YT YouTube2Text

YouTube2Text → Transcripts → Case Interview Fundamentals: Lesson 10 [Brainstormings: The 3-step approach]

Case Interview Fundamentals: Lesson 10 [Brainstormings: The 3-step approach] — Transcript

by CraftingCases · 7,343 words · 1,137 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00Hey there, Bruno here. In this video,
0:02you're going to learn a step-by-step
0:03method to solve one of the most
0:05important building blocks in case
0:07interviews, brainstorming. You might
0:10have heard about other names for the
0:12same type of question, such as the
0:14creativity question or the idea
0:16generation question. But if I were to
0:19give another name to brainstorming, the
0:21name I'd give would be the hypothesis
0:24generation question, because
0:26brainstorming is all about generating
0:28hypothesis in a structured way.
0:31Besides being useful for generating
0:33ideas and hypothesis, brainstorming is
0:36also the first structuring building
0:38block that you're going to learn. So, if
0:39learning to structure cases well is one
0:42of your main goals, you can't skip this
0:44module. So, let me start with a bold
0:47message. It's impossible to get an offer
0:50from a top consulting firm without
0:52knowing how to brainstorm well. And from
0:55what I've seen over the years, the vast
0:57majority of candidates who don't get an
0:59offer because they can't structure well
1:02enough suffer the most in the
1:04brainstorming questions, because there
1:06are no frameworks in case interview
1:08books for brainstorming questions.
1:11Anyway, let me show you what is a
1:13brainstorming question, how it looks
1:15like, before I show you how to solve
1:17those in a way that makes your
1:19interviewer delighted.
1:23If you remember from the video on the
1:24six building blocks, brainstorming is
1:27the first building block related to
1:30structuring.
1:31Let's recall some examples of questions
1:33that would be brainstorming questions.
1:36An iron ore mining company has seen its
1:39revenue drop 30% in the last year,
1:42despite heavy investments. What are
1:45plausible hypothesis that could explain
1:47that?
1:48A large university wants more applicants
1:51for their undergrad program next year.
1:54What are some ideas of what they could
1:56do to get that?
1:58Apple wants to launch a cheaper version
2:01of the iPhone. What are the risks?
2:04What's common between these questions?
2:07Well, they all ask for a list of
2:11hypothesis or ideas.
2:13In most case interviews, you will need
2:15to at some point generate plausible
2:18hypothesis about why something's
2:20happening.
2:21Or the interviewer might ask you to
2:23generate ideas on how to fix a problem
2:25or improve a company, such as in the
2:28example of the university trying to get
2:30more applicants next year.
2:32Or maybe he wants a list of the risks.
2:34It doesn't matter. In fact, a list of
2:37ideas, a list of risks, they're all
2:40lists of hypothesis. A list of ideas is
2:43just a list of hypothesis of solutions
2:46to the problem. A list of risks is just
2:49a list of hypothetical situations that
2:51could be harmful. So, all of these
2:54questions fall under what we call
2:56brainstorming. You use the technique
2:58we'll teach in this video whenever you
3:00need to generate a list. Of course, a
3:03simple list is not what the consulting
3:05firms are looking for. Otherwise, I
3:08wouldn't need to make a video about it,
3:09right?
3:10What you need is a structured,
3:13prioritized list of good ideas or
3:17hypothesis. Let me explain why.
3:20As a consultant, generating ideas or
3:23hypothesis is not enough. You will need
3:26to test them. By having them in a
3:28structured format, you're often able to
3:31test them as bundles of ideas instead of
3:34individual ideas.
3:36And by having them prioritized, you can
3:38start testing the one that makes more
3:41sense first, and perhaps save some time
3:44and be more efficient in your work,
3:45instead of randomly testing them in a
3:48random order. But most people don't do
3:50that. Too many candidates, when faced
3:53with a brainstorming question, will
3:55generate a random list without structure
3:58or priority and fail their interviews,
4:01receiving the feedback that they weren't
4:03structured enough. Here's what they'll
4:05do. Suppose they get the iron ore mining
4:08question. An iron ore mining company has
4:11seen its revenue drop 30% in the last
4:14year, despite heavy investments. What
4:17are plausible hypothesis that could
4:19explain that? A normal candidate who
4:22does not brainstorm well would answer it
4:24saying something like this. Well, it
4:27could be that demand has dropped. Maybe
4:30there's a recession going on, or they
4:33used to sell to steel mills, and steel
4:35mills aren't buying much because they've
4:37accumulated inventory, or maybe their
4:40sales force is not doing a good job.
4:43Um, I don't know. Maybe prices have
4:45dropped.
4:47See the problem? It's not on the ideas.
4:50The ideas are fine. The problem is the
4:53lack of structure and logic. Maybe one
4:56of these ideas is actually what's going
4:58on. They're actually good ideas, but
5:01maybe
5:02these ideas do not contain the real, uh,
5:06answer, or they don't contain the
5:09hypothesis that would actually be
5:10confirmed. And who's going to bear the
5:13risk of having the project delayed, or
5:16even worse, never getting to the root
5:18cause of the problem? Certainly not a
5:20McKinsey partner, which means that if
5:23you ever answer a brainstorming question
5:25like this, you will probably not get the
5:29offer.
5:30Here's what a great candidate would say
5:32instead. Well, if revenue has dropped,
5:35it's either due to a drop in volume of
5:38iron ore sold or the average price. If
5:41it was a lack of volume, I see three
5:43main reasons.
5:45We're not producing enough, the market
5:47is buying less, so the demand is down,
5:50or the market has decided to buy more
5:52from our competitors and less from us.
5:56If the problem is price, I can see three
5:58main reasons as well. Maybe the average
6:00market price is down, or the mix that we
6:04sell has shifted from high value
6:06products to low value products, or we're
6:10not getting market price in our products
6:13because of something that has happened
6:14to our product. Now, because this market
6:18is cyclical, I'd imagine that a revenue
6:20drop of this magnitude
6:23in this short amount of time was caused
6:26by both a decrease in market volumes and
6:29market prices due to a an economic
6:34recession. Do we have data that confirms
6:36this or any of the other hypothesis that
6:39I came up with, or should I think about
6:41more specific hypothesis within each
6:44bucket? Now, if I'm a partner in a top
6:47consulting firm and I'm interviewing
6:49someone and they give me an answer like
6:51this, I'll probably start a slow clap,
6:54get up, give him a hug. I'm just
6:57kidding. But you can can you see the
6:59difference of this answer compared to
7:01the other one? There's not even that
7:03many ideas here, but I know that this
7:06guy would find the right reason in a
7:09real project, and I know that he could
7:12come up with ideas within each bucket
7:14when prompted to do so. And if I have
7:17any doubts, I can just ask them, "Well,
7:19what are your ideas regarding why the
7:21market share would be down?" or
7:22something like that.
7:24Brainstorming questions are extremely
7:27common. I would say that most cases have
7:30between two and four brainstorming
7:33questions per case. This happens because
7:36the skill to generate hypothesis and
7:39ideas in an orderly way is something
7:42that you do almost every day as a
7:45consultant. So, let's dive into the
7:47technique itself. I will show you a
7:50simple approach to brainstorming, the
7:52three-step approach. If you learn this
7:55well and practice until you internalize,
7:57until you learn to generate hypothesis
8:00in a structured, prioritized way every
8:03time, you'll be vastly better than all
8:06the other candidates who are just
8:08creating random lists of unconnected
8:11hypothesis to their interviewers.
8:15Now, before I tell you what the three
8:17steps to brainstorming are, I have a
8:19question for you. What is our goal when
8:22brainstorming ideas or hypothesis?
8:25Well, it is to generate a structured,
8:28prioritized list of good ideas and
8:32hypothesis, right? So, there you have
8:34it. Those are the three steps. First,
8:37define and structure the question.
8:40Second, prioritize the drivers. And
8:44third, generate hypothesis, or if it's a
8:47brainstorming of ideas, generate ideas.
8:50And to explain each step well, I'm going
8:52to use a case as an example. Here's the
8:55case question. Mad, a local martial arts
8:59dojo, has recently seen its revenues
9:02decline sharply. What could have
9:04happened? This is a classical
9:07brainstorming question that could that
9:09could be in the middle of a case, but
9:11most likely would be in the beginning of
9:13the case. Note that this is not a
9:16revenue decline case yet, in the sense
9:18that you have to fix the revenue
9:20decline. So, you have some cases where,
9:23uh,
9:24the revenue has declined and the
9:26interviewer asked you to fix it. This is
9:28not the case. So far, the interviewer
9:31just asked you what might have happened,
9:34which is code for, "Can you come up with
9:36a structured list of hypothesis that
9:39could have caused it?" This case might
9:42evolve into how to solve the problem and
9:44fix the the revenue decline later on,
9:47but for now, it's just a brainstorming
9:49question. Your interviewer just wants to
9:51hear your hypothesis. So, let's do what
9:55he asks, following our three-step
9:57approach. We'll start with defining and
9:59structuring the problem. This is when we
10:02show the interviewer that we're
10:04structured thinkers.
10:07Okay? So, there are three principles to
10:09follow here. Principle number one is to
10:12define the problem specifically.
10:15Principle number two is to structure the
10:18problem with MECE issue trees.
10:21And principle number three is to make
10:23your structure problem-specific.
10:28So, before we even create our structure,
10:30we need to define the problem. It's
10:32impossible to structure the problem well
10:35before we know what the problem really
10:37is. In this case, it's pretty clear. We
10:40need to figure out what has caused a
10:42decrease in revenues.
10:45What makes this case so clear is that we
10:47have a metric to work with, revenues.
10:50Whenever you can, try to reduce the
10:52problem to a metric. So, just to give an
10:54example,
10:55if your interviewer was asking you to
10:57give them ideas on how to make a product
11:00better, better isn't very helpful. Can
11:03you make this a metric? Maybe it's a B2B
11:06industrial product, and a better product
11:08is the one that saves the client more
11:10time and money.
11:12Maybe it's a smartphone, and the better
11:14product is one that serves more
11:16functions well. I don't know if that's
11:18is a good definition of what makes a
11:20smartphone better, but I mean, a
11:22smartphone that has a good camera is
11:24better than one that doesn't. Maybe your
11:27product is a social media app, and the
11:29better product is the one that people
11:31use for more time during their day.
11:34So, often times you won't be able to
11:36reduce the problem to a metric, and
11:38that's okay. Not everything that counts
11:41can be counted, but if it can, you'll be
11:43much better off by making our problem a
11:46metric. In case you can't, try to make
11:49it more specific anyway, using words.
11:53So, what does better mean? Is it harder,
11:55faster, stronger, more user-friendly? I
11:59don't know, but try to make the words
12:01very specific when you can't use a
12:03metric.
12:04Anyway, in this case we're lucky because
12:07we have a very well-defined metric,
12:09revenues. But we can get even more
12:12specific. We can ask how much has
12:14revenue dropped and over how much time.
12:17We can also clarify what are the revenue
12:19streams of this dojo. I'd ask something
12:22like, "I imagine a martial arts business
12:25makes money from monthly subscriptions
12:28to the classes and by selling gear. Am I
12:31right, or are there of other revenue
12:33streams?" Let's say that the interviewer
12:35tells us that yes, that's how they make
12:37money, and that revenues have dropped by
12:4030%
12:41in the past 6 months.
12:44Now, you might be asking yourself, "Hey
12:46Bruno, am I not wasting time doing all
12:49these definition things?"
12:51Heck no. Defining well takes less than a
12:54minute, and is in itself an important
12:57skill that your interviewer will be
12:58evaluating. That's how you make projects
13:01work in consulting, by defining them
13:03really well. It also prevents errors
13:06when the problem is less obvious, so it
13:09prevents you from going down a 30-minute
13:12interview solving the wrong problem, and
13:15it will also help us a lot in the
13:17prioritization step, as you're going to
13:20see. Anyway, now that we know the
13:22problem specifically, we need to
13:25structure the problem. We structure the
13:27problem using MECE trees that are
13:30problem-specific, so principles number
13:32two and three of the define and
13:35structure step.
13:37Let me show you how I do it.
13:39So, in the first level, I'd break down
13:41revenues into number of customers or
13:44revenue per customer. At least one of
13:47those must have dropped, and the
13:49combination of them must have dropped by
13:5230%. So, maybe one was increased, but
13:56the other dropped by more than 30% or
13:59maybe both of them have dropped by
14:01roughly 15% each.
14:04Then, I'd state that if the number of
14:06customers is down, that's either because
14:09more customers are leaving us
14:12or because we've slowed down the number
14:14of new customers that we get each month
14:18to make up for the normal rates that
14:20customers leave. In business, you'd call
14:23these a customer retention problem and a
14:26customer acquisition problem.
14:28You don't need to know these words, you
14:30can just use normal words. Customers are
14:32leaving more often, or we're getting
14:34less new customers. Quick note here to
14:37be clear to people from non-business
14:39backgrounds. It's normal that some
14:42people stop being customers. So, every
14:45business that works on a monthly
14:47subscription basis has a normal rate for
14:50customers leaving them. We call this
14:52customer churn rate.
14:54If, say, 10% of your customers leave
14:56every month,
14:58and that's the normal rate, then we need
15:00to get new customers at the same rate
15:03just to stay flat on revenues. A
15:05customer retention problems happens when
15:08that 10% rate increases and more
15:11customers leave. So, each industry has a
15:14normal rate, each business has a normal
15:16rate. For some of them it's 10%, some of
15:18them is 1%, others it's 50%.
15:22Now, a customer acquisition problem
15:24happens when the rate through which our
15:27customers leave didn't increase, but we
15:30can't find new customers to make up for
15:32the ones we've lost. Got it? Okay. So,
15:36on the average revenue per customer
15:38side, this could have happened because
15:40the average monthly fee for the martial
15:43arts service for the classes has
15:45dropped, or because the average customer
15:48is spending less on gear every month.
15:52This is my structure. So, step number
15:54one, define and structure, ends here.
15:58Before we move on to step number two,
16:00let me quickly go over the two main
16:03principles to have in mind when
16:05structuring.
16:06The first one is that you need your
16:08issue tree, your structure, to be MECE.
16:11MECE means mutually exclusive and
16:14collectively exhaustive. In other words,
16:17your structure must have no gaps and no
16:20overlaps. Do you see how the four
16:22buckets at the end of the tree are
16:24different from each other and show no
16:27overlap with one another? So, customers
16:29leaving is different from getting new
16:32customers. Average monthly fees is
16:34different from uh dollar spent per gear.
16:38Do you know how the four things together
16:40cover the whole revenue problem? There
16:43can't be any revenue problem that's not
16:46in one of those buckets. This means the
16:48structure is MECE.
16:50In this case, it was easy to do that
16:52because revenues is a numerical
16:54variable, and I just broke it down into
16:57a formula.
16:58One more reason to try to make a problem
17:00a metric, a numerical variable. But it's
17:03not always that easy. So, if you're new
17:05to MECE, or if you have trouble ensuring
17:08your structures are MECE, check out the
17:11lesson in this module called the five
17:13ways to be MECE. It will take you to an
17:15article series or a YouTube playlist,
17:18which ones you prefer, that cover the
17:20topic well and show you the five
17:22techniques to ensure MECE-ness when
17:24building your issue trees.
17:27Now, the other feature of this structure
17:29is that it's specific to this problem.
17:32The usual way to break down revenues is
17:35price times quantity. You probably have
17:37seen that formula before, but I didn't
17:40use it here. Why? Because it doesn't
17:42make much sense. Price of what? Quantity
17:45of what? This business has two uh
17:49revenue streams, so the price doesn't
17:51make much sense, and this business is a
17:54service, so you you don't sell services
17:57by quantities, right? It makes much more
17:59sense to break it down into the number
18:02of customers and the average revenue per
18:05customer. And then talking about things
18:08like customer acquisition, customer
18:10retention, and so on. Always make your
18:13structure specific to the problem or to
18:17the client's industry. It does take some
18:19practice, but it ensures you show your
18:22business sense to the interviewer, and
18:24it shows that you're not just
18:26regurgitating some framework that you
18:28read somewhere.
18:30Okay, now that we have a clear structure
18:32to generate our hypothesis on the
18:34problem, let's move on to step number
18:36two, to prioritize drivers.
18:39Prioritization is important because it
18:41shows you can be 80/20. Everyone tells
18:44you you should be 80/20, no one shows
18:46how. This is how you prioritize your
18:49structure. And it's a pretty simple
18:51step. Of the four drivers at the end of
18:54the tree, which one or which ones should
18:57we focus on? Which one or which ones are
19:01probably more important?
19:03There are three main principles here.
19:05First, take the context and the numbers
19:08of the case into consideration. These
19:11are clues of
19:13what should you prioritize in that
19:14specific situation.
19:16Number two, think of which ones of which
19:20drivers could have a bigger impact. And
19:23number three,
19:24think of which drivers have a higher
19:26likelihood this question
19:31were a question on ideas to improve your
19:34revenue, instead of hypothesis of what
19:37has caused them to drop, then instead of
19:40thinking likelihood, we should think
19:42about the feasibility of each driver to
19:45improve. So, maybe it's easier to get
19:48more customers than to to prices.
19:52Now, think for yourself for a second.
19:54Which of these four drivers would you
19:57prioritize to later generate specific
20:00hypothesis of what could have happened?
20:03You can still generate hypothesis about
20:05all four drivers, but which one would
20:07you put most of your focus on?
20:10Here's how I would think about this. And
20:12this is not how what I would speak to
20:14the interviewer, but instead it's what's
20:17happening in my head.
20:19The context that I need to take into
20:20consideration, so the first principle
20:22are two things. This business is in the
20:25martial arts teaching business, and
20:28second, revenues have dropped by almost
20:30a third and pretty quickly. So, if there
20:34was one cause, it must be something that
20:37can take that much revenue away that
20:39quickly. So, this is the context and the
20:41numbers that I take into consideration.
20:44Now, in terms of impact, all four
20:47drivers can have a similar impact except
20:50for sales of gear, which I believe are a
20:53smaller value than the monthly
20:55memberships. So, say that people spend
20:58$500 per month uh to to take these
21:02classes, and they just spend maybe 50 or
21:05$100
21:07uh in gear per month. So,
21:08percentage-wise,
21:10the monthly fees have a much larger
21:13impact than the sales of gear. You
21:15probably can't even
21:18drop your revenues by 30% even if you
21:20cut all sales of gear. That's an
21:23assumption, of course. Finally, in terms
21:25of likelihood, I'd say monthly
21:27membership value probably didn't drop
21:3030% unless people switched to a lower
21:33value plan that we've started. And I'd
21:36say that it's unlikely that customer
21:39acquisition has dropped to a level that
21:41brought a 30% revenue decrease because
21:45even even if we're doing poorly in the
21:48marketing department, we still have our
21:50current customers. So, I'd say that
21:54it's unlikely that we dropped prices
21:57that much. It's unlikely that we're
21:59doing such a bad job in getting new
22:02customers that the overall revenue would
22:04have dropped by 30%, but I'd say that
22:07it's pretty likely that a bunch of our
22:09customers have dropped out for some
22:11reason.
22:12Now, here's how I'd present this
22:15reasoning to my interviewer.
22:17That all was the stuff that was going on
22:19in my head as I was going through the
22:21three principles to prioritize, but to
22:24the interviewer, I'd present it this
22:26way.
22:27So, I'd love to have data on what
22:29happened to each of these four
22:31variables, but if we have no data, my
22:34hypothesis is that the one that's most
22:36likely to have caused the revenue drop
22:39is that our customers started to drop
22:41out more often. And I say this because
22:44even if we're not acquiring customers as
22:47fast, we'd still have our current
22:49customers, so revenue wouldn't have
22:51suffered so much in the short term. And
22:54I also say this because it's unlikely
22:56that we have dropped our prices that
22:59much. Gear sales shouldn't be a huge
23:01part of revenue, so that shouldn't be
23:03it.
23:04If customers are indeed leaving us more
23:06often,
23:08that could be for one of three reasons.
23:10First, they're leaving us for our
23:12competition. Second, they've moved on to
23:16other types of paid physical activities.
23:19And third, they've stopped exercising or
23:22at least they've stopped paying for
23:24exercise.
23:26Now, that's how I'd present it to the
23:27interviewer, and let me be honest with
23:30you. Prioritization is by far the
23:33trickiest step. I've seen a ton of
23:35people communicate this step poorly and
23:38suffer the consequences. So,
23:41if you if you're not that good at
23:43communication, you can skip this step
23:45altogether if you want and go straight
23:47from the structure to the ideas. So, you
23:50can just do steps one and three. But if
23:53you can communicate well, it's very much
23:56worth it because it shows more logic and
23:59it shows that you focus on what's
24:01important. And so, here are three things
24:04you should be careful about this step.
24:06First, make sure you ask for data before
24:09you prioritize. If they don't give you
24:11any data, say that you wish you had
24:14data, but here are the parts that you
24:16think are most important. If you skip
24:19this asking for data part, some
24:21interviewers might think that you're
24:23assuming too much and let you go towards
24:26a wrong path and fail the interview. So,
24:29you should always ask for data before
24:31you do this.
24:32Second, make sure you give the rationale
24:35of why you're prioritizing that specific
24:38bucket. This shows how you think, and
24:41because of that, it's the most important
24:43part. And third, never imply that the
24:46other buckets, the other drivers can't
24:49be the cause. On the contrary, show that
24:52you're aware that the cause may indeed
24:54be in the other buckets, but you would
24:56investigate this one first. Now,
24:59prioritization helps a lot because it
25:02lets you dig deeper into the most
25:04important bucket and build a
25:06substructure for it as I did here. And
25:08it also lets you spend more time on
25:11generating hypothesis for that
25:13prioritized bucket.
25:16And finally, the last step is to
25:17actually generate your specific
25:20hypothesis or ideas, in this case
25:23hypothesis. And this is where most
25:25candidates start. Do you notice how much
25:28important work they have skipped?
25:30This is where you show your creativity
25:33and your business sense
25:35as well. So, let's dig into that. There
25:37are three main principles to bear in
25:39mind to generate ideas or hypothesis
25:42well. First, don't forget to say the
25:45obvious. Obvious ideas are important
25:48because if you don't mention them, you
25:50will leave the interviewer wondering if
25:52you've thought of them at all. This is
25:55even more important when the question's
25:56about plausible hypothesis such as the
26:00question example we're doing here. And
26:02the reason is because obvious things are
26:05often the truth. Keep in mind that most
26:08questions clients have to consultants
26:10have obvious answers and mundane
26:13solutions. Just like most cases that
26:16medical doctors have to deal with are
26:19not outlandish.
26:21Second, think outside the box. If you
26:24just say the obvious things, you won't
26:26show your creativity, which is something
26:29that they're looking for as well. Hey
26:31Bruno, what if I'm not creative? Well,
26:34read more and make more analogies. Maybe
26:37you can't come up with out of the box
26:40hypothesis for why a martial arts
26:42business has lost clients. You don't
26:45know nothing about a martial arts
26:47business, but you might have read about
26:49a weight training business that's been
26:51losing clients. Or you might have read
26:53an article about, I don't know,
26:55millennials dropping out of college. You
26:58can make analogies and come up with
27:00similar hypothesis to things you have
27:03read in the same industry and in other
27:06industries. And the third the third
27:09principle is to stay practical whenever
27:11you give an idea. Creative ideas get no
27:14points if there's no way they can be
27:17implemented. Same for creative
27:19hypothesis that are impossible to
27:22happen. So, when thinking about ideas,
27:25always think about potential impact of
27:28the idea and the actionability or the
27:31feasibility of implementing the idea.
27:34When thinking of hypothesis, think about
27:37the potential impact of the hypothesis
27:40versus the likelihood. Ideally, you want
27:43both impact and likelihood to be high,
27:45but if an hypothesis has high impact and
27:49low likelihood, that's also a good
27:51hypothesis. If it has a high likelihood
27:55but a very low impact, probably not a
27:59good idea or a good hypothesis, at least
28:01in this case because revenues have
28:02dropped by 30% in 6 months. So, given
28:06these three principles, here are some of
28:08the hypothesis that I generate for our
28:12Mad Martial Arts Dojo case.
28:15And just as a note here, to save some
28:17time, I'll just generate hypothesis for
28:20the priority bucket, customer retention
28:23problems. In a real case, I'd start with
28:26the priority bucket and I'd generate
28:28more hypothesis to the priority bucket,
28:31but I'd also generate a few hypothesis
28:34for each of the other buckets as well
28:36just to show the interviewer
28:38that I can generate as many hypothesis
28:41as I need for any kind of topic. So,
28:44starting with clients leaving to
28:46competition, I have a few hypothesis for
28:49that. One is that there's a new
28:51competitor who just opened up shop and
28:54who's better than us in some metric, and
28:57clients are switching to that one. Maybe
28:59they have a better brand name, or their
29:02facilities are nicer, or they have
29:04better teachers, or they're cheaper. Or
29:08maybe we've changed something in our
29:10business that our clients didn't like.
29:13Maybe we've hired new instructors who
29:15aren't as good, or we've switched class
29:18hours and they don't fit our customers'
29:21schedules anymore, or we've raised the
29:23prices and customers don't want to pay
29:26the new price. Or maybe we've changed
29:28locations and people don't want to drive
29:31that far. Or maybe our customers have
29:34switched to other types of physical
29:37exercise, bucket number two. Maybe we
29:39were in a fad market and there was a
29:41huge influx of customers recently, but
29:45they all left to the new fad. Maybe
29:47CrossFit or something like that. Or
29:50another hypothesis within switching to
29:52other physical exercise is that we had a
29:55lot of children among our customers and
29:58then something happened that caused
30:00parents to feel unsafe with martial
30:03arts. Maybe a kid died somewhere while
30:05training and they're all reacting to the
30:07news.
30:08Finally, people could be stopping to pay
30:11for exercise. Maybe there's a huge
30:13economic crisis going on and the parents
30:16are just leaving their kids to exercise
30:19in school or if our clients are adults,
30:22they're moving into cheaper or free
30:24alternatives such as running or weight
30:27training. So, this is it. In a real
30:30case, what I do now is to move to the
30:33other buckets in order of priority and
30:36generate a few hypotheses for each one.
30:39Why we might be getting less new
30:41customers and here the impact should be
30:43really high or why we might have a drop
30:46in average monthly fees.
30:49One thing that I want you to notice,
30:51there's just five bullets in this
30:54screen. I mean, bullet points. But
30:56within each of them, I gave a bunch of
30:59specific hypotheses. If you missed that,
31:02go back a couple minutes in the video
31:04and watch it again.
31:06Now, let me debrief some of the
31:08principles in generating the hypothesis
31:10here. What's the obvious hypothesis for
31:13losing customers?
31:15Wait for it. We've raised prices, of
31:18course, and I've mentioned that under
31:21the more general idea of we've changed
31:23something in our business. I've
31:25mentioned prices there. But I've also
31:27mentioned other obvious changes such as
31:30schedules and teachers and locations.
31:34Another obvious thing that I've
31:35mentioned, well, the economic crisis and
31:38also the new competitor. So, I said a
31:40bunch of obvious ideas. Some are obvious
31:43than others, but all of them pretty
31:46obvious. But I've also mentioned some
31:48hypothesis that you could call out of
31:50the box such as us being in a fad market
31:54and parents scared that some kids died
31:57somewhere. This shows that I can think
31:59creatively and potentially find causes
32:02for things even when they're not
32:04obvious. This is a really important
32:06consulting skill. Now, the key to this
32:09is that all these hypotheses, even the
32:11outside of the box ones, are always
32:14practical. Fad markets exist, especially
32:17in the fitness world, and kids die doing
32:20sports sometimes, and when that happens,
32:23some parents overreact. Now, this last
32:26hypothesis of
32:28uh a kid dying somewhere and parents
32:30being scared and and
32:32uh
32:32taking their kids out is very unlikely.
32:36And I need to make the interviewer aware
32:38of that, that I know it's unlikely,
32:41especially if he challenges me. But it's
32:44still possible. And if something like
32:47that were to happen, the impact on the
32:49business would be very large. It's just
32:52like when an airplane crashes and then
32:54people start stop flying. Same thing if
32:57a kid dies
32:58uh training, I don't know, karate or
33:01jujitsu, probably these types of
33:04business in the
33:05uh nearby locations will suffer a lot.
33:10Now, if you follow these three steps,
33:12you're going to do great in the
33:13brainstorming question. You're going to
33:15have a structured, prioritized list full
33:19of insightful hypotheses that can
33:21explain what's going on or ideas of
33:23solutions to your client's problems. But
33:26great candidates are always thinking a
33:29step ahead. What comes next? So, you
33:32have to have great care on how you
33:34finish [snorts] your answer.
33:36In a candidate-led case, this is
33:38critical because it helps you guide the
33:40case effectively. But even in a
33:43McKinsey-style interview led case, this
33:46is important because it shows your
33:48interviewer that you would be able to
33:50guide the case if you had to or drive
33:53the project forward in a real-world
33:55situation. And the key to finish your
33:58brainstorming is simple. Say what data
34:01you would look for to come to the
34:03conclusion. When you brainstorm, you're
34:06generating hypotheses. So, the obvious
34:09thing to do is to say how would you
34:11transform this hypothesis into a
34:14conclusion.
34:16And there are two ways to do that.
34:19First, with a combination of an analysis
34:23and a hypothesis testing building block
34:26to reach the conclusion of where the
34:28problem is. Here, you mention which
34:31pieces of data you need to know what the
34:33problematic part of your structure is.
34:36So, where does the problem come from? In
34:38the case of the Mad Martial Arts Dojo,
34:41where does the revenue problem come
34:43from? And then the hypothesis testing to
34:45confirm that.
34:47Now, the second way is that often times,
34:50as you brainstorm and you mention one
34:52specific hypothesis, your interviewer
34:55will confirm it to you on the go. So,
34:58say that as you say that you might be
35:00losing customers
35:02because they're going to CrossFit and
35:04you were in a fad market, they might
35:06say, "Hey, that's exactly what's going
35:08on. What would you do next?" Here,
35:12they're sparing you of doing the
35:14analysis and the hypothesis testing as
35:16in the first path. What you should do
35:19next then is obviously to find a
35:21solution to that problem as you already
35:24know the root cause of the problem
35:26because your interviewer just told you.
35:28So, the way to go here is to solve
35:30another brainstorm, this time with the
35:33solution to the specific problem you
35:35found. Now, let me show you how I would
35:38deal with each of these scenarios. Let's
35:40say you've just finished the Mad Martial
35:43Arts Dojo brainstorming and your
35:45interviewer gave you no clue about which
35:48hypothesis is true. Here's what I would
35:51say.
35:52And now that we've have some hypothesis
35:54of why revenues are dropping, we need to
35:57figure out what's actually going on.
36:00What I would do now is to collect data
36:03about uh new customers, customer
36:06acquisition, monthly fees, and revenue
36:10generated from gear sales. Once we
36:13figure out which one of these uh drivers
36:16is causing the revenue to drop, I'd go
36:19one or two levels deeper to find out why
36:24it has dropped. So, for example, if the
36:26problem is that we're losing customers,
36:28I'd try to assess if they're going to
36:30competitors or something else. Let's say
36:33that they're going to competitors. I'd
36:36now call some of these customers who
36:38left us and ask them why they left. If I
36:41want to be more rigorous, we might want
36:43to do a survey with them, but I'd still
36:45call some of them just to make sure that
36:47leads some of them answer the question
36:50as a lot of people just ignore surveys.
36:53Saying this in this way shows the
36:56interviewer that you're actually focused
36:58on solving the problem and that you can
37:01solve problems on your own, not just
37:03answer their questions and then not know
37:06what to do.
37:07Or let's check path number two. Suppose
37:10you're doing the Mad brainstorming and
37:12as you finish, the interviewer tells
37:14you, "Well, suppose they are indeed in a
37:16fad market and all their clients who
37:19left moved on to CrossFit." Here's what
37:22I would say. "Well, if that's the case,
37:25we need to first understand if more
37:26people are going to leave to CrossFit
37:29and we could do that by asking our
37:31current customers how happy they are
37:33with our martial arts services and if
37:36they're
37:37considering doing other physical
37:38activities. If they say that they're not
37:41considering moving on to CrossFit, then
37:43we need to find some ideas on how to get
37:45new customers. Now, if they are
37:48considering moving on to CrossFit, we
37:50need to see how we can both retain these
37:52customers as well as get new customers.
37:57The way I'd go about finding these ideas
37:59is to think of things we can do with the
38:02current martial arts business and things
38:05we can do by expanding our business into
38:07other types of services including
38:09CrossFit, for example. Within the
38:11current business, what we could do is
38:14and then I'd go on into the
38:16brainstorming. So, I just laid out a
38:18quick structure uh on on the how to get
38:22more customers and how to retain the
38:24current customers.
38:26Notice that I mentioned that path number
38:30two is to just brainstorm, but in
38:32practice, I actually first tested the
38:36hypothesis that clients will continue to
38:39leave us and only then I actually
38:41brainstorm. So, if you see the two
38:43bullet points here, the first one is to
38:45the hypothesis testing to define what
38:48the brainstorming is about. Is it about
38:50retaining customers? Is it about getting
38:52new customers or both?
38:54Why did I do that? Well, because the
38:57brainstorming wasn't well defined yet.
38:59All of this is to say
39:02that I'm giving you some guidelines
39:04here. This path number one, this path
39:06number two on screen, are guidelines of
39:09what you could do, but you still have to
39:11think. Each case is different, but the
39:14principles apply. With what you know by
39:16the end of the brainstorming, what would
39:18you do next to help your client the
39:21most? This is how you finish your
39:23brainstorming answer.
39:26Now, before we finish this video and
39:28move on to the drills where you practice
39:31your brainstorming skills, there are a
39:33few questions that appear over and over
39:35again whenever we teach this. So, I
39:38prefer to answer them before they come
39:40to you. Specifically, when should I use
39:43the brainstorming building block?
39:45How to handle what else questions from
39:47the interviewer? And
39:50the third one is kind of a bundle of
39:51many questions.
39:53What if the interviewer gives me no time
39:55to think? What if they tell me not to
39:57use paper? What if they tell me not to
40:00worry about structuring and just say
40:02what's on my mind? Should I still be
40:04structured anyway?
40:06Let's tackle each one individually.
40:08First, when to use it? I've never heard
40:11of interviewers asking candidates to
40:13directly brainstorm, such as why don't
40:15you brainstorm ideas for X? Or why don't
40:19you brainstorm some hypothesis? That
40:21doesn't happen in the interview and it
40:24doesn't happen in the real job either,
40:26even though people brainstorm all the
40:28time. So, you should use the
40:30brainstorming building block in two
40:31situations. First, when the interviewer
40:34asks you to do a list.
40:37Second, when they ask you open
40:39questions, usually phrased as why do you
40:42think that's happening? Or what do you
40:43think is going on?
40:45The first type of question, the list
40:47one, is quite obviously a brainstorming
40:49question. The interviewer is asking you
40:52to generate a list of ideas or
40:54hypothesis. And the common mistake
40:56people make is to actually generate a
40:58list without any structure. The second
41:01scenario is a bit trickier. It is quite
41:03common during an interview for the
41:05candidates to ask a question to the
41:07interviewer, usually asking for data.
41:09For example, uh do we know why customers
41:12have left us?
41:14And have the interviewer send their own
41:16question back to them in a what do you
41:19think format? Or why do you think they
41:21have left us?
41:23Now, what most people do in response to
41:25this type of answer that the interviewer
41:27sends their own question back to them is
41:30to say one potential cause that they
41:32believe in. Or some people might list a
41:34couple of reasons, but that's not what
41:37you should do. You should not give your
41:39opinion on the topic or what you think
41:42is going on because consultants
41:44shouldn't work with opinions. They
41:46should work with hypothesis and with
41:48data. So, what the interviewer is
41:51expecting when they ask you what do you
41:53think is a brainstorming. You have no
41:56data, so you have no idea why this is
41:59happening, which means that you should
42:01generate a list of hypothesis. A
42:03structured, prioritized list of
42:06hypothesis. Now, here's two situations
42:09where you should not use brainstorming.
42:12The first is when the interviewers are
42:14gladly answering your questions. Some
42:16candidates I've coached stopped asking
42:18contextualization questions, such as why
42:20questions, after I taught them to
42:23brainstorm. They'll start brainstorming
42:25without asking these questions first.
42:28But you should always ask first. It
42:30saves a lot of time and you shouldn't
42:33work with hypothesis when you can work
42:35with facts and data.
42:37The second circumstance not to
42:39brainstorm is when you already have the
42:42data or you can get access to the data
42:44by asking for the data. Again, don't
42:46work with hypothesis when you can have
42:49real data.
42:50Now, onto the second question. How to
42:52deal with what else questions?
42:55For those of you who have never been in
42:56a real interview in a real consulting
42:58firm before, you might not be aware of
43:00this, but it's common for interviewers
43:03to press you for volume of ideas.
43:06They'll ask you a brainstorming
43:07question, such as the math dojo one. And
43:10once you finish brainstorming, they'll
43:12ask you, well, what other hypothesis do
43:14you have to explain why your customers
43:16might believe in you? And then you
43:18generate some more hypothesis and
43:20they'll ask you, what else? And once you
43:23generate a few more and stop, they'll
43:25say, what else?
43:26Be prepared for these what else
43:29questions. A lot of interviewers will
43:32press you to have more and more ideas or
43:34hypothesis by asking this simple
43:37question several times. If you sense
43:40your interviewer is shooting for volume,
43:42give them a lot of ideas. They just want
43:45to test where your creative limit is.
43:48And you need to practice for this. When
43:51practicing for breadth of ideas, try to
43:54get 30 ideas for any brainstorming
43:56question. This will almost surely happen
43:59in your interviews, so better be
44:01prepared for it.
44:03Finally, question number three. These
44:06are the scenarios that can happen in
44:07your interviews and that leave the
44:09brainstorming question way more
44:11challenging. Although you can always ask
44:13for 30 seconds to 1 minute to structure
44:16your ideas and most interviewers will
44:18give you that, partners will often tell
44:21you that they won't give you time to
44:22think. Why? Because clients want to hear
44:25your ideas or your hypothesis and
44:28they'll think it's really weird that you
44:30get a piece of paper before you can
44:32speak. No one does that. Sometimes
44:34partners will go so far as to tell you
44:37not to use paper. Partly because that's
44:39a way of telling you that you get no
44:41time to think, but partly because in a
44:44real client meeting, it's weird to talk
44:47while writing, right? They just want to
44:49hear your ideas.
44:50Now, I've been through scenarios and
44:52I've seen candidates go through
44:53situations where partners will say
44:56things that will make you think you
44:57shouldn't even be structured. They'll
45:00tell you, hey, no need to think, just
45:02say what's on your mind. Or even, hey,
45:04no need to be structured, just tell me
45:06your ideas. Now, the most important part
45:09of this video. Don't fall into this
45:12trap. I don't know if they do this trap
45:15on purpose or not, but they will reject
45:17you if you give them an unstructured
45:20answer. I've seen that happen to many
45:22people and it's the most frustrating
45:25thing ever because it seems like you're
45:28doing the right thing by not structuring
45:30the answer, but you're actually not. As
45:34a consultant, you don't ask the client
45:36if they want you to give them an a
45:39structured answer or not. And if the
45:41client tells you, hey, just give me some
45:42ideas, your job is to give them a
45:45structured list of ideas, not random
45:48ones. So, two guidelines regarding these
45:51situations. Always have a structured.
45:54That's number one. Even if you're just
45:56speaking, speak with structure. Even if
45:59you have no paper, speak with structure
46:02by not using paper. Now, the second
46:05guideline is, as you get better at
46:07brainstorming and you can easily
46:09brainstorm issues with time to think and
46:13with a structure on paper, start
46:15practicing how to create structures on
46:18the fly and speak in a super structured
46:21way so that the interviewer can see your
46:23structure. They can see your issue tree,
46:26even if you don't use paper. This is an
46:28advanced skill and it's only worth
46:30developing it after you can do it with
46:33time to think and with paper. But it's
46:36well worth it, especially before final
46:38rounds. And one quick tip here, the best
46:41way to learn how to do this is to
46:43whenever you would need to give ideas or
46:45generate hypothesis in real life
46:48or at school or at work, you speak with
46:52structure and with prioritization. All
46:54right?
46:55So, this is how you brainstorm. This is
46:58the three-step approach to brainstorm.
47:01And now you should go ahead and apply
47:03that to our practice drills
47:06that you'll find right after this video.
47:08I hope this video was useful to you and
47:10I hope it helps you learn brainstorming
47:13well because this is a super important
47:16and useful skill. And I'll see you in
47:18the next one.

About this transcript

This page contains the full transcript of Case Interview Fundamentals: Lesson 10 [Brainstormings: The 3-step approach] by CraftingCases, generated from the public captions YouTube serves with the video. The transcript has 7,343 words across 1,137 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.