In a candidate-led case you drive the analysis: you decide what to look at next, and the interviewer mainly responds. In an interviewer-led case, the interviewer asks a set sequence of questions (structure, exhibit, maths, recommendation) and you answer each one well. The core skills are the same, but pacing, how much structure you lay out and what "leading" looks like are different, so practise both.
Which firms use which format?
As a broad, widely reported pattern, McKinsey is known for interviewer-led cases, while BCG and Bain are generally described as candidate-led. McKinsey's official guidance calls its case the "Problem Solving Interview" and pairs it with the Personal Experience Interview in client-facing interviews. Treat the BCG/Bain label as a tendency, not a rule. Styles vary by office and interviewer, and many boutique and Big Four strategy teams mix both. Ask your recruiter or contacts what to expect.
BCG's online case (the "Casey" chatbot) is effectively interviewer-led: the chatbot sets each question and you can't steer the case.
What does a candidate-led case feel like?
The interviewer gives you the prompt, you clarify and structure, and then you're expected to say where you want to go: "I'd like to start by understanding whether the decline is revenue- or cost-driven. Do we have a revenue breakdown?" The interviewer gives you data when you ask for the right thing, and may let you wander if you ask for the wrong thing.
What's being tested: prioritisation, hypothesis-driven thinking, and the ability to keep the case moving towards an answer.
Common failure: asking for data without a reason ("Can I see costs?") and losing the thread, so that after 25 minutes you haven't tested your main hypothesis.
What does an interviewer-led case feel like?
You'll typically get a prompt followed by a series of discrete questions:
- "What factors would you consider?" (structuring/brainstorming)
- "Here's an exhibit. What do you take from it?"
- "Calculate the expected annual savings."
- "What would you tell the CEO?"
What's being tested: depth and quality on each question, creativity in brainstorming, and accurate quantitative work. Each question is scored in its own right.
Common failure: building a giant opening structure that the interviewer then ignores, or giving shallow, list-like brainstorms.
How should you adapt your structure?
For candidate-led cases
Your opening structure is your roadmap. Make it hypothesis-driven, prioritise a starting branch, and refer back to it: "We've ruled out pricing, so let's move to the cost branch." Summarise every few minutes.
For interviewer-led cases
Keep the opening structure clear but proportionate, because the interviewer will direct the next steps anyway. Prepare for mini-structures. When asked "What are the risks of this acquisition?", structure your answer into categories (financial, operational, regulatory, people) instead of listing ideas at random.
How should you adapt your pacing?
In candidate-led cases, you set the pace. Watch the time and push towards a recommendation by around the 25–30 minute mark, even if your analysis is incomplete. In interviewer-led cases, the interviewer sets the pace, so your job is to answer crisply and avoid over-running on any one question.
How do you show leadership in an interviewer-led case?
Even when the interviewer asks the questions, you can show case leadership:
- Link each answer to the overall objective ("This matters because...").
- Close each answer with a "so what".
- Suggest what you'd look at next, even if the interviewer then redirects you.
How do you avoid getting lost in a candidate-led case?
- Keep a visible issue tree on your page and tick off branches.
- State your current hypothesis explicitly and update it as data arrives.
- If you're unsure where to go, summarise what you know and ask yourself what would most change your answer.
How should you practise both formats?
- Use firm materials. McKinsey publishes practice cases on its careers site (at the time of writing: Beautify, Diconsa, Electro-Light, a national education system case and Talbot Trucks). These show the interviewer-led rhythm.
- Ask partners to run the format you need. Tell your practice partner, "Run this candidate-led, and only give me data when I ask for something specific."
- Mix in AI practice. An AI interviewer such as caize.ai can run cases in a structured, question-by-question flow or let you drive, which is useful for getting volume in both styles before human mocks.
- Practise the online case. For BCG specifically, practise timed chatbot-style cases, as there are no hints and the clock is tight.
FAQ
Is one format easier than the other?
Neither is objectively easier. Candidate-led cases demand more self-direction, and interviewer-led cases demand depth on each question.
Does McKinsey always use interviewer-led cases?
That is the widely reported norm, but individual interviewers vary, and McKinsey has piloted AI-assisted cases that some observers describe as more free-flowing. Prepare to lead either way.
Do I still need a framework in an interviewer-led case?
Yes, but structures will be shorter and question-specific rather than one big tree.
How do I know which format I'll get?
Ask your recruiter, check firm websites, and speak to current consultants in your target office.
