You have prepared your structure, your calculations and your synthesis, and the interview starts differently from what you expected: the interviewer asks a first precise question, then a second, then a third, without letting you sketch the whole problem. Or, the other way round, they hand you the prompt, say nothing and wait for you to move forward alone. Both situations exist, forums call them candidate-led and interviewer-led, and they do not call for the same pace. This article explains the logic of each, how to recognise the format in the first minutes and how to adapt your preparation.
Two formats, two logics
The common ground is well known: a case interview presents a realistic business situation and assesses the quality of your reasoning. The Boston Consulting Group sums it up on its case interview preparation page 🔗: the candidate must structure their approach, ask thoughtful questions, analyse data and do quick calculations, while communicating clearly. McKinsey, for its part, describes on its interviewing page 🔗 a "Problem-Solving Interview" that presents a business scenario to judge how you approach a complex problem.
What these pages do not say is who holds the wheel. That is the difference between the two formats:
- the candidate-led case: the interviewer sets out the problem, then you decide the order of the analyses, the data to ask for and the moment to conclude;
- the interviewer-led case: the case is cut into successive questions (structure, estimate, read a chart, recommend), and the interviewer decides how they follow one another.
No firm needs to commit to a single format, and practice varies by office, interviewer and round. Treat these two models as poles, with many cases sitting somewhere between them. The skills assessed are the same; what changes is where you are expected to shine: on initiative in one, on precision and responsiveness in the other.
The candidate-led case
In this format, the interviewer states the situation ("An industrial group sees its profit fall; what do you recommend?"), answers your framing questions, then hands you the floor. They intervene little: to give a figure you asked for, to nudge you if you drift, to test a hypothesis.
What this demands of you:
- structure alone: propose a framework, get it validated in a few words, then follow it; the library of frameworks is a starting point, never a mould;
- announce your next step: "I will start with the breakdown of revenue, then look at costs" gives the interviewer something to follow;
- ask for data instead of waiting for it: without a request, nothing arrives;
- manage time: nobody will tell you ten minutes are left; it is up to you to decide when to stop exploring and when to conclude;
- conclude on your own initiative, with a recommendation, two or three reasons and the main risks.
The natural trap is dispersion: exploring every branch of the tree without ever prioritising. A candidate who steers well links each analysis to the hypothesis being tested, and quickly drops a lead that goes nowhere.
The interviewer-led case
Here, the case looks like a series of mini-exercises. The interviewer asks a precise question ("How would you structure the analysis?", "Estimate the size of this market", "What does this chart tell you?"), lets you answer, then moves on to the next. The thread of the case is in their head, not in yours.
What this demands of you:
- answer the question asked, not the one you would have preferred: a full structure when you are asked for a figure wastes everyone's time;
- stay precise and concise: the answer often fits in a minute or two, with the conclusion in the first sentence;
- handle a calculation with no warm-up, because the question comes without preparation; the article on mental maths in consulting interviews details the techniques that hold up;
- keep the overall view despite the slicing: tie each answer back to the general thread of the case, for instance by saying in one sentence what your figure changes for the recommendation;
- react quickly to a follow-up or a correction, without defending yourself.
The trap is the opposite of the previous one: answering mechanically, question after question, without ever saying where you are heading. On the last question ("What do you recommend to the client?"), you still have to produce a synthesis that holds across the whole case.
Recognising the format in the first minute
Nobody announces "this case will be led by me". The signals are in the first minutes.
In the prompt. An open prompt, with a single objective and few details ("How can this client restore profitability?"), points rather to a case where you steer. A prompt that ends on a narrow question ("What questions would you ask before starting?") points rather to a guided sequence.
In the silence. After your structure, if the interviewer says nothing and waits, the floor is yours. If they immediately move on with "All right, let us calculate the revenue of segment A", they are the one driving.
In the nature of the interventions. An interviewer who gives you data only when you ask leaves the initiative to you. One who produces a chart or a table at the right moment sets the path.
If doubt remains, remove it plainly. After rephrasing the prompt, suggest: "I propose to start with the structure, then move to the numbers; tell me if you would prefer another order." The answer settles it in ten seconds, and framing things this way is itself a sign of maturity. BCG's pages, incidentally, list clarifying assumptions and active listening among the good reflexes.
One last point: the format can change along the way. A case starts open, then the interviewer takes back control to speed things up. Do not cling to your plan if it is taken away.
Adapting your preparation
Preparing for a single format makes you fragile on the other. A few pointers for working on both.
For the case where you steer, practise out loud, timer in hand, with a prompt your partner does not elaborate on. Work on three gestures: setting out a structure in two minutes, announcing each step, and deciding on a conclusion at a time fixed in advance. Profitability cases suit this well, because the breakdown tree forces you to prioritise.
For the case where the interviewer chains questions, work on isolated questions: an estimate, a chart to comment on, a recommendation in thirty seconds. Time each answer. Ask your partner to put the questions in an order you do not know.
For both, practising in pairs remains the most effective way: ask your partner to switch format from one session to the next, even in the middle of a case, and to tell you what they observed about your pace. Also keep a record of your sessions: format played, weak point, correction decided.
Finally, think about your application. The case is only one step; what you will be asked in the fit interview and at the first CV screening rests on clear bullets that set out problem, method and result. The strategy consultant CV template shows this way of writing. The ConseilCV editor takes care of the layout for you and produces a PDF that firms' screening software reads correctly, which leaves your preparation time for the cases.
Mistakes specific to each format
When you steer:
- not announcing your plan and leaving the interviewer to guess where you are going;
- asking ten framing questions in a row without ever proposing a hypothesis;
- waiting to be given the data instead of asking for it;
- not watching the time and reaching the conclusion by improvising;
- pursuing an unproductive branch for fear of "losing" the analysis already done.
When the interviewer leads:
- treating every question as a chance to recite a full structure;
- answering without starting with the conclusion, which forces the interviewer to look for it;
- losing the thread of the case and giving, at the end, a recommendation unrelated to your figures;
- bristling when the interviewer corrects or redirects you: it is often help;
- staying silent during a calculation, when talking while calculating lets the interviewer correct an error on the way.
In both cases, the costliest flaw is the same: ignoring what the format asks of you at that precise moment. Read the situation, adjust, and do not replay a learned script.
Frequently asked questions
Which case format is the most common?
It cannot be stated in general: practice varies by firm, office, interviewer and round, and firms publish no statistics on the subject. Prepare for both rather than betting on one.
Should I ask the interviewer which format they use?
Not directly, but you can propose your sequence ("structure first, then numbers") and leave them the option of changing it. Their reaction tells you what you need without making you look hesitant.
Can the format change within a single case?
Yes, and it is common. An open case can become guided if time runs short, or the other way round if the interviewer wants to see your autonomy. Follow the movement without insisting on your initial plan.
Should I use a framework in each format?
In both, a tailored structure beats a generic model. It weighs more when you steer, because nobody completes it for you.
Sources
- Boston Consulting Group, Case interview preparation 🔗
- McKinsey & Company, Interviewing at McKinsey 🔗