single case.

 

Can AI grade my McKinsey case performance accurately?

 

AI gives structured feedback that's directionally useful, but it tends to be too generous. Always force the AI to identify three specific weaknesses rather than just give you a score. Cross-check AI feedback with feedback from a real partner or coach before your actual interview.

 

Will McKinsey know if I practiced with AI?

 

No. There's no way for McKinsey to know how you prepared. The firm actively endorses candidates using AI for prep, as long as you're not using it during a live interview. According to McKinsey's careers page, the firm considers AI a valuable preparation tool when used responsibly.

 

How do I make AI cases harder?

 

Add complexity to your prompt. Ask for a multi-part case with two exhibits, a tight time limit of 25 minutes, and an aggressive interviewer who interrupts. Force the AI to push back on your hypothesis at least twice. Demand strict grading where a 2 out of 5 is allowed if your answer is weak.

 

Is the McKinsey AI Interview the same as practicing with AI?

 

No, they are completely different. Practicing with AI means using ChatGPT, Claude, or Google Gemini to simulate a case interview as part of your prep. The McKinsey AI Interview is a new component of the McKinsey hiring process where candidates use McKinsey's internal AI tool, Lilli, during a live interview to demonstrate how they collaborate with AI.

 

Everything You Need to Land a Consulting Offer

 

Need help passing your interviews?

• Case Interview Course: Become a top 10% case interview candidate in 7 days while saving yourself 100+ hours
  
  
• Fit Interview Course: Master 98% of consulting fit interview questions in a few hours
  
  
• Interview Coaching: Accelerate your prep with 1-on-1 coaching with Taylor Warfield, former Bain interviewer and best-selling author

  

Need help landing interviews?

• Resume Review & Editing: Craft the perfect resume with unlimited revisions and 24-hour turnaround

 

Need help with everything?

• Consulting Offer Program: Go from zero to offer-ready with a complete system

 

Not sure where to start?

• Free 40-Minute Training: Triple your chances of landing consulting interviews and 8x your chances of passing them