Practise McKinsey cases with a stronger partner.

McKinsey describes problem-solving, personal-experience and role-dependent expertise interviews. Use your recruiter’s instructions to choose what you need to practise. Generalist case practice will not cover every requirement of a specialist role.

For an interviewer-led problem-solving session, agree that your partner will guide the sequence of questions. Your job is still active: clarify the objective, build a useful approach, interpret the data and explain implications. When the interviewer moves to a new question, follow it. You can preserve the overall objective without forcing the conversation back to your preferred branch.

The official examples are a useful starting point because they show the kinds of questions a partner needs to handle. Prepare them from the interviewer side as well as the candidate side.

Start with your university, alumni network or a peer platform such as PrepLounge. Ask for one trial session with a clear exchange.

I'm preparing for a McKinsey generalist interview in English. I can practise Wednesday, 19:00–20:30 IST. I'd like one case each, with ten minutes of feedback after each attempt. My current focus is explaining calculations and answering follow-up questions directly. Can we each prepare an unfamiliar case to give the other person?

Replace the details with your situation. Include any casebooks you have already worked through so the partner can avoid familiar material. If you only have 45 minutes, take turns receiving a case across two appointments. Protect the debrief.

Before giving a case, solve the numerical work yourself. Separate the opening prompt from information available on request and from later exhibits. Identify what each question is trying to reveal: a structure, a calculation, a business interpretation or a recommendation.

Know the case well enough to distinguish a different valid approach from a genuine error. Otherwise your feedback can punish sound reasoning simply because it differs from the answer sheet.

Do not rescue your candidate the instant they pause. Give them room to reason. If you provide a hint, note what help was needed and discuss it afterwards. That creates a better diagnosis than pretending the answer was fully independent or turning the session into a contest.

Here is an original short practice exercise. A retailer is considering a self-service returns portal to reduce processing costs while preserving customer service.

First ask how the candidate would assess it. A useful approach covers eligible returns, likely adoption, genuinely avoidable processing cost, software cost and the effect on service. “Customer, company, competition” would need substantial explanation before it could guide this decision.

Next provide the numbers. There are 100,000 eligible returns per year. The portal is expected to handle 40%. Each migrated return saves £4 in avoidable processing cost, and the portal costs £100,000 annually. Ignore setup costs initially.

The annual benefit is 40,000 × £4 − £100,000 = £60,000. Now ask what happens if adoption is only 20%. The answer becomes an annual loss of £20,000. Break-even is 25,000 migrated returns, or 25% adoption.

Then introduce a £90,000 setup cost. At expected adoption, first-year net benefit becomes negative £30,000. A candidate who previously recommended immediate rollout should update the recommendation or explain the investment horizon. The £60,000 recurring benefit has not disappeared; the first-year economics have changed.

Finally ask what a pilot must establish. Good answers investigate adoption, resolution rates, repeat contacts and whether the £4 saving can actually be removed from the cost base. Faster processing does not automatically reduce staffing expense.

The useful numerical response is more than “£60,000.” It might be:

At 40% adoption, the portal saves £60,000 annually after its running cost. Adoption needs to exceed 25% to break even before setup costs. I would test adoption and achievable processing savings before committing to rollout.

Your partner should notice whether you labelled the time period and costs included. They should also notice whether your recommendation changes when the assumptions change.

Use demonstrations to study clarification and interpretation alongside calculation. Pause at a question, answer it yourself and compare what each answer helps the client decide.

A case session can crowd out personal-experience work. Book a separate appointment if that is what it takes to give one story proper attention.

Ask your partner to interrupt with specific questions: What did you personally do? What did the other person want? Why did you choose that approach? What would have happened if you had done nothing?

Suppose you describe persuading a team to reduce a launch's scope. “I aligned stakeholders” hides the important material. Explain whose support was necessary, why they resisted, what evidence changed the discussion and which compromise you accepted. Keep the facts stable while answering in a different order.

At the end, choose the mistake with the greatest effect on the answer. Replay that moment immediately. For the portal case, that might mean restating the recommendation after learning about setup cost.

Test the correction later on a different business. Keep a regular partner who notices your habits, and add unfamiliar partners who make you explain everything clearly from scratch. If coordination becomes difficult, use the small practice-group format to share the work.