Interviewer-Led vs Candidate-Led Cases: Two Different Games

In an interviewer-led case, the interviewer sets the agenda and feeds you data section by section, steering the conversation toward whatever they want to test next, which is the format McKinsey is best known for. In a candidate-led case, you own the direction, decide what to analyze first, and pull data by asking targeted questions, the format BCG and Bain lean toward. The two formats test overlapping but different skills: interviewer-led rewards adaptability and listening for redirection, candidate-led rewards independent prioritization and comfort setting your own agenda. Most candidates prepare for one and get caught flat-footed by the other.

I conducted well over a hundred interviews at McKinsey, all interviewer-led, and I have read enough candidate debriefs and prep material from the candidate-led firms to know exactly where the two formats diverge and where candidates from one background stumble in the other. This is the breakdown.

What actually changes between the two formats

The case content barely changes. A profitability problem is a profitability problem regardless of who is steering. What changes is who holds the steering wheel.

In an interviewer-led case, the interviewer typically has a case document with specific exhibits and a specific sequence in mind. You build your structure, propose a starting branch, and the interviewer often redirects you: "That's reasonable, but let's look at the cost side first" or "Before that, here's an exhibit on market share." The case moves in a series of prompts and responses, closer to a structured conversation with a clear conductor.

In a candidate-led case, the interviewer gives you the prompt and then largely waits. You propose the structure, you propose which branch to open first, and you ask for the data you think you need. The interviewer answers what you ask, sometimes offering a nudge if you're stuck, but the sequencing is yours to build and defend. If you never ask for a piece of data that turns out to matter, in a candidate-led case you might simply never see it. That is a real risk the interviewer-led format does not have, since there the interviewer will surface the exhibit regardless.

Interviewer-led cases: what is actually being tested

Because the interviewer holds the sequence, the skill under a microscope in an interviewer-led case is how you handle redirection. I watched candidates build a perfectly reasonable structure, get redirected toward a branch they hadn't prioritized, and either adapt smoothly or visibly resist because they were attached to their own plan.

The strong response to redirection is quick and calm. Something like, "Got it, let's start with costs then, I'll come back to the revenue side after," said in the same breath as pivoting, with no defensiveness and no re-litigating your original plan. The weak response tries to explain why your original branch was actually the better starting point. That reads as someone who cannot take direction, which is a bad signal for a job that is entirely about taking direction from partners and clients.

Interviewer-led cases also test how well you listen for embedded hints. Interviewers steering the case will often plant a clue in how they phrase a redirect or which exhibit they hand you next. If they hand you a cost breakdown unprompted, that is rarely an accident. Candidates who treat every exhibit as neutral information miss half the signal being sent to them.

Candidate-led cases: what is actually being tested

A candidate-led case is testing something closer to how you would behave with an actual client in the room: given an ambiguous problem and no one telling you what to do next, can you decide what matters, ask for it directly, and adjust your own plan as new information comes in.

The skill most exposed here is prioritization under silence. Nobody will rescue you if you ask for the wrong data first. If your structure has four branches and you spend fifteen minutes down the least important one because nobody redirected you, that time is gone and the case clock does not pause to give you a second chance. Interviewers running candidate-led cases are watching whether you propose a starting point with a reason attached, not just any starting point.

The second skill is question quality. In an interviewer-led case, you can often get by with a reasonable structure and let the interviewer's exhibits do some of the navigating for you. In a candidate-led case, vague questions get vague or unhelpful answers. Asking "can I see some data on the market" gets you less than asking "can I see the revenue split by customer segment for the last three years." The specificity of your questions is graded as directly as your math.

Interviewer-led vs candidate-led, side by side

Where candidates get tripped up in each format

Candidates who only practiced with McKinsey-style partners in interviewer-led mode tend to freeze slightly in a candidate-led case, because nobody hands them the next exhibit. The silence feels like a test failure even when it isn't. The fix is to treat that silence as normal working time, the same way you'd sit with a blank page before a real client meeting, and to narrate your thinking so the interviewer can follow along even while you're deciding what to ask for next.

Candidates who only practiced candidate-led cases sometimes struggle in an interviewer-led one for the opposite reason. They're used to owning the whole sequence and treat a redirect as an interruption to push back on, when it should be treated as new information to fold in immediately. If an interviewer-led interviewer redirects you twice and you keep steering back toward your original plan, that reads as rigid, not confident.

Here's a short checklist for handling either format cleanly:

• Ask early in the interview, or infer from the firm, which format you're likely facing, since McKinsey leans interviewer-led and BCG and Bain lean candidate-led
• If redirected, acknowledge in one sentence and move, never defend your original branch
• If the case goes quiet, treat it as your working time, not as a signal you're doing something wrong
• Ask for data using full, specific requests instead of general ones
• Keep a running note of open questions so you can circle back to a branch you set aside
• At the end, synthesize everything you covered into one recommendation regardless of who chose the path to get there

How to adapt your structure in real time

The core move that works in both formats is the same: build your structure so that any branch could plausibly be entered first, rather than building a rigid sequence you're mentally committed to. If your issue tree has revenue and costs as top branches, you should be equally ready to start on either, because in an interviewer-led case the interviewer might pick for you, and in a candidate-led case you need a real reason to pick one yourself rather than defaulting to whichever branch you always start with.

Practice building the same structure and mentally rehearsing two different opening lines from it, one where you propose starting with revenue and one where you propose starting with costs, each with a one-sentence reason. That rehearsal is what makes redirection feel like a minor adjustment instead of a derailment, and it's the same skill that makes candidate-led prioritization feel like a choice instead of a coin flip. For the full method on building that kind of flexible structure from scratch, see how to structure a case interview, and for the underlying frameworks that feed into it, the case interview frameworks guide is the deeper reference.

Practicing for both formats

If you only have practice partners from one firm background, you will only get reps in one format. Ask your practice partner directly to run the case as interviewer-led one session and candidate-led the next, using the exact same case prompt both times. You'll notice quickly which format exposes your weaker instinct. If you tend to over-explain when redirected, that shows up in the interviewer-led run. If you tend to wander without a clear next question, that shows up in the candidate-led run.

It also helps to watch or read a full case end to end before your first mock, so you have a mental model of pacing in both formats rather than guessing. A full sample case interview, annotated line by line walks through exactly where structure, redirection, and data requests happen in a real case, and what is a case interview is the right starting point if you're still building the basic mental model before worrying about format differences. Once you've got the shape down, a rotating set of practice prompts helps more than memorizing one case cold, and 25 case interview questions with answer outlines gives you enough variety to drill both formats without repeating the same prompt.

The bottom line

Interviewer-led and candidate-led cases test the same underlying business judgment through two different lenses, one measuring how well you take direction, the other measuring how well you set your own. Build a structure flexible enough to enter from any branch, treat redirection as new information rather than an interruption, ask for data in specific terms when nobody is handing it to you, and practice both formats with a real partner before you assume you know which one you'll get. If you want the mechanics of building that flexible structure down cold before you worry about who's steering, start with how to structure a case interview.

Go deeper

Format is only half the battle. The CaseMap method inside the course is built to work regardless of who's steering the case, because it trains the structure itself rather than a script tied to one interviewer style.

Get the free resume module →

Get the complete Cut to the Case course →

One payment, lifetime access, 30-day guarantee.