1500755700 1480091280

Candidate-led Case Interview

By Kim Tran October 17, 2016Case Interview2 Comments

Candidate-led cases are interesting but also very difficult, especially for beginners. I personally really like these cases because they represent very well how an actual consulting project works and the role of the engagement team in it.

So rather than talking specifically about how a candidate-led case works, let me introduce you to the LOGIC behind problem solving in management consulting. It’s ironic that many experienced candidates, having practiced 20 to 30 cases, actually don’t grasp this very well. That’s why even with that much practice, they can still struggle in certain cases and still feel like they need even more practice.

Well, practice is good, but only when you have mastered the basics. We talked a little bit about this in the Case Interview 101 video  but here is the much more in-depth explanation. To make this as easy to follow as possible, I divided the whole concept into bite-sized elements with numbers.

So here we go: the core logical foundation of how management consultants solve problems.

  • No.1: The problem has to be defined
  • No.2: To solve a big problem, we, management consultants, don’t look for solutions right away. Instead, we try to find the ROOT CAUSE. This ensures us that we have completely eradicated the problem and have a long-lasting impact.
  • No.3: There can be millions of possible root causes. To effectively and efficiently find the right one, we use a top-down and MECE approach. This is called an “issue tree”. See our MECE and Framework video for more details.
  • No.4: In order for each branch to exist in the issue tree, there HAS to be a chance that the Root-cause is in it. In other words, for every branch, there must be at least one hypothesis associated with it.
  • No.5: Now assume we have a structured, MECE, and hypothesis-based issue tree. Pick the best hypothesis,or,  in other words, choose a big branch to begin with. Then test if the root-cause is in there.
  • No.6: Depending on each case, different methods are used to test the hypotheses. But benchmarks are a powerful tool, too. Two main types of benchmarks used are the historical and competitors’ data.
  • No.7: If data suggests that the root cause is indeed IN the testing branch, go down one level deeper and repeat the same process: breaking down the branches into sub branches in an MECE, hypothesis-driven way and test each sub-branch using data. See case interview examples and answers
  • No.8: At any point where the data suggests that the root-cause is NOT in the testing branch or sub-branch, move to a parallel branch or sub-branch on the same level.
  • No.9:  Keep doing this until the whole issue tree has been covered or until the interviewer would like you to switch gears.
  • No.10: Lastly, once you have identified one or many root-causes, think of solutions to resolve them!

In theory, the above approach always works. But the following two conditions must be met.

  • Condition No.1: each and every single part of the issue tree must be perfectly MECE.
  • Condition No.2: the issue tree, or in other words, the breakdown must somewhat properly isolate the root-cause.

A few tips to meet these conditions:

Condition No.1: each and every single part of the issue tree must be perfectly MECE. Let’s look at an example of a movie theater. Suppose that after a movie show, there’s an anonymous feedback that the sounding was bad from his or her spot. As the theater’s consultant, you divide all SEATS in the theater into three areas: left, middle, and right. No matter how much investigation and testing you do, you still can’t find the root-cause.

It turns out that the anonymous customer was in the disability area. The original breakdown was not MECE.. or more specifically: not exhaustive.

Sometimes it can be hard to be perfectly MECE for the whole process.

Condition No.2: the issue tree, or in other words, the breakdown must somewhat properly isolate the root-cause. For example, in the theater example above, if unfortunately seats with audio problems were in fact the seats in the back, then breaking down the theater into left, middle, and right areas will give any candidate a hard time tracking down the problematic seats. Testing each branch of the issue tree would only yield weird results. You will find the partial problems in each branch and cannot isolate the problem unless you adjust the issue tree or use another framework.

Well, here are a few tips to meet those conditions:

1. Make sure you understand the concept of MECE really well. Please refer to our MECE video for more details.
2. Try to improve your business intuition in order to be able to pick good frameworks or correctly draw spot-on issue trees. We devoted a whole eBook in our End-to-end program for this.
3. Most importantly, develop the habit of aligning yourself with the interviewer. No matter how good you are in relation to the above tips, there will always be cases that are hard to be MECE inside out and hard to draw frameworks that are spot on. The interviewer is actually a great resource you can use.

If you would like to see more great materials, check out our Youtube Channel or subscribe to our emailing list.


See our Case Interview End-to-End Secrets Program for exclusive practice cases and insightful tips!

Save valuable time on the Case Interview by boosting up your math speed and accuracy!

Easy killer tips to improve your Case Interview right away!

Keep in touch with us

Get insights and updates from us on McKinsey PST and other Consulting


  • Max Wen

    I love the fact that you come up with the logical approach of solving problems. I personally think we should integrate this approach into our daily lives, not just limited to consulting work. I would appreciate if you could have further instructions on how to test the hypothesis by means of benchmarking or any other tools. Thank you so much, Kim!

  • Izzy

    Very inspiring!