YT YouTube2Text

YouTube2Text → Transcripts → Case Interview Fundamentals: Lesson 17 [Frameworks: The 4-step approach]

Case Interview Fundamentals: Lesson 17 [Frameworks: The 4-step approach] — Transcript

by CraftingCases · 7,504 words · 1,168 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:01In the last video, I've shown you the
0:03fundamentals of frameworks. I walked you
0:06through how a framework is nothing more
0:08than a map to solve a problem. I've also
0:11shown you the difference between issue
0:13trees and conceptual frameworks and when
0:16to use each one. And I've shown you the
0:19first and only conceptual framework I
0:22will ever teach you, the 3 C's. More
0:25than that, I've shown you the anatomy of
0:27the 3 C's, why it is the way it is, and
0:30I've shown you how to create relevant
0:33questions for each category of the 3 C's
0:36or any other framework for that matter
0:38without having to memorize anything
0:40using the 5 W 2 H.
0:44The point was not to have you memorize
0:45the 3 C's or have you use it in every
0:48single location. The point was to show
0:51you a tool before I can teach you how to
0:53build one specific for your cases. But
0:57before we do that, we need to know how
0:59to use these tools, how to use these
1:02frameworks. In this video, I'm going to
1:05show you a simple, easy-to-use, and
1:07effective four-step process to build and
1:11communicate any framework regardless of
1:14it being the 3 C's or something else
1:16that you come up with as I will show you
1:18how to in the next lessons in this
1:21module. Anyway, let's not waste any more
1:24time and jump right in.
1:27So, the easiest, most effective way to
1:29build and use a conceptual framework is
1:31in four steps. This is true when
1:34applying the 3 C's to a business-related
1:36competitive strategy problem or when
1:39using any other framework created by you
1:42or not.
1:43The four steps are: Number one, define
1:46and structure the problem. Number two,
1:49raise key issues for each category.
1:52Number three, have a hypothesis for each
1:56key issue. And number four, create tests
2:00for these hypothesis.
2:02Defining and structuring is important
2:05because it will enable you to come up
2:07with a good structure for your case.
2:09Most candidates fail already in this
2:12step. Not defining and not understanding
2:15the problem well or forgetting key
2:17categories. For now, we only have the 3
2:20C's as categories, but I'll show you how
2:23to come up with more soon enough.
2:25Coming up with key issues means
2:28basically asking relevant and insightful
2:30questions on each category that you come
2:33up with in step number one. This is
2:35basically a checklist of questions that
2:37you must ask to understand the problem
2:40and solve your client's problems. Think
2:43of when you go to a doctor with a
2:44problem, they'll ask you a bunch of
2:46questions, right? To figure out what
2:48disease you have or what problem you
2:49have. So, this is just like when you go
2:51to a doctor. You have questions around
2:55uh different areas to your client or in
2:57your head, and in this step, it's where
3:00you talk about those specific questions
3:03to the problem.
3:04Now, the third step is to create
3:06hypothesis for each issue that you raise
3:09in step number two. Think about it. When
3:11you go to a doctor and they ask you
3:13questions, they usually have a
3:15hypothesis of what problem you have in
3:18their mind, otherwise they wouldn't be
3:20asking that question. So, each question
3:22they ask has a purpose, and the purpose
3:24is to test that hypothesis. Same for
3:27consulting. You only ask questions that
3:30the answers to it could change the final
3:32recommendation.
3:33Which means that you need a hypothesis
3:36for each question you ask of how that
3:38could change the recommendation even if
3:41you don't proactively communicate all of
3:43your hypothesis to the interviewer. You
3:46need to have them at least in the back
3:48of your head. And finally, in step
3:50number four,
3:52you're going to create tests for your
3:54hypothesis. So, to keep up with the
3:56doctor analogy, your doctor is going to
3:58confirm or disprove his his hypothesis
4:02with data from uh whatever you tell
4:05them, but also from clinical exams,
4:08imaging tests, lab tests, or even
4:11genetic tests.
4:13Uh now, the importance of having a
4:15hypothesis here is if your doctor has no
4:18hypothesis, they're just going to ask
4:20more and more and more tests and exams.
4:23Now, if they have hypothesis, they know
4:25which ones to ask. So, it's similar here
4:27for consulting. You need the hypothesis
4:30and then you test those hypothesis. You
4:32don't have to communicate all of the
4:34tests of your hypothesis to your
4:36interviewer, but and you don't even have
4:38to think of them beforehand, but you
4:40better be able to test each and every
4:42one of them if prompted. Each and every
4:45hypothesis. An untestable hypothesis is
4:48not a true hypothesis.
4:51Now, let's go back and look at each step
4:54into more depth.
4:57I'll start with step number one,
4:59defining and structuring. Just like in
5:02brainstorming, you need to define the
5:04problem before you structure it. Why?
5:07Well, because you can't structure a
5:08problem that you don't understand. So,
5:11how do you define a problem? And is
5:13there a situation where you don't have
5:16to bother with it?
5:17First of all, you need to have a clear
5:19understanding of any problem before you
5:22start structuring it. But problem
5:24definition becomes absolutely critical
5:28when the problem is poorly defined.
5:30Contrast these two questions, for
5:32example.
5:34I got this one from McKinsey's website
5:36practice cases, and it's a pretty
5:39typical of a McKinsey first round case.
5:42I'm not going to read through the whole
5:43thing because the point is that they
5:45give you a ton of information. The
5:47problem is pretty well defined and
5:49there's little room for mistakes here.
5:51Now, take a look at this case question
5:54which was given in a final round at BCG.
5:58The CEO of a bank has just visited one
6:01of the bank's branches and found it to
6:04be too crowded. And that's it. The
6:07interviewer didn't say anything else.
6:09There's not even a case question in this
6:11case question.
6:13In the second case, you obviously need
6:15to clarify and define the case much more
6:19than in the first one.
6:20The first one is much better defined, so
6:22you don't have to spend as much time
6:24doing it. So, this is the first lesson.
6:28Some cases are better defined than
6:30others, and you need to spend more time
6:32defining the case when it comes to you
6:35in a poor, poorly defined matter. A
6:38quick hint here. When the case comes
6:40ill-defined, that usually means that the
6:43interviewer expects to see your problem
6:45definition skills. That usually happens
6:48in final rounds because partners really
6:51value this skill as it's one of the key
6:53things that they do to their clients to
6:55define their problems or help them
6:57define their problems.
6:58Now, the second lesson in problem
7:00definition is that you define the case
7:02by using clarifying questions. And the
7:05third lesson that goes together with the
7:07second one is that you should aim to
7:09have three to five clarifying questions
7:12to define the problem. Try to have no
7:15more than five unless the case is so
7:18absurdly amorphous that you really need
7:20more than that. Let's dig into these two
7:23last lessons.
7:24Clarifying questions are broad questions
7:27aimed at defining the problem more
7:30specifically.
7:31You need to understand the specifics of
7:33the problem so you can create a
7:34framework specifically to the case you
7:37get. And that's true even if you're just
7:39going to use the plain 3 C's because you
7:42need the details for steps two, three,
7:45and four of building and communicating
7:47the framework. You can use any question
7:50to clarify, but there are three types
7:52that are particularly useful.
7:55Number one, those questions that explain
7:58the objective of the case in detail.
8:01This would be useful, for example, to
8:02the BCG case that I just showed you. In
8:05that case, the CEO seemed concerned with
8:07crowded brain bank branches. So, you
8:10could ask your interviewer, "Okay, I
8:12understand his concern with that. Is
8:14there anything specifically he's
8:16concerned with? Is he concerned with
8:18customer retention, for example, or with
8:21something else?" So, here you're trying
8:23to dig deeper and more being more
8:25specific into the objectives of the
8:28case. Number two, questions related to
8:31understanding the business model of the
8:33client or their industry. You can't
8:36structure a case without understanding
8:38how that business works. At a minimum,
8:42you need to understand how that business
8:44generates revenues, so everything
8:46related to its products, marketing, and
8:49how they sell, and also how that
8:51business operates, which is what will
8:54determine the business cost structure.
8:55So, everything related to how they
8:58manufacture and and and transport the
9:01products, for example.
9:03Number three, questions that ask for
9:06specifics of generic statements. And
9:09this includes asking for critical
9:11numbers when your interviewer gave them
9:13and asking about definitions of words
9:16that you might not know. So, if your
9:18interviewer said that you need to figure
9:20out how to grow a client's sales, you
9:23want to ask by how much and in which
9:26time frame, and if it's in dollar amount
9:29or percentage, things like that. If the
9:32interviewer says that customer churn has
9:34risen, you need to understand exactly
9:37how is customer churn defined in that
9:40client and by how much it has risen.
9:43So, even if you don't know what churn
9:45means,
9:46>> [clears throat]
9:46>> you should ask. But even if you know
9:48what churn means, you still need to know
9:51how that specific client defines churn
9:54because different industries define the
9:56number different differently. Also, in
9:59this category are questions related to
10:01units of measurement. So, if you need to
10:04get profits to last year's level, for
10:06example, you need to understand if it's
10:09the profits in the dollar amount or as a
10:11percentage of revenues or something
10:14else. Now, another guideline for good
10:17clarifying questions. Density matters.
10:20What I mean by density here is that each
10:23question that you ask must add a lot of
10:26value in your understanding of the
10:28situation. You don't want to ask 15
10:31different questions. Otherwise, 5 or 10
10:34minutes of your interview will have gone
10:36by and your interviewer will will assume
10:38that you're not going to create a
10:40structure to the problem. So, you want
10:42every question to count. And this is
10:45easier said than done.
10:46Practice will help you with that as well
10:49as focusing on the three types of
10:51questions that I just mentioned.
10:54Three to five well-formulated questions
10:57seem to be the sweet spot in most case
10:59interviews. You can't usually do less
11:02than three because most interviewers
11:04leave some issues ambiguous or
11:06ill-defined to see if you're going to be
11:09able to define the case well.
11:11In the vast majority of cases, you're
11:13going to be able to fully understand the
11:15situation with three to five questions.
11:19One more thing. Explicitly say that
11:23you're going to clarify some things that
11:25weren't fully clear before you pause and
11:27structure the case. This lets the
11:29interviewer know that you're aware that
11:32you should structure your thinking in a
11:34case interview and that leaves them a
11:36bit less anxious that, you know, is this
11:38guy going to stop and structure or is he
11:41going to just going to keep asking me
11:42questions?
11:44And my final and most important point.
11:46The role of clarifying questions is not
11:50to solve the problem.
11:52Any questions that will help you solve
11:55the problem should be within your
11:57structure. The role of clarifying
11:59questions is to be the minimum set of
12:02questions that will enable you to define
12:05and understand the problem so that you
12:08can structure it.
12:10Now, after you're done with clarifying
12:13and defining the case, it's time to
12:15structure it.
12:17In practice, the first thing you're
12:19going to do is to ask your interviewer
12:21for a couple minutes so you can
12:22structure the case.
12:24In case you're not exactly sure how the
12:26dynamics of asking for time and being
12:28silent while structuring the case work,
12:31you can watch videos of case interview
12:33examples. To help you out, Julio has
12:36curated the best videos of case
12:39interview examples from the web and
12:41reviewed each one of them. Just Google
12:43for Crafting Cases, Case Interview
12:45Examples, or something like that and
12:47you'll find it.
12:48In these couple of minutes in which
12:50you're silently structuring, you will
12:52want to write down on paper steps one
12:56and two, your categories and your
12:58issues. And you'll make a mental note of
13:00steps three and four so that you don't
13:02waste time writing them down but are
13:05able to say your hypothesis and tests to
13:08the interviewer. Anyway, to structure
13:10the case, you can have two paths. Number
13:14one, creating an issue tree in case the
13:16problem is simpler, shorter term, and
13:19more prone to be solved by problem
13:21isolation. Number two, creating a
13:24conceptual framework in case the problem
13:27is more complex, multifaceted,
13:30long-term, and prone to be solved by
13:32understanding what's going on from
13:34multiple angles. If you create an issue
13:37tree, it should always be built from
13:39scratch. If you choose to create a
13:42conceptual framework, you again have two
13:44paths. You can adapt from a proven
13:47framework such as the three Cs as I will
13:50teach you in the next video, or you can
13:52create one from scratch as I will teach
13:55you later in this module. With a
13:57well-defined problem and some practice
14:00building frameworks, you should be able
14:02to come up with the categories in about
14:0415 seconds. The rest of the time you get
14:06to create your framework is mostly
14:08dedicated to the next few steps.
14:12So, let's move on to step number two,
14:15raising key issues. As I mentioned
14:18before, the way to raise good key issues
14:20is to ask insightful questions using the
14:24five Ws and two Hs. But there's a bit
14:26more to it. These questions should be
14:29asked in a problem-specific
14:31way.
14:32I'm going to give examples of generic
14:34questions you could ask in a case
14:37and then show you the same questions
14:39specifically for an airline strategy
14:41case. But before I do that, I must warn
14:44you. Almost no candidate who learns to
14:47memorize frameworks raises key questions
14:50in a problem-specific way. 95% of them
14:54will just ask generic questions that
14:56could fit any case. It's funny. A couple
15:00days ago I was watching one of the
15:01videos at McKinsey's recruiting page to
15:04see what they're currently talking about
15:06their own recruiting process. One of the
15:08consultants in the videos said that what
15:11differentiates great candidates from the
15:14others is that when they talk about a
15:16client situation, they're talking about
15:19that specific client situation with
15:22things relevant to that industry. So,
15:24why don't more candidates do that? Well,
15:27one reason is that they've memorized
15:29frameworks and pre-made framework
15:31questions are generic almost by
15:33definition. Another reason is that it's
15:36tough to learn to make your questions
15:38problem-specific.
15:40But here's the good news. Once you learn
15:42to do that, you instantly become two to
15:46three times better than everyone else.
15:49This is one of those things that with
15:51just a little effort will bring you a
15:53ton of results. And the reason is that
15:56consultants never ask generic questions
15:58to their clients. They always ask
16:01problem-specific questions. By doing the
16:04same, you will sound more like a real
16:06consultant and less like a clueless
16:08candidate. So, let's run through some
16:11examples so I can show you that in
16:13practice. Suppose that I'm thinking of
16:16key questions for the customer part of
16:18the analysis for an airline strategy
16:20case.
16:21Instead of asking, "Who are my customers
16:24and how much do they buy?" I could ask,
16:27"What are the key segments within
16:29business and leisure travel and how
16:31large is each in terms of number of
16:34seats sold and revenues?" Instead of
16:37asking, "What products do they buy and
16:39what do they want?" I could ask, "What
16:42routes and add-on services are in large
16:45and growing demand and what are the the
16:48customers' main complaints?" Instead of
16:51asking, "Where and when do they buy?" I
16:54could ask, "What main channels do they
16:56use to buy tickets and how price
16:58sensitive are they regarding to when
17:01they buy?"
17:02Notice how the questions on the left
17:05could work in any industry, while the
17:07questions in the right will only work in
17:10an airline case. This improves your
17:13interviewer's perception of your skills
17:16due to a bunch of different reasons. For
17:18one,
17:19it just makes the problem-solving
17:21better. Specific questions have specific
17:23answers and they're better than generic
17:25questions. Another reason, it shows that
17:28you can interact with real clients.
17:31No one likes the consultant who comes in
17:34and doesn't seem to be thinking about
17:36the problem.
17:37Also, it's impossible to assume that
17:40you're not thinking about that specific
17:42problem when you're asking very specific
17:44questions. Just by doing this, you will
17:47immediately see your interviewer's face
17:49in relief as they realize that you're
17:52not a framework robot.
17:54Notice that you don't need to be full of
17:56knowledge about the airline industry and
17:59not even know specific terms related to
18:01it. All you need is to put yourself in
18:04the shoes of someone who buys plane
18:06tickets. Just being aware that when you
18:09board a plane, there are tourists and
18:11business travelers, or reminding
18:13yourself that each route is a different
18:16product, and that airlines would then
18:18try to sell you extra luggage and
18:20priority boarding and Wi-Fi and so on.
18:23Or reminding yourself about the last
18:25time you tried to buy a plane ticket and
18:28the prices kept going up on your
18:30favorite website.
18:32Just having these basic things in mind
18:34will help you a lot on being specific.
18:38Now, being problem-specific is really
18:41important, so let me give you another
18:43example. Usually, when you're doing
18:45competitor analysis, you want to know
18:47about your competitor's key resources
18:50because they give you a clue about what
18:52what your competitors can do and are
18:54likely to do.
18:56The generic way to do this is to ask,
18:59"What are my competitors' key
19:01resources?" But suppose you're in a case
19:03in the mining industry. You might ask,
19:06"What are my competitors' mines and how
19:08are they different from others?"
19:10Or if you're in a case in the retailing
19:12industry, you might ask something
19:14different such as, "What are my
19:16competitors' brands, store locations,
19:19and suppliers' contracts?" And if
19:21[clears throat] you're in a case on the
19:23pharma industry, you guessed it, the
19:25questions are going to be different
19:27again. Something like, "What are my
19:29competitors' brands, patents, and R&D
19:32pipeline?" Now, a lot of candidates feel
19:34that this is riskier. They'll say, "What
19:37if I forget a type of resource that is
19:39important in that industry? Am I not
19:42better off just by asking the generic
19:44question and having the interviewer tell
19:47me what resources are important?" And
19:49the answer is no. You can and probably
19:52should mitigate some of the risk by your
19:55working on how you phrase the question.
19:57Something like, "I want to know about my
19:59competitors' key resources. For example,
20:02what mining sites do they own and how
20:04are they different from each other?"
20:06This keeps the conversation open for
20:09them to suggest other types of resources
20:11that might be relevant. However, just
20:13being generic and not giving any
20:16specifics is the riskiest strategy
20:19because it will almost guarantee that
20:21you're not going to get the offer. If
20:23you're lucky, your interviewer will give
20:25you a second chance and ask you, "What
20:27types of resources do you think would be
20:29relevant to take a look at?" But why
20:32count on luck?
20:33Now, one thing that you might have
20:34noticed is that the variable that I used
20:37in these examples to make the issues
20:39problem-specific is the industry that
20:41the client is in.
20:43This is probably the most important
20:45variable, but you don't have to limit
20:46yourself to it. Let's dig deeper into
20:49the mining example.
20:52Some types of minerals are mined in
20:54thousands of mines across the globe,
20:57while others have just a few possible
21:00mines. We call these industries
21:02concentrated. And if the mineral that
21:05you're dealing with is in a concentrated
21:07industry, you might want to ask if your
21:09competitors have the ability to control
21:11prices as a reaction of your strategy.
21:15There's just a few competitors, just a
21:16few mines, so they might uh
21:20put
21:21increase or decrease the the supply so
21:24to change prices. Another example, say
21:26that the mineral you're dealing with has
21:29plastic substitutes. So, aluminum is a
21:32good example here. Many of its
21:34applications are on food packaging, for
21:36example, and that can be switched to
21:38plastics by the industrial companies
21:40that buy it. Now, in that case, you
21:43might want to add a question regarding
21:45your substitute competitors' production
21:48facilities, including their capacities
21:51and the cost of production. Final
21:53example here. Suppose that we're dealing
21:55with a mineral that has few buyers.
21:58Maybe it's a mineral that is only bought
22:01by, I don't know, chip manufacturers.
22:03And there's just a few of those around
22:05the globe. In this case, a key resource
22:08would be the relationship your
22:10competitors may have with these
22:12companies and their buyers, and you'll
22:14want to ask about that as well. Now, you
22:17should always think about how you can
22:20make your issues case-specific,
22:22including dimensions that aren't the
22:24industry itself. You can think of types
22:27of products sold as I did in this
22:29example, but you can also think of the
22:31country the company is in. So, a company
22:33in China is different than a company in
22:36Europe. You can think of the specific
22:38goals of the client or the context of
22:41the case. So, for example, if a case
22:43happens in a recession, this is
22:45different uh than if it happens in an
22:48economic boom. One last thing regarding
22:51the issues. A good list of key issues
22:54for each category of your framework
22:55should have five characteristics in
22:58order of importance. First, the question
23:01should be problem-specific.
23:02This is what I've shown you so far.
23:05Second, you should always have the most
23:07obvious issues to solve the problem.
23:10They might be obvious, but they're
23:11obvious for a reason.
23:13Third, you should have a few non-obvious
23:15issues as well, so interviewers can see
23:17that you can think outside the box.
23:19Fourth, the list should be as exhaustive
23:22as possible.
23:24Because we're dealing with qualitative
23:26concepts here, it's impossible to be
23:28completely exhaustive. This is not a
23:30math equation, but still, having a list
23:33of questions that seem to be
23:35comprehensive is much better than one
23:37that doesn't. Finally, number five is an
23:40extension of number three. You should
23:42have all the non-obvious issues that are
23:44critically important to solve this
23:47problem. Every case has a few of those,
23:50the incredibly important issues that no
23:53candidate mentions because they're so
23:54specific to that particular case. I put
23:57this as the fifth in this list, not
23:59because this isn't important. These
24:02issues are actually the ones that make
24:04or break real consulting projects, and
24:06especially in final rounds, they're
24:08really, really important. I put this as
24:11a fifth because less than 1% of
24:13candidates can pull that off
24:15consistently. It requires amazing
24:18business sense. Now, because this is so
24:20difficult, I highly recommend that you
24:23focus on the other four factors first,
24:25and after you're good there, you can
24:27focus on this fifth. The best way to
24:29improve in all of these five factors is
24:31to practice structuring extensively with
24:35access to well-developed answers. This
24:37is one of the reasons why you find
24:40comprehensive drills in this chapter,
24:42and it is also one of the reasons why
24:44we've created a course dedicated to
24:47structuring that has a huge practicing
24:49components through tons of drills.
24:52And hey, one quick tip here. If you're
24:54having trouble coming up with good
24:56frameworks and you get all the
24:57categories right, having a checklist of
25:00these five points on your side as you
25:03practice will help you figure out where
25:05to improve. Okay, cool. With the
25:08categories and the key issues laid out,
25:11we have the bulk of our framework ready
25:13to go. But even though we know what are
25:16the key areas to look at in order to
25:19solve the problem, and even though we
25:20know what are the key questions within
25:22each area, we still don't have a
25:24hypothesis-driven approach.
25:27Which leads us to our third step,
25:29creating hypotheses for each key issue.
25:32Now, a lot of people get confused about
25:34why you need hypotheses and why
25:36consulting firms have a
25:38hypothesis-driven approach in the first
25:40place. In fact, a lot of people don't
25:42even understand what makes a hypothesis
25:45a hypothesis.
25:47Let's take care of those concerns. There
25:49are two key reasons why you need a
25:51hypothesis that underlies each question
25:54that you ask when you're raising key
25:56issues.
25:57Reason number one is that because you
25:59want to be effective and efficient in
26:02your work, each question must be
26:04important enough to potentially change
26:07the final recommendation. You need to
26:09explain or to be able to explain to your
26:12interviewer why you're asking each
26:15question.
26:16The thing with people drawn to
26:18management consulting is that many of
26:20them are curious people. They want to
26:22know everything about the situation, the
26:24industry, and the client. I'm like that,
26:27you probably are, too. However, you
26:30don't have the time to learn everything
26:32in your typical consulting project,
26:35which means that you should focus on
26:36what's going to move the needle. You
26:38need a mechanism to sort the critically
26:41important from the merely interesting.
26:44And that's one role of the hypothesis.
26:47If you have an underlying hypothesis for
26:49each issue, that makes it more likely
26:53that each issue is critical, not just
26:55interesting.
26:57Now, you don't need a formal hypothesis
26:59to do that. You could just explain why
27:01each issue is critical, and that's one
27:04way to go. If you mention why you need
27:07to answer
27:08each of the questions that you you're
27:11raising the key issues step, that's
27:13already good enough for most standards.
27:17But there's a second reason why you need
27:19a hypothesis behind each issue.
27:22You need to answer the question in a
27:24very tangible, specific way. You're not
27:28just answering your questions in a
27:30generic way, you actually need to answer
27:33them in a data-driven, fact-based,
27:36specific way. And to show your
27:39interviewer that you can do that, you
27:41need a testable hypothesis.
27:44Now, this is all very theoretical so
27:46far, so let's bring that to practice.
27:49I'll use an example of the airline
27:50strategy case that we discussed a few
27:52minutes ago.
27:54Remember that one of the
27:55problem-specific
27:56issues that we had under the customer
27:59analysis was, "What are the key segments
28:02within business and leisure travel, and
28:05how large is each in terms of number of
28:08seats and revenues?"
28:10Well, when you mention that issue to
28:12your interviewer,
28:14they don't know if you're asking that
28:16because you have a purpose in answering
28:18that question or because you read in
28:20some case interview book that you're
28:22supposed to talk about customer segments
28:25in strategy cases. Now, there are
28:28several potential hypotheses that would
28:30justify this question.
28:33And you need to show at least one of
28:34them to your interviewer.
28:37So, I'll show you a simple one that
28:39works. If you want to learn more about
28:41how to formulate hypotheses and how to
28:44test them, we've got you covered in the
28:46hypothesis testing module. We've
28:48actually got a whole module for
28:51generating and testing hypotheses
28:53because doing that is a skill that
28:55appears in multiple parts of the case,
28:58not just in the framework. So, it's a
29:00building block in itself. Anyway, here's
29:03a simple hypothesis. After stating your
29:06issue, you could say to the interviewer,
29:09"And my hypothesis is that there must be
29:11at least one highly profitable segment
29:14in the market that is not served well by
29:17anyone, us or our competitors, and that
29:20we could target that segment from now
29:22on. An example of that would be
29:24last-minute business travel between
29:27large business centers."
29:29Now, this is a good hypothesis not
29:32because I've used the word hypothesis in
29:35it, but because it makes our issue super
29:38relevant, because it is testable, and
29:41because if it's true, it moves the case
29:45forward. If we found such a segment that
29:48is highly profitable and underserved,
29:51we'd have way more focus on how to
29:53create this airline's strategy. Now, the
29:56most important of these three things is
29:59that it's testable. If this hypothesis
30:02were not testable, it could never make
30:05the issue relevant nor drive the case
30:07forward in practice, which brings us to
30:10the fourth step, creating tests for your
30:14hypothesis.
30:16But before we do that, let's get
30:17something straight here. You're probably
30:19worried about how to create and present
30:22a framework with all these steps in
30:24enough time. I'll address that directly
30:27right after walking you through creating
30:30tests for your hypothesis. But still,
30:33it's important for you to be aware that
30:35you only have to do the hypothesis
30:37testing step as prompted by your
30:40interviewer. In a candidate-led case,
30:43you might prompt yourself and choose to
30:46test a specific hypothesis right after
30:48you present the framework so you can
30:50drive the case forward, but still,
30:53you're doing it as prompted and only one
30:56hypothesis at a time.
30:59Let me show you how to test the
31:00hypothesis that I just raised.
31:04I showed the interviewer that I can do
31:06it by showing them a hypothesis testing
31:08plan. It is basically a list of
31:10questions I want to answer with the data
31:13that I need in order to answer it. I'd
31:16first say that I'd build a list of
31:18segments in different ways, such as the
31:21reason for them to be traveling or the
31:23routes that they take or how many
31:26flights these segments uh flight per per
31:29year or the demographics of the person.
31:32So, I'd segment
31:34the population in different ways and try
31:36to find a segment that is large and high
31:40paying. This will ensure that they're
31:42highly profitable. So, I'd look at their
31:44size by both the number of tickets sold
31:47to this segment and the revenues they
31:49generate, and I'd double-check their
31:51price sensitivity by looking at how much
31:54they pay per ticket versus the average
31:57ticket price on the same flights.
32:01So, an example of a segment like this
32:03would be last-minute flights by
32:05executives in business travel routes.
32:08Another example could be high-net-worth
32:10individuals who travel every weekend to
32:14their leisure homes, for example.
32:16Then I'd survey the customers from the
32:18most promising highly profitable
32:21segments that we find to see what are
32:24their main frustrations
32:26with the flights they take. And I'd also
32:28look for what they desire the most if
32:30they could change anything about their
32:32air travel experience. So, what I'm
32:35looking for here is for that
32:37uh highly profitable segment to have
32:40lots of frustrations, which would
32:42indicate that they're poorly served
32:44currently. And finally, as a last step,
32:48I'd see our company's competencies to
32:51see which of those segments that are
32:53poorly served and highly profitable that
32:56we can serve better without reinventing
32:59ourselves. So, this is an example of a
33:02hypothesis testing.
33:04Now, one quick thing here. Some would
33:06say that all of this is obvious. It's
33:09obvious that you want to find a highly
33:11profitable segment with lots of
33:13frustrations that it's easy for your
33:15company to serve. And that's true, but
33:17it's not the point. The point is that I
33:20can
33:21uh show the interviewer that I can find
33:23a way to analytically prove that
33:27hypothesis and quantify its impact
33:30instead of just knowing what is obvious
33:32to everyone.
33:34Right? So, proving in a very systematic
33:37way is important because any gaps in
33:39reasoning show,
33:41and if there aren't any gaps in
33:43reasoning, it's easy to convince your
33:45clients to take action. It's It's much
33:47better to have that than to just have
33:50intuition about what's obvious.
33:54So, that finishes our four-step approach
33:57to frameworks.
34:00But there's still one concern left to be
34:03answered. How the heck are you going to
34:05do all these four steps in two or three
34:08minutes and still have the time to
34:10communicate all of it?
34:12Well, let's break that down.
34:15I'll start with the best practice on
34:16creating the framework in under 3
34:18minutes and then show you how to
34:20communicate it so you can maximize your
34:23interviewer's perception of your work.
34:26So, let's start with creating first. You
34:29got your piece of paper in the
34:30interview. You've defined the case
34:32clearly, and you've clarified any
34:35questions that you had, and you've asked
34:37your interviewer for a couple minutes to
34:39structure your case.
34:42Now, the first thing you'll do is step
34:43number one, creating the categories.
34:47You'll think which categories you want
34:49to consider. For now, you only know how
34:51to create the three Cs and other
34:53pre-made frameworks that you might have
34:55learned about before taking this course,
34:58uh but in the next video, I'll show you
35:00how to adapt the three Cs, and later
35:03you'll know how to create structures
35:05from scratch. So, you'll have more
35:06flexibility in creating the categories.
35:09And you'll write that down on paper.
35:12Thinking which categories you want to
35:14use and putting them on paper should
35:16take you about 15 to 30 seconds, no more
35:19than that.
35:20Now, the next thing you'll do is to
35:23think about key issues for each category
35:27and to write them down. You don't have
35:29to write them down as whole phrases as
35:31you see in the drills that I do, but uh
35:34but you just have to write down short
35:36snippets of text that will help you
35:38remember what are the key questions so
35:42that when you're presenting them to your
35:43interviewer, you can speak in a whole
35:46phrase format.
35:48So, write short snippets so you don't
35:50waste time, but uh be able to speak uh
35:54about these key issues in a whole phrase
35:56format. You should be able to come up
35:58with and write down each key issue in 5
36:02to 10 seconds each once you have the
36:04practice to do so. So, this is quick,
36:07but it's doable.
36:09It also means that you should be able to
36:11write down uh 12 key issues in a minute
36:15or two.
36:17While there's no right or wrong number
36:19of issues in your framework, you'll see
36:21through the examples in this module that
36:23most of my frameworks have between 10
36:27and 15 issues to consider, which is why
36:29I used the number 12 as a reference.
36:32Now, as you write down these key issues,
36:35you should also think about the
36:37hypothesis that underlie them. However,
36:41you don't need to spend time writing
36:44down these hypotheses. All you need to
36:47do is to have a hypothesis for each
36:49issue you write down so you can remember
36:52why you're asking each question, what is
36:55the logic behind it.
36:57For that reason, it shouldn't take you
37:00any extra time on top of
37:02uh creating the the issues to do that.
37:04Your mind will come up with the
37:06hypothesis at the same time that you
37:08think through the issues, right? They're
37:11inherently
37:13uh related to each other.
37:15Finally, you don't need to spend any
37:17time even to think about the test for
37:21each hypothesis in the few minutes that
37:23you have in silence to think through the
37:25problem.
37:26Although this is a critical step for the
37:28framework to work, it is a step that you
37:31only do after you present the framework
37:34for the few hypotheses that your
37:36interviewer will point you out to test.
37:39So, this whole process of creating the
37:41framework should only take you a couple
37:43minutes, no more than 30 seconds for the
37:46categories and about 2 minutes for uh
37:49writing the issues down and thinking of
37:52the hypothesis in your head.
37:54Will you be able to do this in a couple
37:56minutes as you're starting out? No,
37:58you'll probably need 5 or 10 or 15
38:01minutes to create a decent framework
38:04before you get more experience. But once
38:07you master it, it will be easy to do so,
38:09which is why you should practice
38:11structuring separately from whole cases
38:14so you get more and more practice on it.
38:17One question that comes up a lot, isn't
38:203 minutes too much to spend in silence
38:23creating the framework? I heard from
38:25someone that I shouldn't take more than
38:271 or 2 minutes. You'll hear that from a
38:29lot of people. Now, here's the deal.
38:32Which do you think is better from an
38:34interviewer's perspective? A candidate
38:37who has spent 3 minutes and created an
38:40amazing framework or one that has spent
38:431 or 2 minutes and created a mediocre
38:46one?
38:47It's really, really hard to come up with
38:50a great framework in 1 or even 2
38:52minutes. Even I can't do it.
38:55And most people can't do it in 3
38:57minutes, either. So, they didn't have
38:59the practice, but as you go through this
39:02course and if you want to take a step
39:04further and and join our structuring
39:06course, you'll be able to do an amazing
39:10framework in 3 minutes. And that extra
39:12minute, if spent to make your framework
39:15better than 99% of frameworks out there,
39:18is the best well-spent minute you can
39:21have in any interview. The results
39:24you'll get are going to be much better,
39:26and this will set the tone for the rest
39:28of the interview. The interviewer is
39:29going to be really impressed with you.
39:31So, they'll help you out more
39:34uh and and they'll just have a better
39:36impression overall.
39:37Now, eight or nine out of 10
39:39interviewers will be glad that you took
39:42a little bit extra time and actually did
39:44a good job.
39:45In fact, most of them won't even notice
39:47you took a minute longer. They'll be
39:49answering email in their smartphones or
39:51they'll be just hanging out. I have
39:54never seen or heard of an interviewer
39:56using a stopwatch or anything like that.
39:59Now, from time to time, you will find a
40:02more hurried interviewer that will cut
40:04you off and ask you to start presenting
40:07the framework to them
40:09uh at minute 1 and 1/2 or 2. Now, you
40:12won't lose any points if this happens.
40:15All you have to do is to say that, you
40:17know, you were taking a bit of extra
40:19time to think through some issues
40:21specific to that client situation and
40:24that, sure, you can start showing your
40:26work to them
40:27uh and then think of more issues as you
40:30present it to them.
40:32Now, once you create your framework and
40:34announce to the interviewer that you're
40:36ready, you need to communicate it to
40:39them.
40:40Here's how you do it.
40:42First, you walk them through each of the
40:45categories that you've written down on
40:47paper and explain why they're important.
40:50This should take you half a minute and
40:53it's important because it gives your
40:54interviewer a high-level view of your
40:56structure. You walk them through each
40:59category before you even start talking
41:01about the issues.
41:03Then, you go category by category
41:06explaining the key issues. Now, for only
41:09a few of them, you should explain what's
41:11the underlying hypothesis.
41:14The other hypothesis will be just in
41:16your head in case they ask.
41:19Just to make this extra clear, you will
41:21walk them through all of the issues
41:23you've raised and for a few of the most
41:26important, you will also mention what's
41:29your hypothesis behind that issue. Why
41:31is that? Well, you want to go through at
41:34least a few hypotheses so that you can
41:36show your interviewer that you're
41:38hypothesis-driven
41:39and very purposeful about the questions
41:41that you ask. You want to show them that
41:44there's some thinking behind each issue.
41:47And in most interviews, you don't want
41:49to go through all of the hypotheses
41:52because that would take too much time.
41:54You basically want to show your
41:55interviewer that you can do it and that
41:57you have hypotheses for every issue in
41:59your head and if they want to know your
42:01hypothesis for all the issues, they're
42:04going to ask follow-up questions.
42:06Now, going through all of these issues
42:08and a few hypotheses should take you
42:11from two to four minutes depending on
42:13how much detail you put into everything.
42:16Now, you must be thinking, what about
42:18the tests for these hypotheses? I don't
42:21need to think about them when creating
42:23the framework nor to communicate it to
42:25the interviewer. So, what's the point of
42:27that step? Well, what you need to make
42:29sure
42:31is your hypotheses are testable, but you
42:34don't have to come up with the specific
42:36test that you're going to use until you
42:38actually have to test the hypothesis.
42:40And that will come after you present
42:43your framework.
42:45In an interviewer-led format, the
42:47interviewer will ask you, how would you
42:49test a few specific hypotheses in your
42:52framework? They will do this after you
42:55present the framework as follow-up
42:57questions. And in a candidate-led
43:00format, they will ask you, well, what
43:02would you like to do next as soon as you
43:04finish presenting your framework? And
43:06guess what you'll do next? That's easy.
43:09You pick one or two critical hypotheses
43:12and test them to reveal critical
43:15information to help your client solve
43:17their problems.
43:18Testing the prioritized hypothesis is
43:21always the next step after presenting
43:24the framework unless your interviewer
43:27just asks you to do something else.
43:29One important thing though, I highly
43:31recommend you to practice this
43:34hypothesis testing step a lot when
43:37practicing structures or when doing mock
43:40cases. And the reason is that most
43:42candidates never ask for tests as
43:45follow-up questions as often as
43:47interviewers do. So,
43:50when you're practicing with other
43:51candidates during mock interviews,
43:53they'll never ask you to test
43:54hypotheses.
43:56And this means that if you don't make a
43:58deliberate effort to practice, you'll
44:01get no practice before your interviews
44:03and this is going to be tough on you.
44:05You're going to have a hard time knowing
44:08how to test hypotheses.
44:10So, to wrap up this video and the
44:12four-step approach to frameworks.
44:16First of all, make sure that the problem
44:18is well-defined before you try to
44:20structure it. You can't structure a
44:22problem that you don't understand.
44:24Once you have a well-defined problem,
44:26you'll know what type of framework to
44:28use as well as how to come up with the
44:31right categories and the most relevant
44:33issues and hypotheses within each
44:36category. Of course, you're going to
44:37learn how to create these categories
44:40uh in the next video on how to adapt the
44:41three Cs, but this step is critical.
44:45Second step.
44:46Make sure your key issues for each
44:49category are problem-specific.
44:52Of the five characteristics of a good
44:54list of key issues, this one's the most
44:56important. Remember to add specificities
44:59about your client's industry, country,
45:02or the specific context that they're
45:04going through to make your framework
45:06sound more specific and less rehearsed
45:10and memorized.
45:11Third thing. Use hypotheses for key
45:14issues to show why they're critical to
45:17your investigation. Hypotheses not only
45:20help you show the interviewer
45:21specifically what's on your mind
45:23regarding that issue, but also will help
45:26making the broader question of your
45:29issue specific enough so it can be
45:31tested. And finally, know how to create
45:35tests for the hypotheses to show your
45:37interviewer that you will know not only
45:39which questions to ask in a real
45:41project, but also how to answer them. A
45:44consultant that can generate hypotheses,
45:46but that cannot find ways to test them
45:49is not doing their job right. And
45:52regarding communication, when creating
45:54the framework, focus on the first three
45:57steps. And when presenting it, focus on
46:01the first two, but also sprinkle in some
46:04hypotheses to the most important key
46:06issues so your interviewer knows you're
46:09being hypothesis-driven
46:10in how you create your questions.
46:13Now you know how to use frameworks. You
46:17also know from the previous video how to
46:19manipulate the three Cs so you can
46:21actually create a framework. But how can
46:23you create more specific frameworks to
46:26your problems?
46:27How can you guarantee that you're
46:30touching on all categories relevant to
46:32the problem you have at hand?
46:34Well, the next video will show you how
46:36to adapt the three Cs. Learning to do
46:39that will not only show you how to make
46:42your business strategy structures
46:45different and better from those of all
46:48other candidates who have memorized
46:50frameworks, but it's also an essential
46:52step in the in learning how to create
46:55frameworks from scratch so you can solve
46:57non-business problems or very different
47:00business problems as well. So, uh this
47:03is the end of this video. Jump into the
47:05next one and I'll teach you how to adapt
47:07the three Cs to create better
47:10frameworks.

About this transcript

This page contains the full transcript of Case Interview Fundamentals: Lesson 17 [Frameworks: The 4-step approach] by CraftingCases, generated from the public captions YouTube serves with the video. The transcript has 7,504 words across 1,168 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.