McKinsey Case Interview Format: What Actually Happens in the Room

A McKinsey case interview runs 45 to 60 minutes and has three distinct phases: the opening, the analysis, and the recommendation. Each phase tests a different skill. Failing one phase doesn't end the interview, but stumbling in the same way across all three signals a pattern, and patterns get remembered.

Think of it like a cooking competition where the judges are scoring three separate rounds: your prep work, your cooking, and your plating. A great dish presented badly scores lower than a good dish presented confidently. McKinsey is scoring all three rounds.

There's also a personal experience interview (PEI) woven in, usually 15 minutes at the start or end. This guide focuses on the case portion, but know that the PEI is scored just as seriously as the case. Interviewers share notes.

The interviewer reads you a prompt. It's usually two to four sentences describing a client situation and a question. Then they stop and wait.

What most candidates do: start talking. What top candidates do: ask one clarifying question, then take 60 seconds of silence to structure.

Here's why the silence matters. The interviewer is watching whether you default to comfort or to rigor. Talking immediately signals that you'd rather appear engaged than actually be right. Taking time signals that you understand the cost of going in the wrong direction.

The clarifying question has a specific purpose: confirm the objective. Not "can you tell me more about the client?" but "when you say they want to grow profitability, are we focused on the next 12 months or is this a multi-year strategic question?" One question. Targeted. Then structure.

When you present your structure, walk the interviewer through it out loud: "I've broken this into three areas: revenue, cost, and market dynamics. My working hypothesis is that the problem sits on the cost side, specifically fixed costs, but I want to validate that against revenue data first. I'd like to start with revenue. Can you share what's happened to volume and price over the past two years?"

The interviewer is taking notes. That last sentence, asking for specific data, is what marks a candidate as someone who can run a project.

Once you have data, you're in the analysis phase. This is where the case usually lives for 20 to 30 minutes. The interviewer hands you exhibits (charts, tables, numbers) and watches what you do with them.

Most candidates make two mistakes here. First, they describe the exhibit instead of interpreting it. "Revenue declined from $4.2B to $3.8B" is a description. "Revenue declined 9%, but volume only fell 3%, which means price realization is down more than volume, suggesting a pricing or mix issue rather than demand" is an interpretation. McKinsey interviewers are trained to spot the difference.

Second, candidates drop their structure when the data gets complicated. They start following the data wherever it leads instead of testing their hypothesis. Think of it this way: you're not a tourist who goes wherever the signs point. You're a navigator who knows the destination and uses the signs to confirm or revise the route.

When you get a new exhibit, do three things before you speak. First, identify what the chart is measuring. Second, find the most important number or trend. Third, connect it back to your hypothesis: does this support it, contradict it, or not tell you anything yet?

The interviewer is scoring: structured thinking, quantitative comfort, hypothesis discipline, and communication. They're not scoring whether you got the "right answer" on any individual exhibit. They're scoring whether your thinking process is one they'd trust in front of a client.

Practice this on a case like Netflix's 2007 strategic pivot, a case with multiple exhibits where the data pulls you in different directions before the real lever reveals itself.

The interviewer cuts the analysis phase and asks for your recommendation. Usually there's been no warning. "Okay. So what would you tell the client?"

This is the plating round. You have 90 seconds to turn everything you've done into a crisp, defensible answer.

Bad recommendation: "Based on everything we've discussed, there are a few factors to consider. On the revenue side we saw some decline, and on the cost side there were some issues. Overall I'd say the company should probably focus on costs, but they should also think about revenue."

Good recommendation: "My recommendation is that the client should cut fixed costs in the hardware division by 20%, primarily through renegotiating manufacturing contracts. This addresses the primary driver of the margin decline, specifically fixed cost absorption, which the data showed is the real issue, not demand. The risk is supplier relationship damage, which I'd mitigate by offering longer contract terms in exchange for lower rates."

The structure is: one-sentence recommendation, then the primary reason stated with a number, then the risk and how to manage it. Three beats. If you can't do it in three beats, you haven't synthesized. You've summarized.

McKinsey uses a structured scorecard. Interviewers score candidates on four dimensions, each on a scale.

Problem-solving: Can you move from an ambiguous question to a rigorous answer? Do you form hypotheses or just collect data?

Personal impact: Are you convincing? Would a CFO trust your recommendation? Confidence and clarity of communication. This is about how you say it, not just what you say.

Leadership: Do you drive the conversation, or do you wait to be led? Do you proactively share what you need and why?

Entrepreneurial drive: Is there energy behind your thinking? Does it feel like you actually care about getting the answer right?

None of these are about getting the case "correct." McKinsey knows the answer. They're watching how you think, not what you conclude.

Practice the opening separately. Take 20 prompts. For each one, practice only the first 90 seconds: one clarifying question, 60 seconds of silence, structured walkthrough. The opening sets the tone for the whole interview. It deserves its own drill.

Time your recommendations. After every practice case, write your recommendation in 90 seconds without referencing your notes. If you can't do it, you didn't synthesize. You just analyzed.

Practice with real data exhibits. Most case prep uses verbal-only cases. McKinsey almost always has exhibits. Get comfortable reading a chart, identifying the key number, and interpreting it out loud in under 30 seconds.