Practice case interviews with an interviewer who shows you the exhibits
38 interviewer-led McKinsey, BCG and Bain style cases from the Kellogg, Wharton and Booth casebooks and The Ultimate Case Interview Workbook. You talk it through by voice, the interviewer puts each exhibit on your screen when the case reaches it, and the report grades you against the case's own answer key. In your browser, or on Google Meet and Zoom.
A full MBB mock case on Tough Tongue AI, exhibits and all.
38
Interviewer-led cases, live tonight
20 to 40
Minutes per case, the length of a real first round
129
Cases indexed from 5 MBA casebooks
How a case interview runs
A case gives you a business problem and about half an hour to solve it out loud. The problem is usually a real client situation in disguise: should a drink maker enter a new market, why is an airline losing money faster than its rivals, what should a scooter cost in Vietnam. The interviewer is watching how you structure an unfamiliar problem, whether you read data carefully, whether your math holds up under a clock, and whether you can commit to an answer and defend it.
The Wharton Consulting Club casebook budgets a single case at about 20 to 25 minutes. Real first rounds with fit questions run longer, and the hard cases in our collection go to 40. The order is nearly always the same.
- 1 · 2 to 3 min
Prompt and clarifying questions
Restate the problem, confirm the objective, ask what you need.
- 2 · 4 to 5 min
Framework
Lay out how you will attack it, then say where you would start.
- 3 · 3 to 5 min
Building the solution
Probe, prioritize, brainstorm in buckets.
- 4 · 8 to 10 min
Exhibits and math
Read the chart, find the trap, set up the math before you calculate.
- 5 · 2 to 3 min
Recommendation
Answer first, then reasons, risks and next steps. Hold it when pushed.
Interviewer-led or interviewee-led
Cases come in two shapes, and the firm you are interviewing with mostly decides which one you get. In an interviewer-led case the interviewer holds the order of questions, hands you one exhibit at a time and scores each part on its own. In an interviewee-led case you drive: you decide which branch of your framework to test and ask for the data yourself.
Interviewer-led
McKinsey, Accenture, Oliver Wyman
You are scored question by question, so a weak framework does not sink you, but every exhibit is a separate test. Practice reading charts cold and setting up math before you calculate.
Interviewee-led
Bain, BCG, Deloitte, L.E.K., PwC
Your framework is the map, and asking for the right data at the right time is part of the score. Practice saying your hypothesis and moving the case forward without waiting to be asked.
Firm groupings follow the Wharton Consulting Club casebook, which calls them generalizations from candidate experience. Strategy& shows up on both sides.
What practice has to reproduce
Reading a casebook is learning to drive in a parked car. Practice transfers when it puts you in the moving one. These are the four things we built the case interviewer around.
The exhibit lands on your screen
The interviewer flips a slide deck to each chart or table at the moment the case reaches it and asks what you see. Nothing is read out to you.
The case runs its full length
Twenty to forty minutes, with the second calculation, the brainstorm, and a push-back built from what you said. That back half is where most candidates lose the case.
You find your own math errors
When a number is off, the interviewer asks you to walk through it rather than correcting you. That is what a real interviewer does.
Feedback against the answer key
The report puts your numbers next to the casebook's, scores structure, math, judgment and communication, and quotes the moments that decided it.
A lot of first rounds now happen on video. On a paid plan the interviewer can join a Google Meet or Zoom call as a participant, with a voice and an optional live avatar face, and run the case there, so you rehearse in the same window and on the same camera as the real thing.
All 38 practice cases
Every case is interviewer-led, uses the casebook's own exhibits, and is graded against its answer key. Click a case to start it in your browser. Difficulty is the casebook's own label where it gives one.
Kellogg Consulting Club 2024 Casebook (13 cases)
All cases in this casebook →-
Avalon
A government picks which power plants to build: capacity math and wind, solar and gas costs.
Opportunity assessment · Energy and utilities Run → -
Busch's Barber Shop
A barbershop launch in Evanston: market sizing, two sites, and a loyalty program decision.
Market entry · Hospitality Run → -
Bubbly LLC
A hand-soap maker at full capacity fixes its product mix and weighs West Africa.
Market entry · Retail and CPG Run → -
Chic Cosmetology
A cosmetology college weighs a new campus: breakeven, customers and market share.
Market entry · Education Run → -
Chicouver Cycle
A bikeshare startup picks neighborhoods and docking or dockless across four trip types.
Market entry · Transportation Run → -
Dark Sky
A small drone maker wants growth: new-product economics with cannibalization.
New product · Aerospace and defense Run → -
DigiBooks
An e-reader launch in a new country: segments, channels and channel profitability.
Market entry · Retail and CPG Run → -
Events.com
A ticketing platform gains share while profit falls: find the pricing root cause.
Profitability · Technology Run → -
Garthwaite Healthcare
A health insurer's costs outrun the market: size the commission bill and redesign it.
Profitability · Healthcare payer Run → -
Health Coaches
A payer with 5 million members weighs a disease-management program.
Operations · Healthcare payer Run → -
Healthy Foods
A wholesaler looks for growth in its own customers: three exhibits, almost no math.
Growth strategy · Food wholesale Run → -
High Q Plastics
Profit fell from $50M to $20M: rebuild the plant P&L under lean manufacturing.
Profitability · Automotive parts Run → -
Kellogg Capital
Strong returns, flat revenue: convert performance fees and find the fund that doubles it.
Growth strategy · Asset management Run →
Wharton Consulting Club Casebook 2024-25 (9 cases)
All cases in this casebook →-
Electro Chargers (case 1)
A battery-swap maker picks a city to enter and sizes the fixed investment.
Market entry · EV infrastructure Medium Run → -
Happy Bellies (case 2)
A meal-kit pioneer looks for growth through a new product launch.
Revenue growth · Food and beverage Hard Run → -
Smart Radiators (case 3)
A Japanese manufacturer picks which smart-valve startup to buy, and prices the deal.
Acquisition · Consumer electronics Medium Run → -
Compressor Co. (case 4)
A gas compressor maker with flat revenue reads competitor and refinery exhibits.
Revenue growth · Industrial goods Medium Run → -
PSU Bank Co. (case 5)
A bank's mutual fund arm grows slower than the industry: three exhibits, two brainstorms.
Revenue growth · Financial services Hard Run → -
Fruit Drink Co. (case 6)
A pomegranate drink loses share and money: one per-unit cost exhibit.
Profitability · Consumer packaged goods Easy Run → -
Race to the Bottom (case 7)
A one-route German airline loses profit faster than rivals: revenue-per-flight math.
Profitability · Airlines Medium Run → -
Coolant Co. (case 8)
A Chinese cooling maker weighs the US: direct entry or partnership on a two-year payback.
Market entry · Data-center cooling Medium Run → -
Lux Rides (case 9)
An Italian scooter brand prices for Vietnam: landed cost, rivals, willingness to pay.
Pricing · Automotive, luxury Hard Run →
Chicago Booth MCG Casebook 2024-25 (3 cases)
All cases in this casebook →-
Army Hotel (case 1)
A PE firm weighs a 400-room hotel on an army base: per diem pricing and breakeven.
Market entry · Hospitality Medium Run → -
Breast Cancer Surgery Device (case 2)
A device maker sets a revenue-maximizing US price from an adoption-by-price exhibit.
Pricing · Medical devices Easy Run → -
Burger Palace (case 3)
A Chicago chain sizes an airport location from gates and load factors, then breakeven share.
Market entry · Restaurants Medium Run →
The Ultimate Case Interview Workbook (13 cases)
-
Aquador
A $2B beverage producer weighs zero-calorie flavored water.
Market entry · Beverages Easy Run → -
Swift Enterprises
A jet ski maker's revenue falls 9% a year: diagnose it, then share and promotion math.
Revenue decline · Consumer goods Easy Run → -
Get-A-Ride
A Moscow taxi company decides whether to buy new airport permits.
Market entry, capacity · Mobility Medium Run → -
GDS Systems
A chip maker decides whether to renew its corporate jet lease.
Cost reduction · Technology Easy Run → -
AG&E
A utility meets a renewables mandate: buy credits or build a wind farm, then a twist.
Cost comparison · Utilities Easy Run → -
Hens Co.
A PE firm screens a chicken farm against two explicit criteria.
Acquisition screen · Agriculture Medium Run → -
Surfside
A surfboard maker lands a 25,000-board order with capacity for 5,000.
Capacity, operations · Leisure Hard Run → -
Polystore
Is the loyalty program working? Four charts and a selection-effect correction.
Program evaluation · Retail Medium Run → -
Bushel Co.
A backpack maker weighs moving production to Vietnam: layered profitability math.
Outsourcing · Manufacturing Hard Run → -
Prytown
Should a PE firm buy a city's parking meter rights, and what should it bid?
Investment, valuation · Private equity Medium Run → -
Telcast
A satellite TV provider weighs broadband: build or lease, then expected value.
Market entry · Telecom Medium Run → -
Burger Corp
The #3 US burger chain weighs buying a chicken chain: synergies and cannibalization.
Acquisition · Fast food Medium Run → -
Pink Cross
A payer-provider fixes falling profit: segmented sizing and two break-even problems.
Profitability · Healthcare Hard Run →
Prefer the app's view? The same cases are in the MBB Case Interview collection.
Casebook indexes
MBA consulting clubs publish casebooks every year, and they are still the best source of realistic cases. We read five of them, the three our live cases come from plus Fuqua and Tuck, and indexed every case by type, industry, format and difficulty, so you can pick the next one on purpose. The cases themselves stay in the books.
By firm
McKinsey practice cases
McKinsey's five published sample cases, the five question types in an interviewer-led case, and where to drill each one. →
2024
Kellogg Consulting Club, Northwestern
36 cases in 439 pages · 13 live on Tough Tongue AI
See every case →2024-25
Wharton Consulting Club, University of Pennsylvania
18 cases in 280 pages · 9 live on Tough Tongue AI
See every case →2024-25
Management Consulting Group, Chicago Booth
45 cases in 385 pages · 3 live on Tough Tongue AI
See every case →2023-24
Duke MBA Consulting Club, Fuqua
18 cases in 176 pages · build any of them yourself
See every case →2024
Tuck Consulting Club, Dartmouth
12 cases in 142 pages · build any of them yourself
See every case →A practice loop that works
Most consulting clubs suggest 20 to 40 full cases, weighted toward the last six weeks. The number matters less than the loop. Ten cases with calibrated feedback, each redone until the fix sticks, beat thirty with a friend who goes easy on you.
- 1
Run a full case
End to end, out loud, with the clock running. Start with an easy one from the list above.
- 2
Read the report
Find the one thing that cost you the most: a missed footnote, a framework that did not survive the second exhibit, a recommendation you dropped when pushed.
- 3
Redo the same case
A day or two later. If the same mistake shows up, it is a habit, and habits are what you are here to fix.
- 4
Then a new case
Same case type, different industry. Move to medium and hard cases once easy ones stop surfacing new mistakes.
Build a case from any casebook
If the case you want is not in the list, build it. Tough Tongue AI has an MCP server that works in Claude, ChatGPT, Codex and Cursor, and the open-source case-interview-builder skill turns a case from a casebook PDF, a doc or pasted text into an exhibit deck and an interviewer-led case with the answer key hidden from you, a weighted rubric and a study guide.
It needs the Tough Tongue MCP server and the Gamma connector for the deck. The companion case-interview-study-guide skill writes a prep sheet for the same case.
# the core plugin plus the MCP server
npx plugins add tough-tongue/toughtongue-skills
# the community interview skills
npx plugins add tough-tongue/ttai-community-skills
# then, in your agent
"Turn case 12 of the Wharton casebook into an interviewer-led case interview."
Case interview guides
MBB Case Interview Prep with AI
What a case actually tests, one exhibit walked through, and how the AI case tools compare.
Read the guide →The Best AI Case Interview Tools in 2026
Thirteen tools compared on exhibits, case length, grading, Zoom and Meet, and price.
Read the guide →Practice Case Interviews by Yourself
A tested ChatGPT and Claude interviewer prompt, where it breaks, and a five week plan.
Read the guide →Market Sizing Questions, Worked
Six estimates worked end to end, top-down and bottom-up, with every assumption stated.
Read the guide →The Profitability Framework
The profit tree, five diagnostic questions, and a profitability case worked with the math.
Read the guide →Preparing for PM, system design or coding rounds too? See AI interview practice for the other 500+ mock interviews.
Frequently asked questions
What is a case interview?
A case interview is the main round in consulting hiring at firms like McKinsey, Bain and BCG. The interviewer describes a business problem, usually a real client situation in disguise, and you structure it, work through data the interviewer hands you as charts and tables, do arithmetic out loud, brainstorm, and end with a recommendation they push back on. A first-round case usually runs 25 to 40 minutes, often after a few minutes of fit questions.
What is the difference between an interviewer-led and an interviewee-led case?
In an interviewer-led case, the interviewer holds the order of questions: after your framework, they tell you which area to look at, hand you one exhibit at a time, and score each part separately. McKinsey runs cases this way. In an interviewee-led (candidate-led) case, you decide where to go next and ask for the data yourself, which is more common at Bain and BCG. The exhibits and the math are the same skills, but the pacing is different, so practice both.
How do I prepare for a case interview?
Learn the common case types and a few starting structures, then spend most of your time running full cases out loud: 20 to 40 complete cases is the usual range consulting clubs suggest, and the Wharton Consulting Club's casebook puts typical preparation at 60 to 80 hours. After each case, read specific feedback, redo the same case to check the fix stuck, then move on. Drill case math and chart reading separately, because those are where most candidates lose points.
Where can I find McKinsey practice cases?
McKinsey publishes a few sample cases on its careers site, and MBA consulting club casebooks such as Kellogg, Wharton and Booth include McKinsey-style interviewer-led cases. On Tough Tongue AI you can run 38 interviewer-led cases from those casebooks and The Ultimate Case Interview Workbook with a voice interviewer that shows each exhibit on screen and grades you against the case's answer key.
Can I practice BCG and Bain style candidate-led cases here?
Partly. All 38 cases run interviewer-led, which is closest to McKinsey. Many of them come from cases the casebooks write as interviewee-led, so the content matches what Bain and BCG ask, but the AI interviewer still holds the order of questions. For true candidate-led practice, where you ask for every piece of data, a human partner is still the better drill, and the casebook pages here mark which cases are written that way.
Can AI run a realistic case interview?
Yes, if it does what a real interviewer does: puts the exhibit in front of you instead of reading it out, runs the whole case including the second calculation and the push-back on your recommendation, waits while you find your own arithmetic errors, and grades you against the case's answer key. What AI cannot fully judge yet is presence, so add a few human mocks in your last two weeks.
How much does case interview practice on Tough Tongue AI cost?
New accounts get 25 free minutes, which is enough to try the format, but free sessions run 10 to 12 minutes, so a full case needs a paid plan. Starter is $20 a month for 150 minutes, about five full cases, and includes running the interview on Google Meet or Zoom. Pro is $99 a month for 900 minutes.
Can I turn a case from my own casebook into an AI mock interview?
Yes. The open-source case-interview-builder skill runs in Claude Code, Codex, Cursor and other agents. Give it a case from a casebook PDF, a doc or pasted text, and it builds an exhibit deck, then creates an interviewer-led case on Tough Tongue AI with the answer key hidden from you, a weighted rubric and a study guide. It needs the Tough Tongue MCP server and the Gamma connector.
Your first real case should not be your first live one
Pick an easy case, run it tonight, and read where the points went.