Interviewer-Led vs. Candidate-Led Case Interviews: What Changes

Candidates prepping for multiple firms discover mid-process that "the case interview" is really two different formats. McKinsey typically runs interviewer-led cases; BCG and Bain lean candidate-led (with plenty of variation by office and interviewer). The underlying skills overlap heavily, but the format changes how those skills get graded — and a candidate who has only practiced one format can genuinely stumble in the other.

What happens in an interviewer-led case

The interviewer owns the agenda. You get the prompt, deliver a structure, and then the interviewer walks you through a planned sequence: "Let's look at the market first — here's an exhibit," then a math question, then a brainstorm, then a synthesis. Each segment is a discrete question with a fairly defined right answer, and the interviewer will move you along regardless of where your structure pointed.

What this format grades hardest: question-by-question excellence. Crisp structures on demand ("what factors would you consider?"), clean narrated math, level-4 exhibit reads, and the ability to switch topics without losing the thread of the overall case. The comfort is that you can't get lost — the discomfort is that there's nowhere to hide a weak segment, because every candidate gets the same segments.

What happens in a candidate-led case

You own the agenda. After your opening structure, the interviewer asks the quietly terrifying question: "So where would you like to start?" — and from then on you choose the branch, request the data, interpret what you get, decide what that means for where to go next, and eventually declare yourself ready to recommend. The interviewer mostly follows you, supplying information when asked.

What this format grades hardest: drive. Passivity is the classic failure — candidates who analyze whatever is in front of them competently but never push toward an answer, waiting for a prompt that isn't coming. The format also punishes structure-abandonment: your opening framework is supposed to be the map you actually navigate by, and interviewers notice when it was ceremonial.

The three skills that transfer, and the one that doesn't

Structuring, case math, and exhibit reading transfer between formats almost unchanged — they're graded in both, just packaged differently. The skill that doesn't transfer automatically is navigation: explicitly closing a branch ("prices are flat, so the revenue problem is volume — that rules out pricing"), stating what that implies, and naming the next move. In interviewer-led cases the interviewer navigates for you; in candidate-led cases navigation is the test. If your practice has been mostly interviewer-led (most casebooks are written that way), this is the muscle you haven't trained.

How to prep when your target firms use both

• Train the shared core first. Structures, math, exhibits — they're most of the grade in both formats. The drills in our math and exhibits guides apply everywhere.
• Then get reps in each format deliberately. Tell your practice partner or coach which format to run. In Key Takeaways, cases run interviewer-led, as a live spoken conversation with the AI interviewer: use them for the shared core, say the transition sentence at every branch close anyway, and let the scored breakdown show whether your structure and synthesis held up.
• Drill the transition sentence. The navigational tic that makes candidate-led cases feel controlled: "So we've established X, which means Y — next I'd want to look at Z." Say it at every branch close until it's automatic. It also makes interviewer-led answers sound more senior.
• Match your mock mix to your interview schedule. Heavier candidate-led reps before BCG/Bain rounds, heavier discrete-question reps before McKinsey — final week only; before that, the mix should follow your weaknesses.

The bottom line

Same skills, different steering wheel. Build the core once, then practice both formats on purpose — and if you've never heard "where would you like to start?" under pressure, make sure the first time isn't in the real room.