The big difference between the McKinsey First and Second Round is in the second round you'll be expected to lead the process. It will be a candidate-led interview. Rather than testing the component level skills as in Round 1, they'll be testing how you pull all the skills together cohesively.
In terms of preparation, the general feedback from those getting offers is the Look Over My Shoulder® examples are extremely reflective of the dynamic of cases in McKinsey Rounds 2 & 3 (same feedback for BCG and Bain).
Listening actively to Look Over My Shoulder® and doing some practice cases with another person would be the recommended practice. Specific skills to work on that are relevant in Rounds 2 and 3, but not always in Round 1:
- Use of a framework or creating a custom issue tree
- Switching frameworks when you've exhausted the usefulness of a framework or issue tree
- Taking breaks from quantitative analysis to do qualitative analysis when the quantitative analysis is able to tell you what happened, but can't tell you why it happened or how it happened... and you're unable to progress in the case until you get the answers to "why" and "how"
- Asking for data -- generally very little data will be provided to you up front, you are expected to ask for what you want (and what you ask for should ideally be in support of a specific hypothesis, especially after the first 5 or 10 minutes of a case)
All of the key skills above are demonstrated multiple times in Look Over My Shoulder® at varying levels of candidate performance (so you can see the difference).
The one other point I will mention with Look Over My Shoulder®, is that it is a very good demonstration of the key principles behind good case interview performance.
A principle is a rule of thumb -- under situation X, you will want to consider doing Y.
But it is rare that the actual case you will get will be exactly the same as the cases you'll hear in Look Over My Shoulder®. They are always different in some way.
The key in going through Look Over My Shoulder® is to not only hear what people do when they perform well, but listen to my explanation as to why that was the right thing to do at that particular time.
It is very important to understand the reasoning behind each micro-decision made every four minutes in a case because you will not get the exact same case in a real interview. However, the reasoning behind the principles will apply and transfer to the case you will get in your actual interview.
If you understand the underlying rationale of each little case habit and behavior, when you get a case involving a situation similar (yet different) to what you're familiar with, you'll be able to adapt what you knowto that specific situation.
To put it in math terms, in Look Over My Shoulder®, you will see an example of "When you see X, you need to do Y." In your real case, you might see a situation that is described as X + Z, a set of circumstances you've never seen before.
But, if you have gone through Look Over My Shoulder® enough times, you will have internalized the principle that in X situation, you do Y in the case. This will lead you to realize that the case is referring to a situation X + Z, so you'll hopefully notice both the similarity to what you know and the difference, and do Y + Z.
In particular, I would encourage you to especially pay attention to the cases with issue tree diagrams (cases #2, #4, and #5). The frameworks in my frameworks handouts covers 65% of the cases you will see in an interview.
The remaining 35% of cases are either a variation of those frameworks (e.g., one that requires both frameworks, one that requires a custom issue tree within one branch of a framework, one that starts with a custom issue tree and then you discover it does actually fit one of my regular frameworks) or a custom issue tree from start to finish.
As you look at the issue tree diagram, you'll want to pay attention at two different levels. What I did in that particular case and notice the underlying principle of the issue tree process.
If you can master frameworks + issue tree process, you cover 95% - 99% of all case scenarios from the Top 3 firms. The issue tree process in particular is the one you really want to grasp as you can use an issue tree to structure just about any kind of problem -- even the totally weird ones -- so you can "ace the case".