McKinsey case interview: practice cases with worked examples (2026)

A McKinsey case interview is an interviewer-led business problem. The interviewer describes a client situation, then takes you through it one question at a time: a structure, an exhibit or two, some maths, perhaps a brainstorm and a closing recommendation. Each interview commonly runs for about an hour and pairs one case, often 25 to 35 minutes, with a Personal Experience Interview (PEI). Candidates commonly face two interviews in a first round and two or three in a final round with partners, and many first take McKinsey's online Solve assessment. Formats vary by office, role and year, so check your invitation.

Interviewer-led changes what good looks like. You do not decide what to analyse next; the interviewer does. That makes every question a separate test. A strong answer is short, structured and finished with a so-what, and then you stop and let the interviewer steer. Candidates also commonly report that calculators are not allowed, so the maths happens on paper, out loud.

Our library has four cases written in the McKinsey style: interviewer-led, one question at a time, with numbered exhibits handed over as the case moves on. Each has a guided walkthrough with feedback on every answer, and each can be run live with the AI interviewer.

Start with Nordwind: its maths is single-step and its exhibits are clean. Atlas and Northvale are harder, with multi-step maths and a judgement call at the end.

All three firms test the same core skills: structure, maths, reading data and judgement. What differs is who drives, how much data you get and how fit is tested. Read this table as a tendency, not a rule: interviewers vary, and every firm uses every case type.

If you are interviewing with more than one firm, practise both rhythms. Our BCG case interview guide covers driving a case yourself and evolving a hypothesis across exhibits.

An interviewer-led case is a chain of distinct question types, each marked on its own. Work out which one you are being asked, answer that, and stop.

Each worked example takes a case to the end of its first exhibit: prompt, structure, exhibit, first calculations and the hypothesis you would say out loud. The rest is yours to play.

Nordwind Airlines: the vanishing margin

The prompt. Nordwind is a short-haul carrier in northern Europe. Over three years passengers have grown 18%, but operating profit has fallen by roughly 40%. The CEO wants to know why and what to do. Management blames a new low-cost entrant's fares.

A structure that fits. Split profit into revenue and cost, then make the cost side airline-specific. A growing airline's costs must be compared per flight, and they behave in three ways: some grow with time, some with each turnaround and some with flying.

First calculations. Test management's hypothesis before accepting it. The average fare fell EUR 3, so on 2025's 22.7m passengers the lower fare cost EUR 3 × 22.7m = EUR 68.1m of revenue. That is real money. But revenue still grew EUR 181m (EUR 1,363m to EUR 1,544m), while operating costs grew EUR 242m (EUR 1,211m to EUR 1,453m). The EUR 61m profit fall is costs outgrowing revenue, 20% against 13%. Note one more line: flights grew 25% for 18% more passengers, so load factor fell from 88% to 84%.

McKinseyNordwind Airlines: the vanishing marginProfitability · Medium

Nordwind Airlines: the vanishing margin

The best first McKinsey-style case. Play it guided first, then live with the AI interviewer.

Interviewer

Our client is Nordwind Airlines, a short-haul carrier operating out of northern Europe. Over the past three years passenger numbers have grown 18%, but operating profit has fallen by roughly 40%. The CEO wants to understand what is driving this and what to do about it. How would you approach the problem?

Start the caseGuided, step by step

Kingsmere Hospitals: the surgery waiting list

The prompt. Kingsmere is a public hospital trust with two acute hospitals. Its waiting list for planned surgery has grown from about 6,700 to 9,600 patients in three years, and the board is about to approve GBP 9m for two new operating theatres. The chief executive wants to know what is really limiting surgical activity and what it would take to fix it.

A structure that fits. This is a capacity problem, so follow the patient. Name each stage, give it a capacity in one unit (operations a year) and look for the stage with the least. Treat the two theatres as an option to test, not the starting point.

First calculations. The list moves by the gap between the lines: in 2025, 13,400 added minus 12,000 done is 1,400, exactly the growth from 8,200 to 9,600. Since 2022, additions rose 1,000 (8%) while operations fell 600 (about 5%). Just to stop the list growing, Kingsmere needs 13,400 operations a year. Its target, which you get by asking, is about 5,600 within three years, so it must also clear 4,000 extra patients, about 1,330 a year: roughly 14,700 operations a year, about 23% more than today, if additions hold at 13,400.

McKinseyKingsmere Hospitals cuts its surgery waiting listOperations · Medium

Kingsmere Hospitals cuts its surgery waiting list

An operations case built on one unit across every stage. Find the binding constraint before anyone spends GBP 9m.

Interviewer

Our client is Kingsmere Hospitals, a public hospital trust running two acute hospitals. Its waiting list for planned surgery has grown from about 6,700 to 9,600 patients in three years, and the board is about to approve GBP 9m for two new operating theatres. The chief executive has asked us to find out what is really limiting surgical activity and what it would take to fix it. How would you approach this?

Start the caseGuided, step by step

Atlas Logistics: does a last-mile deal clear?

The prompt. Atlas Logistics, a regional freight and warehousing operator with $1.1bn of revenue, is considering buying Kestrel, a last-mile delivery company, for $340m. The CFO thinks the price is high. Should Atlas proceed?

A structure that fits. Keep what Kestrel is worth on its own apart from what it is worth to Atlas. Mixing the two is how buyers end up paying the seller for value only the buyer can create.

First calculations. Start with the CFO's concern. Kestrel's 2025 EBITDA is $210m × 9% = $18.9m, so $340m is 18.0x EBITDA. On its own record Kestrel looks solid rather than special: revenue grew from $182m to $210m, about 7% a year, and the margin edged up from 8.4% to 9.0%. Park two lines for later: 85% of the drivers are contractors, and the top three customers bring 38% of revenue.

McKinseyAtlas Logistics eyes a last-mile acquisitionM&A · Hard

Atlas Logistics eyes a last-mile acquisition

A hard M&A case: watch the base of every percentage, and decide how much synergy to pay away.

Interviewer

Our client, Atlas Logistics, is a regional freight and warehousing operator. They are considering acquiring Kestrel, a last-mile delivery company, for $340m. Their CFO thinks the price is high. Should they proceed?

Start the caseGuided, step by step

Northvale Energy: a motorway charging network

The prompt. Northvale, a national electricity utility supplying 3 million households, is weighing a $240m plan to build 150 ultra-rapid charging hubs along the motorways over three years. The CEO wants to know whether to go ahead.

A structure that fits. For an infrastructure entry, the total market matters less than whether one site makes money and how many good sites exist. Four questions, in order:

1. Market: how big is motorway charging demand now and in 2030, and how fast is it growing?
2. Hub economics: capex, margin per kWh, site costs and throughput, against the board's payback hurdle.
3. Site access: which sites can Northvale actually get, and what does each type sell?
4. Right to win and entry mode: build, buy, partner or supply power to other operators.

2028 and 2030 are the industry's central forecast. First calculations. From 2026 to 2030 the fleet is forecast to grow 3.3 times, from 1.2m to 4.0m: about 35% a year, since 3.33 to the power of a quarter is about 1.35. From 2022 to 2026 it grew 2.4 times, about 24% a year. So the forecast needs growth to speed up from about 24% to about 35% a year, which makes adoption the first sensitivity to test. The plan itself implies $240m ÷ 150 = $1.6m a hub.

McKinseyNorthvale Energy weighs a motorway charging networkMarket Entry · Hard

Northvale Energy weighs a motorway charging network

A hard market-entry case with four calculations. Throughput by site type decides it.

Interviewer

Our client, Northvale Energy, is a national electricity utility that generates power and supplies 3 million households. It is considering a $240m plan to build 150 ultra-rapid electric vehicle charging hubs along the country's motorways over the next three years. The CEO wants to know whether to go ahead. How would you approach this?

Start the caseGuided, step by step

Two parts of the McKinsey process are not cases at all, and both commonly carry real weight in the decision.

The PEI

The Personal Experience Interview is the part of each interview where you describe one experience and the interviewer probes it in depth, commonly for 10 to 20 minutes. Each PEI targets one theme. As commonly described these are personal impact, entrepreneurial drive, inclusive leadership and courageous change, though McKinsey revises the labels, so check your invitation. Prepare six to ten true stories with numbers, real resistance and your own actions at the centre. Our McKinsey PEI guide has sample questions and a story template, and the fit interview and PEI lesson teaches the method with checks.

Solve

Solve is McKinsey's online, game-based assessment, commonly sent after your application is reviewed. Candidates commonly report two or three games, most often Red Rock, Sea Wolf and the Sustainable Futures Lab, taking roughly an hour to an hour and a half in all. McKinsey's own description is less specific, so your invitation is the authority. Solve is not a case, but it rewards the same habits: read the objective, collect only the data you need and do the maths carefully. Our McKinsey hub covers the whole process, and the Red Rock Study guide walks through the data-analysis game candidates most often meet.

Run this list after every practice case. Anything you can't tick is your next drill.

• I asked two or three clarifying questions that changed my structure, not ten.
• My structure was built for this client, and I said where I would start.
• I said each calculation's setup before doing it, and finished with what the number meant.
• On each exhibit I led with the one insight that mattered, not a tour of every number.
• I answered the question I was asked, then stopped.
• My hypothesis changed when the data said it should, and I said so out loud.
• I closed with the answer first, three reasons with numbers, a risk and next steps.
• My PEI story held up through five rounds of follow-up questions.

Play each case twice. Guided mode puts the exhibits beside each question and gives feedback on every answer. Live mode, with the AI interviewer, is the closest thing to the real rhythm: questions you can't see coming and maths said out loud.