In an interviewer-led case, the interviewer usually chooses the next question. In a candidate-led case, you take more responsibility for deciding what to investigate next. Both require you to explain your reasoning and turn findings into a useful answer for the client.
The practical difference appears at the end of an analysis. Do you answer the question you have just been asked, or propose the next question worth answering? You need to be comfortable doing both, because a conversation can change direction halfway through.
I would practise these as two behaviours rather than memorising two separate ways to solve a case. Your business reasoning should survive either format.
When the interviewer chooses the question, finish that question
Suppose the client is considering a price increase. You have proposed examining customer response, economics and implementation. The interviewer says, “Let's calculate what happens to monthly contribution.”
Calculate contribution. Your structure does not give you permission to spend another three minutes explaining customer segments before answering.
A complete response has an approach, a result and an implication:
“I'll compare subscribers multiplied by contribution per subscriber, before and after the change. That gives £80,000 before and £84,600 after: a £4,600 monthly increase. The increase is financially attractive for this cohort, provided the retention assumption holds.”
Then give the interviewer room to respond. They may ask about retention, introduce a chart or move to implementation. You can carry the £4,600 finding into any of those tasks without insisting on your preferred sequence.
The weak version is a number with no interpretation. The equally weak version is answering a different question because you planned to cover it next.
When the next move is yours, explain why it matters
Now imagine the interviewer says, “What would you look at first?” You need to make a choice.
“I'd like to look at customers” names a topic. It does not explain what you want to learn or how that knowledge changes the decision.
A better response is:
“The price increase works only if enough customers renew. I would first estimate retention at the new price, then compare the contribution from those retained customers with today's contribution. Do we have evidence from a renewal test?”
This is a manageable request. It names an uncertainty, links it to the decision and asks for relevant evidence. The interviewer can supply data, tell you it is unavailable or challenge your priority.
You are leading a discussion, so leave space for that response. An uninterrupted tour through every branch of your framework makes it harder to solve the case together.
Quick Prioritisation Drill
A ferry operator is considering a new late-evening sailing on its island route. The interviewer asks what you would look at first.
- Cost per sailing
- £1,800, fixed by contract
- Average fare
- £12 per passenger
- Ship capacity
- 400 passengers
Which question should come first to decide whether the sailing pays?
Use the answer before asking for more data
Suppose you learn that 94% of customers would renew. Say what that means before reaching for another exhibit:
“We would retain 9,400 of the 10,000 customers. I now want to see whether the extra contribution per retained customer offsets the 600 lost subscriptions.”
If the next calculation supports the increase, close that part of the analysis:
“The economics improve for this cohort. The remaining decision is whether the same renewal response is likely across the rollout, and how we introduce the change.”
You have moved from uncertainty to evidence to a narrower set of remaining questions. That progression is what makes a case feel directed.
Pause to state what the evidence now supports when you find yourself collecting facts without changing your view. A short conclusion gives the next question a reason to exist.
Respond when control shifts
A candidate-led discussion can contain a very specific calculation request. An interviewer-led discussion can end with an open invitation to identify risks or recommend action.
| What you hear | What I would do |
|---|---|
| “What does this chart tell you?” | Interpret the chart and its implication before requesting another exhibit. |
| “Which issue would you investigate next?” | Select a question, explain its importance and ask for the evidence. |
| “Assume demand is sufficient. What about capacity?” | Accept that instruction for the exercise and analyse capacity. |
| “We have two minutes left.” | Bring the current findings together and make a supported recommendation. |
| “Is there another way to approach this?” | Offer a genuinely different route and explain what it would reveal. |
Treat a redirection as new information about the task. You can still flag a material concern briefly: “That works if the demand assumption holds. Taking it as given, I would now check whether we can supply it.” Then move on.
Quick Answer Repair Drill
You are leading a case on a meal-kit subscription firm whose profit has fallen. Your analysis has produced these findings.
- Profit decline
- £2.0m this year
- From Northern cancellations
- £1.5m
- Northern change
- New courier since March
Told “two minutes left”, a candidate asks for cancellations by recipe. What is the flaw?
Avoid turning firm labels into a script
You will hear broad descriptions of McKinsey as interviewer-led and BCG or Bain as candidate-led. Those labels can help you select practice material. They cannot tell you exactly what the next person will ask.
Read your invitation and the relevant office's instructions. Practise taking the interviewer through your reasoning, listening and drawing intermediate conclusions, whichever person chooses the next task.
A direct question takes priority over a memorised firm stereotype. An open question calls for initiative even if you expected the interviewer to lead.
Rehearse the same case in two ways
Choose one unfamiliar problem and a partner. In the first run, ask your partner to select the sequence: structure, calculation, interpretation, recommendation. Your job is to answer each task and connect it to what you have already learned.
In the second run, use fresh numbers or a different business. Your partner should supply information when requested and leave you to choose the next useful analysis. Review whether each request had a purpose and whether you used the answer.
For a focused starting point, work through a customer-retention problem, then a capacity decision. Ask your partner to direct the first discussion and let you choose the next investigation in the second.
Frequently asked questions
Telling the two formats apart
What is the real difference between interviewer-led and candidate-led cases?
The main difference is who chooses the next question. In an interviewer-led case, the interviewer usually selects it. In a candidate-led case, you take more responsibility for deciding what to investigate. The difference shows at the end of each analysis: you either answer the question you were just asked or propose the next question worth answering.
Is McKinsey always interviewer-led and BCG or Bain always candidate-led?
Those are broad descriptions, not reliable rules. They can help you choose practice material, but they cannot tell you exactly what the next interviewer will ask. Read your invitation and the relevant office's instructions for specifics. Prepare to handle both styles rather than rehearsing a single script based on a firm's reputation.
Can a case switch between the two styles partway through?
Yes. A candidate-led discussion can include a very specific calculation request. An interviewer-led discussion can end with an open invitation to identify risks or recommend action. Listen to what you are being asked at each point. A specific request needs a direct answer. An open prompt needs you to choose an issue and justify it.
Which skills matter in both formats?
Some skills matter whoever chooses the next task: taking the interviewer through your reasoning, listening carefully and drawing intermediate conclusions. Both formats also require you to turn findings into a useful answer for the client. Practising these habits prepares you better than memorising which firm is supposed to use which format.
When the interviewer sets the task
I have already laid out a structure. Can I walk through it before doing the calculation I was asked for?
No. A direct question takes priority. Suppose you proposed examining customer response, economics and implementation, and the interviewer then asks for monthly contribution. Calculate contribution first. Your structure does not give you permission to spend time on customer segments before answering. You can return to the other areas once the requested task is complete.
What does a complete answer to a calculation request include?
It includes an approach, a result and an implication. State how you will calculate it, for example subscribers multiplied by contribution per subscriber. Give the figure, then say what it means for the client's decision. A number without an implication leaves the interviewer to do the interpretation that the question was testing.
When you choose the next step
When asked ‘What would you look at first?’, is naming an area enough?
No. ‘I'd like to look at customers’ names a topic, but it does not say what you want to learn or how it changes the decision. Link the analysis to the decision instead. A price increase works only if enough customers renew, so you would first estimate retention at the new price. That explains both your choice and its relevance.
Should I ask for more data as soon as I get a number?
No. Say what the number means first. If 94% of 10,000 customers would renew, state that you retain 9,400 and lose 600. Then explain your next question: whether the extra contribution per retained customer offsets those lost subscriptions. Interpreting before you ask shows the interviewer why the next exhibit is worth requesting.
When should I close off one part of the analysis?
Close it once the result answers the question that part of the analysis was meant to test. If the next calculation shows that the economics improve for the cohort, say so explicitly before moving on. A clear intermediate conclusion tells the interviewer what you have established. It also stops the case drifting into further data requests with no clear purpose.
Practising both styles
How can I practise both styles with a partner?
Work through an unfamiliar problem twice. In the first run, your partner chooses the sequence: structure, calculation, interpretation, recommendation. You answer each task and link it to what you have already learned. In the second run, your partner supplies information only when asked and leaves you to choose the next useful analysis.
Why use fresh numbers or a different business for the second run?
If you repeat identical inputs, you can rely on memory of the first run rather than judgement. Changing the numbers or the business makes you decide again which analysis is worth doing next. That is the skill the candidate-led run is meant to test. It also shows whether your approach works in a new situation.