YT YouTube2Text

YouTube2Text → Transcripts → Case Interview Fundamentals: Lesson 5 [Estimation: The 5-step approach]

Case Interview Fundamentals: Lesson 5 [Estimation: The 5-step approach] — Transcript

by CraftingCases · 7,447 words · 1,154 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00Hey guys, how are you doing? In today's
0:02video, I'm going to show you our
0:05five-step approach to doing estimations,
0:08to answering estimation questions.
0:10Just before before we start, I want to
0:12clarify something to you that people ask
0:15me all the time. So, I just want to get
0:16this out of the way.
0:18And the question everyone asks me is,
0:21"Why is it that we call them
0:22estimations, while everyone else calls
0:24them market sizing?"
0:26And that's because
0:29but not all quest not all estimation
0:31questions you you'll get will be market
0:33sizing. Many of those are going to be to
0:35estimate anything else. And the reason
0:37for that is
0:40when you're in the consulting job,
0:41you're going to estimate a lot a lot a
0:43lot of stuff. And the least of those are
0:46going to be market sizing. Although,
0:48market sizing estimations are going to
0:50be the ones that are most relevant to
0:52the project, the ones that are probably
0:54going to go in
0:57uh the slide deck that the client's
0:59going to get. Uh in the everyday life,
1:01you're going to be doing other
1:02estimations, like how much work some
1:04initiative's going to take, how much
1:06time, how much money some initiative's
1:08going to take, or how much whatever
1:10resource something will take, or
1:12whatever result something will bring.
1:14And it could be saving time, saving
1:16money, saving work, anything. And
1:19um it's so it's not always going to be a
1:22market sizing. Consulting consultants
1:24are doing estimations all the time. If
1:27even to decide what part of the project
1:30they want to tackle when they don't have
1:31a lot of time or anything like that. So,
1:34that's why they're going to give you
1:36other things to estimate, not only
1:38market size. And that's why I want you
1:39to practice not only market sizes. And
1:43the the way I'm going to show you this
1:44is I will
1:47quickly go over the steps, and then I'm
1:49going to do a drill myself.
1:52Uh I hope you try to do that drill
1:54before I do it, so that you can get an
1:57even better understanding of the steps.
1:59And then on the third step, I will
2:02explain each step of our approach
2:05thoroughly with with examples from this
2:07drill. So, you're going to understand
2:08this a lot deeper if you try doing the
2:11drill before me rather than only
2:14watching it and then watching the steps
2:15and just trying to absorb everything
2:18passively. Okay? So, first before I
2:21start this,
2:23I want to tell you why I'm giving you a
2:25way step-by-step process rather than a
2:27set of principles or a bunch of
2:29examples.
2:30So, that's because it's easier to do
2:33everything right and not get lost if
2:35you're following a step-by-step. And
2:38it's also easier to show the interviewer
2:41that you're getting all the important
2:42drivers, that you're going to get to the
2:44answer, and that you can discuss each
2:46issue in depth if you're separating each
2:50one of the steps.
2:52Uh
2:53I've coached
2:55candidates for over 2 years, and I've
2:58always started with estimations, and I
3:00have absolutely no doubt that doing
3:03clearly defined steps is the best way to
3:06do your estimations. And it's all
3:08spelled out to you here. So, just follow
3:10these steps and practice, and you'll be
3:11ready for any estimation cases.
3:14Now, let's go. Let me show you the five
3:16steps that I want you to follow.
3:19The first step is clarify the estimation
3:22scope. So, just make sure that you know
3:24exactly what it is that you're trying to
3:26estimate.
3:27Second one, structure the problem. So,
3:29what factors you're going to use and how
3:32you're going to use them. Don't put any
3:34numbers yet. Just indicate what you're
3:36going to do. Third, choose your
3:39assumptions. So, assign values to the
3:41variables that you just
3:43uh
3:44put together.
3:45And next, calculate. So, actually get to
3:48an answer based on the values you got
3:50and the
3:51math structure you said you were going
3:53to use.
3:54And last, check your answer. So, does it
3:56make sense? Compare it to something else
3:59to know whether it makes sense or not.
4:02Now, here goes the drill.
4:03Case question is, your client is a
4:06private equity fund who has been faced
4:08with a decision of investing in a
4:10rapidly growing candy company called
4:12Jupiter.
4:14Jupiter's main bets for their growth
4:16over the next year is the quality of
4:18their bubble gum. Your team was assigned
4:20to help this client decide whether or
4:22not they should invest in Jupiter.
4:25After the first problem-solving meeting,
4:27you're individually assigned with
4:28quickly estimating the size of the
4:30bubble gum market in the US.
4:32How would you do that?
4:34Now, take your time. Uh
4:37structure your answer, come up with your
4:39full answer to all five steps, and only
4:42then watch my answer and the explanation
4:46on each step.
4:48Just um before I start, I have a quick
4:50question that I want to ask. And that
4:52is, do I want to estimate only bubble
4:55gum as in candy or chewing gum like
4:58nicotine gum or other functional gums,
5:00too?
5:01And the interviewer is going to ask me,
5:03he's going to tell me actually,
5:04uh that's okay, you can estimate only
5:06bubble gum as in candy.
5:09Uh we don't
5:11uh we're not looking at other gum
5:14markets at all. Okay. All right. So,
5:17first thing I want to do is structure my
5:19problem, and the way I want to do this
5:21is
5:22bubble gum market in the US in dollars
5:25per year
5:27is
5:28number of pieces sold times the price
5:30per piece.
5:32And the number of pieces sold in a year
5:34is the number of consumers times the
5:36average yearly consumption per consumer.
5:39And the number of consumers is the total
5:42population times the penetration rate of
5:45bubble gum.
5:46Now, the Now, I want to start assigning
5:50value to my assumptions. And I want to
5:53start with total population. I know that
5:55the total population in the US is 320
5:58million, approximately. And now to the
6:01penetration rate of bubble gum. So,
6:03I believe that not that many people do
6:06not chew gum, even though teens and
6:09children chew more than adults.
6:12A lot of people chew gum. So, I will go
6:14with 80% although that might even be
6:17pessimistic.
6:19But that's my assumption here.
6:22And now to the average yearly
6:24consumption per consumer. So,
6:27I chew around
6:30five pieces of gum per day on average,
6:33but I'd say I'm way above average. So,
6:36I'm going to go with
6:38frequent consumers
6:40consuming around 20 pieces per week. Low
6:43frequency consumers consuming around
6:45four to five pieces per week. And low
6:49frequency consumers are probably more
6:50frequent. So,
6:53uh eight to 10 pieces per week on
6:56average is my assumption here at 50
7:00weeks per year. I'll go with 400 pieces
7:04per year on average per consumer. Now,
7:08to price per piece. So, I believe
7:11premium brands are priced on average
7:13around 25 cents a piece. And I'm
7:16thinking about those
7:20uh
7:21I'm thinking about single pieces, not
7:24uh little packs with several pieces
7:26inside. And low-cost brands are priced
7:29at around 5 cents a piece, I think.
7:32That's based on my experience. If we
7:33have any data on that, that would be
7:34really helpful.
7:36And I'm going to assume that they are
7:38sold at a 4:1 ratio. So, I'd say that
7:41the average piece of gum costs 9 cents.
7:45Okay? Now,
7:47uh let me just do some quick
7:48calculations here to get to my answer.
7:50So,
7:51320
7:53million * 80% that should be close to
7:55320.
7:5680% of
7:59300 is 240. 80% of 20 is 16. So, I'll go
8:02with
8:03So, this is 256
8:07um million consumers.
8:09times 400
8:12That will be 250 times 400 will be
8:151,000. So, 100,000. So, that's
8:20100,000 million, 100 billion. It's got
8:23to be close to 100 billion, but
8:26I can tell here I I can see that 4 * 256
8:31That's 2 to the 10th 2 to the 10th. So,
8:33that's
8:351,024. So, that's going to be 102.4
8:38billion. I'll go with 102 billion.
8:41And this is 10% less than 102 billion.
8:46So,
8:47um I'm
8:49uh Can I round this? I'd ask the
8:51interviewer in this case. Can I round
8:53this
8:54uh down a little bit to 9.2 billion? And
8:59actually round up a little bit. And
9:01yeah, in this case interviewer would say
9:03okay cuz I'm rounding
9:05uh uh less than 1% in difference. So,
9:09they'll say that's fine. And now,
9:11just before I say that this is my final
9:14number, let me check to see if this
9:16makes sense. So,
9:18mm I know that each American spends
9:22around $30 per year in bubble gum based
9:25on this value. So, that's
9:29uh 9 billion divided by 300,000
9:33300 million and that would be around $30
9:35per American. And that's between one and
9:38three meals, depending on where you are
9:41and what kind of meal you're getting,
9:43but
9:44I probably spend the equivalent to five
9:46meals in gum per year. So, this sounds
9:48reasonable given that I chew gum above
9:51average. Of course, maybe in the US
9:53people chew more gum than in Brazil.
9:56Maybe my experience is a bit skewed in
9:58this sense, but it should be in the
10:01ballpark.
10:02And if there's anything way off here, I
10:05believe it has to be in either price or
10:08average consumption per consumer. Those
10:10are my variables that I'm the least
10:12certain about and that could change the
10:14most.
10:15And now, as I said, we're going to go
10:17through
10:19each step of the process so that I can
10:22show you with examples what they mean.
10:25So, first one is clarify the
10:27estimation's scope. Um and this means
10:32make sure you know what you're trying to
10:34estimate. And the first thing I want you
10:35to learn here is ask any scope-related
10:39questions you may have. So,
10:41there are two two main examples that
10:43come to my mind in scope-related
10:45questions. And that is one, potential
10:49market. So,
10:51it's
10:52not uncommon that the interviewer is
10:53going to ask you to estimate the
10:54potential market for something. And if
10:56you're not sure what that means,
10:59uh I can explain it to you now. Like, it
11:01would mean the total market that could
11:06exist for that. So, people that could
11:07consume would consume if you do the
11:09right thing to have them consume or
11:11something like that. But, always ask
11:13what that means if you're not sure. And
11:15the second example here is even
11:18more important. And that is sometimes
11:21candidates will estimate the market size
11:24of something
11:27uh not in the way the interviewer
11:29expects. So,
11:31usually market size is a metric that's
11:34in dollars per year in total revenues.
11:38Some candidates will estimate that in
11:40number of customers. And they won't ask
11:43anything to the interviewer. They'll
11:44just come up with their structure, come
11:45up with their number, and say that their
11:47answer is
11:48uh 10 million customers or something.
11:50And that doesn't make any sense cuz
11:52that's not that's not what the
11:53interviewer is expecting. Sometimes the
11:55the interviewer wants the market size in
11:56units, which might make it a lot easier
11:59to do, but you won't ask and then you
12:02just go and do it in dollars and it it
12:04you've just lost time. You've proved
12:06that you're a consultant that will lose
12:07time instead of asking what you're
12:09supposed to do. So, if you're not sure
12:13what it is that you're trying to
12:14estimate or what metric it is or how you
12:17or anything like that, then just ask
12:19rather than start doing it. And next,
12:22beware of gray areas in your scope. So,
12:26uh to give you an example of this, uh in
12:29this drill, I asked about the
12:33market size if I'm trying to estimate
12:35market size of only bubble gum as in
12:37candy or as in other utility chewing
12:41gum. And it could be that other utility
12:45chewing gum are a possibility for this
12:47client and uh for to enter and to grow,
12:51and so they might want me to estimate
12:53that too.
12:54Uh another example would be
12:57TVs. If I'm trying to estimate the TV
12:59market size, it's quite common for
13:01people to go and estimate how many TVs
13:04there are, households, uh what's the
13:06replacement rate of those TVs, and based
13:08on that get to the market size of TVs.
13:11But if you don't ask, it could be that
13:14TVs sold to businesses are very
13:17important to this manufacturer
13:18specifically that that's your client or
13:20to the market in general, or maybe they
13:23just don't want to leave that out. So,
13:25you have to ask them if they want to
13:28know only if you can do only TV sold to
13:30consumers or are the TV sold to
13:32businesses important, too. So, like
13:35their TVs, you'll go into fast food
13:37chains and there are several TVs showing
13:39what the menu is or things like that.
13:41So, those TVs could be important. At
13:43least you have to show the interviewer
13:45that you're
13:46worried about what is actually the scope
13:48of the problem, so that you can do it
13:51correctly and that everyone will be
13:53satisfied. That's an important
13:55characteristic of a consultant.
13:57Okay, now next
13:58structure the problem.
14:00And this means, as I said, um
14:05show your plan to get to your answer
14:07from beginning to end. So, the
14:10interviewer has to analyze a bunch of
14:11things
14:12uh while they're interviewing you. Will
14:14you discuss a certain issue with enough
14:17details? Will you miss some of the
14:20important issues? Will you get to the
14:22final answer? Will you make some math
14:24mistakes?
14:26And that's confusing. It's a lot of work
14:28to evaluate all that at the same time in
14:31a candidate. So, that makes the
14:33interviewer never 100% sure you did
14:36everything right if everything's mixed
14:39up together. So, this brings me to my
14:41first point here. Have your interviewer
14:43understand the full logic of what you're
14:45going to do before you start estimating.
14:48So, show your full plan, your full
14:49structure first.
14:51And you're allowing them to look at two
14:54of every of all those things that they
14:55have to analyze. And that is
14:58uh if you're going to miss some
14:59important variable and if you're going
15:01to get to the final answer. And then
15:03they can get that out of the way of
15:05their mind and then they can evaluate
15:07you properly in all every and everything
15:09else that they want to evaluate. And
15:11that way the interviewer can be 100%
15:13certain that you're a great candidate by
15:15the end of the interview.
15:16And then they can advocate even harder
15:19for you.
15:20Uh you want to help the interviewer
15:22evaluate you. You two are the same team
15:24here, which is getting a good consultant
15:27into his firm. You want to help him with
15:29that. Okay? And next, use top-down
15:33structures to make sure that you're not
15:35leaving anything out and that the
15:37interviewer understands your ideas
15:38easily. So, best way to communicate an
15:41estimation structure is a top-down
15:44algebraic tree structure.
15:46If that doesn't come easily to you at
15:48first, it definitely pays off to spend
15:50some time practicing this.
15:53Um, if you're having a bad if you're
15:55having a hard time, it always helps
15:57trying to do different structures for
15:58the same estimation to test out your
16:00ideas. So, you can take one specific
16:02variable and then break it down in three
16:05or four different ways and then choose
16:08whatever you find best there. And then
16:10you take those two variables or three
16:11that you're breaking it down into and
16:13try to break them down into several
16:15different ways to again. And then you
16:19eventually you get the experience so
16:21that the tree structures will come
16:23easily to you. Let me show you my
16:24structure again to make sure we're both
16:26on the same page on what's an algebraic
16:29top-down tree structure.
16:31So, the what I showed you here is I have
16:34bubble gum market in the US in dollars
16:37per year. And I'm going to break it down
16:39into a formula, into a a product of two
16:43different variables, which are number of
16:46pieces sold and price per piece. Number
16:48of pieces sold per year times dollars
16:50per piece, that gives me dollars per
16:52year. And then I I broke down pieces
16:55sold into number of consumers times
16:58average yearly consumption per consumer.
17:01So, I have
17:02uh, people and
17:05uh, bubble gum per person per year. And
17:09those two together, bubble gum per year.
17:12Okay?
17:13Um, now
17:15uh, what I want you to see here is this
17:18is a top-down structure because it
17:20starts from the final result and it only
17:25in the end gets to the actual
17:27assumptions or the variables that I want
17:29I need value for.
17:31Number two, this structure is algebraic.
17:34Not only
17:36Uh, uh, actually this structure is
17:38algebraic. This means that I every
17:41single breakdown is a
17:45uh, math operation, algebra operation.
17:47And that's what you want to do cuz if
17:49you do that, you're making sure that
17:50you're always MECE.
17:52And third thing that I want you to see
17:54here, this is only a structure. There
17:56are no numbers here at all. Uh, I could
17:59have broken down average yearly
18:01consumption into
18:04average weekly consumption times number
18:06of weeks in a year. That would be okay.
18:09Uh, and I could and could not put 50 or
18:1152 or 53 weeks in a year. That's not
18:14what matters. Uh, I just don't want you
18:16to start you starting to discuss
18:19assumptions before you've shown the
18:21interviewer your full tree, your full
18:23structure. Okay? Now, let's move on to
18:28uh,
18:29another step that I want you to see here
18:31and that is
18:33um,
18:35explicitly state the algebra in your
18:37structure. So, especially when that's
18:39not obvious,
18:41I I need you to actually say to the
18:43interviewer, "I'm going to multiply this
18:46I'm going to divide this this by that."
18:49So, in this case, they're all obvious.
18:51But, one example of something that's not
18:53exactly obvious and you might want to
18:55explicit to the interviewer instead of
18:58just saying,
18:59uh, "one and the other." You want to say
19:01what you're going to do. So,
19:03uh,
19:04imagine you're doing a case where you're
19:06trying to estimate how many shoes one
19:09specific person buys per year and you
19:14you're going to use the replacement rate
19:15of formal shoes for to determine how
19:19many formal shoes they're going to buy.
19:20So, they have three formal shoes in
19:22their wardrobe, and you want to know
19:26how frequently they replace.
19:28So, and that's how many shoes they buy
19:30per year.
19:31Mhm. So,
19:33you could say I'm going to find the
19:35replacement rate of formal shoes as a
19:37function of the lifetime of the shoe.
19:40But, another way to say this is
19:42uh I'm going to find the replacement
19:44rate of formal shoes as the
19:47uh reverse of lifetime of formal shoes.
19:50So, one over the lifetime. Now, I'm
19:52saying what I'm going to do with this
19:54number, and this is much stronger, and
19:56this makes it clear to the interviewer
19:58what I need. Okay? Now,
20:01um
20:02let's move on to the next step. Choose
20:05your assumptions. So, now you've got a
20:08clear scope, you've got your structure
20:11with everything that you're going to do.
20:12You only need to plug in values in the
20:15ends of your tree structure, calculate,
20:17and get your answer. Let's talk about
20:19plugging in those values.
20:21The first thing that you need to know
20:23here is you have to use data and logic
20:25to back up your assumptions. So, you
20:27don't have to overcomplicate, but you do
20:30have to use logic to turn a number that
20:32you trust into something that you can
20:34use. So, let me show you an example of
20:36this.
20:37So, I said I chew around five pieces of
20:40gum per day, but I'd say I'm way above
20:43average. So, I would you I would assume
20:46that high-frequency consumers consume
20:48around 20 pieces per week, low-frequency
20:51consumers consume around four to five
20:52pieces per week. So, given that
20:55low-frequency consumers are more
20:56frequent, uh I would say that the
20:58average consumes around eight to 10
21:01pieces per week.
21:02Uh times 50 pieces 50 weeks per year,
21:05that will give me around something like
21:07400 pieces Uh uh per year for the
21:11average consumer. So, I use data and
21:14that is how much gum I chew, how much
21:17gum I think other people chew based on
21:19that. That's not the most most
21:20trustworthy piece of data, but it's
21:22still data. And I use logic. So, some
21:25people chew more, some people chew less.
21:27They have a different distribution.
21:29There are more people in one category
21:32than on the other. This is how I'm going
21:33to calculate them. And then I got to a
21:36value. That is not the most trustworthy
21:39value, but it's more trustworthy than if
21:41I had just said uh I don't know, maybe
21:44500, maybe 1,000, whatever.
21:47Uh this makes this much stronger and it
21:49shows that I have a skill for building
21:51actual quantitative models. Actually, I
21:54really built a quantitative model here.
21:56So, I showed the interviewer that if I
21:58happen to know these these numbers,
22:00maybe because someone told me, maybe
22:02because I did my research, or maybe
22:03because um
22:06uh
22:06for some reason I happen to know one of
22:09them, then I already have a model. I'll
22:11just plug in actually high-frequency
22:13consumers are 20% of total consumers and
22:16they consume 30 to 40 pieces per week
22:19and low-frequency consumers actually
22:22chew uh 10 pieces per week and they're
22:2580%. So, and well, the math what I'm
22:28going to do is already clear here. The
22:29interviewer knows that if I happen to
22:31stumble upon these numbers, I would have
22:34not only a trustworthy number, a
22:37really really useful number here that
22:39could be used in a project. Okay? Next,
22:42uh focus on what matters. So, you want
22:45to show the interviewer that you're
22:47going to spend your time as a consultant
22:49doing the most relevant work to the
22:51client. And spending time in only in the
22:54most relevant assumptions will show that
22:56to the interviewer. So, if one
22:57assumption won't change much its value
23:00or doesn't change much the final result,
23:03you don't want to spend several minutes
23:05in it. And that would mean several days
23:08in a real project. You don't want to do
23:09that. You want to go as quickly as you
23:11can through it. Let me show you how I
23:13did this in this case. So, I said that
23:16penetration rate of bubble gum uh
23:19probably around 80% because not that
23:22many people chew gum, and even though
23:24teams and children chew more, uh
23:27not that many adults don't chew gum. And
23:3180% might even be pessimistic, but the
23:33reality is it could be 90, it could be
23:3670. It wouldn't be feasible to be like
23:38under 50.
23:40Uh so, I'm not going to spend so much
23:42time in here. 80% is fine. versus price,
23:46for example. Uh I don't
23:49I don't know much about price of gum. Um
23:52and I Well, at least off the top of my
23:55head. And given that I'm doing this per
23:57piece, I know even less cuz I'm used to
24:00buying packages of gum. So,
24:02I could I
24:04I could, if I were uh not paying
24:07attention, trust the price to be 1 cent
24:09or 50 cents
24:10uh versus 10 cents a piece if I did not
24:13think a little about it. And there are
24:15many different types of gum,
24:17positioning, uh amount in a box, etc.
24:19So, that could be confusing. And the
24:21same is true for quantity a year. It's
24:24not obvious at all. Like, I can't think
24:25how much gum I eat per day. I even can
24:28think how much gum I chew per week, but
24:30per year?
24:32That's not intuitive at all. So, I want
24:34to spend more time in those assumptions
24:36cuz I won't just look at them and say
24:38this is off. If I were to look at
24:40penetration rate of bubble gum, and it
24:42said 5 or 10%, I would say definitely
24:44say this is off, something's wrong.
24:46Okay?
24:47Uh next thing I want you to learn here
24:50is you can get creative with your
24:52numbers. Here,
24:54uh you can watch my answer again for
24:55this. I did compound average as it I did
24:58segmentations, I made up ratios based on
25:00my experience. And you can do anything
25:02as long as you're backing up your
25:04assumptions with data and you're
25:06following some logic. And you you're
25:08going to see many examples of drills,
25:10many examples of this in other drills.
25:12And to be honest, the only way to get
25:15good at this is practicing. If you're
25:17having a hard try hard time,
25:19try and do many estimations, try and
25:21come up with several ways to do the same
25:23assumption, and eventually you'll get
25:25you get
25:27proficient in this. It's not that hard
25:30and it's honestly kind of fun. So,
25:33go ahead and do it several times if
25:34you're having a hard time. And last,
25:37it's okay to ask. You can ask for the
25:40interviewer for some of the numbers. I
25:42didn't ask in this case and I'm not
25:44going to ask in any of the drills cuz of
25:46course I don't have an interviewer in
25:47front of me, but you can always ask and
25:50maybe they'll know something.
25:51Maybe they won't, but that won't hurt.
25:54Of course, if they just say no to
25:55everything if you're asking every single
25:57assumption. Do you have this number? Do
25:58you have this number? And they'll say
26:00they don't have any data, get the hint
26:02and find your own assumptions.
26:04But for things like
26:06the population of a city that you're not
26:08used to or of a country that you don't
26:10know very well, you can ask. That's
26:12fine.
26:13If they need you to come up with your
26:15own values, then you're of course going
26:17to have to find
26:18a city that's similar to that that you
26:21think is similar to that, a country that
26:22you think is similar in size or in
26:24population, and just use use that value
26:28if they say that's fine. Next,
26:31calculate.
26:32I don't want to spend too much time on
26:34this because
26:36well, the goal of this video is not to
26:39turn you into a human calculator and
26:43I'm going to go over just some
26:44principles which if you're not having
26:47specific problems with math and you're
26:51not making math mistakes in every single
26:53case, they are pretty much everything
26:55you need to know right now about
26:57calculations so that you can finish
26:59estimations comfortably.
27:01So, the first principle is
27:04uh calculations must be precise and
27:06efficient. Precise is more important
27:09than efficient.
27:10So, don't make mistakes, but if you do
27:13make them, you have to catch them, and
27:14I'll talk about this in a second.
27:16And the thing with efficiency is it
27:19doesn't mean
27:20it
27:21it doesn't mean being able to do 13 13 *
27:2549 really quickly. It means taking
27:28shortcuts. That's what makes a
27:30consultant different in their way of
27:31making
27:33uh calculations. So, it doesn't mean
27:36that you don't need to use a paper. It
27:37doesn't mean that you're really quick at
27:39it. It means that you can take
27:40shortcuts. So, instead of doing 13 * 49
27:43in your mind really quickly, you can go
27:4613 13 * 50 is 650.
27:50Um
27:51so, I'm just going to subtract now 2% of
27:54that, and that's going to give me what
27:5613 * 49 is. Okay?
27:59Uh so, let me show you how this these
28:03two showed in this drill. And in this
28:06first
28:07uh in the second calculation that I did
28:09here, I could instantly see that 4 * 256
28:14is 1,024 cuz that's 2 to the 10th. And
28:17that gave me a really large shortcut cuz
28:20I didn't need to do any math on paper or
28:22in my mind. I could just easily get to
28:25102 billion. And in the next
28:28calculation, what I did was instead of
28:31doing 9 cents * 102 billion,
28:34I got 10% of one of 102 billion, which
28:38is 10.2 billion. And then instead of
28:41taking 10%
28:43exactly 10% out, I took 10 I took out
28:4610% of 10 billion and that brought me to
28:499.2. Now, this is not correct. It should
28:53be 9.18 billion, but I didn't want to
28:57lose time for those point 02 billion
29:01which were not going to be so so
29:03relevant in my answer. We're not going
29:05to be relevant at all.
29:07Uh, so I just did it. So, I was still
29:09precise, but most important I was
29:11efficient without giving up uh
29:15relevant precision cuz this is not
29:18really going to change whether
29:20uh um this market is correct or not cuz
29:23I'm not that sure of any of my
29:24assumptions in order to uh
29:27spend so much time on that little on
29:31that little little part of the problem.
29:33Okay?
29:34The next thing I want you to learn in
29:36calculations is always always always
29:39reality check before you start
29:41calculating. So, what does that mean?
29:43That means doing a ball park calculation
29:45of your math before
29:48uh you actually do the math if you need
29:50to.
29:51Uh that's always going to show you if
29:53you're wrong. And the interviewer is
29:54never ever going to catch one of your
29:56math mistakes. You're always going to
29:58make math mistakes. It doesn't matter if
30:00you're an economics PhD, a physics PhD,
30:04it really doesn't matter. You are going
30:06to make mistakes. You want to catch
30:07those mistakes before the interviewer.
30:09The way you do that, reality checking.
30:11Let me show you how I did this in this
30:13case. Those are simple calculations. You
30:15probably wouldn't need them so much, but
30:18it's still important to do to get the
30:20habit and you're really going to need to
30:22do those in the drills that you're going
30:24to see next. Now,
30:26uh here in the first calculation I
30:29instead of just doing 320 million times
30:3180%, I first said this should be close
30:34to 320 million cuz 80% is close to 100%.
30:38And then in the next calculation,
30:41instead of just going for my my what was
30:44in my mind which was
30:464 * 256 is 1024 and then I just have to
30:50deal with the zeros. First I said 4 *
30:53250 is 1000. So that should be 100
30:57billion and only then I did something
31:00something and then I got 202 billion and
31:03I was like okay okay this is close.
31:05I could have messed up in this math
31:08because there are a bunch of zeros and
31:09got into 10.2 billion or 1.02
31:14trillion
31:16I wouldn't have known if I didn't do
31:19this rally check before. Okay, next step
31:22check your answer.
31:24This last step is what will set you
31:26apart from 95% of candidates. You want
31:29to check if your answer makes sense and
31:31consultants do that all the time because
31:33you can't really
31:35audit complex models quickly all the
31:37time. So you just see if the final
31:39result makes sense whether you're
31:41checking a teammates
31:44model or your own model
31:46you want to see if it makes sense if the
31:48result makes sense just see if the model
31:50working properly and showing an
31:53interviewer that you already have this
31:55skill in you will put you ahead of all
31:57those candidates. I say it will put you
31:59ahead of 95% of candidates because
32:02most people don't even know that they
32:03should be doing this and most several
32:06people who know are not doing this
32:08right. So let me show you how to do this
32:11and this will be a great leap in your
32:13performance. The first lesson I want you
32:15to learn here is compare your value to
32:17something to get an idea of whether it
32:19makes sense or not.
32:20The only way to make sense of a number
32:22either to know what is right or wrong or
32:25if it's high or low is to compare it
32:27with some other number. So take blood
32:29pressure. If I tell you my
32:32my blood pressure is 120 over 80 how do
32:35you know if it that makes sense or not?
32:37What if I tell you it's 520 over 120?
32:41You can only know it if you know the
32:42blood pressure of a normal healthy
32:45adult. And if you're a doctor or if
32:48you've been in touch with that before,
32:49then you will know it. If you don't, you
32:51have to ask someone, you have to do some
32:53research, you have to find something to
32:55make sense out of those numbers. Uh
32:58the same goes for potential market for a
33:00new product or expected sales in a new
33:02drugstore. Uh
33:05Let's talk about potential market for a
33:06second. Let's say you're trying to
33:08estimate the potential market for
33:11high-protein potato chips. So, instead
33:13of a protein shake, you'd eat potato
33:16chips. That's the new product that I
33:18just came up with. And you came up with
33:20a potential market of $500
33:24in the US, of course.
33:26Uh I only know that that's too much
33:29because I'm implicitly comparing it to
33:31something. So, I could be implicitly
33:33comparing it to GDP figures or to other
33:36markets, like the total pro- whey
33:38protein market is 30 billion.
33:40Uh
33:41or to companies' revenues, like Amazon's
33:43revenues is 100 billion. Uh or
33:47I could be look look at this number and
33:49say, "Well, 500 billion that means that
33:51each Americans will spend
33:54uh $1,500
33:56on high-protein potato chips per year."
33:59So,
34:00all of those are comparisons and I would
34:02not be able to say that 500 billion is
34:05too much or too little if I were not
34:07comparing it with something.
34:10Um and you see how I manipulated these
34:12numbers, how I changed them to
34:15do it my reality checks,
34:18to make sense of them.
34:20Uh you can do that.
34:22You can do whatever you want to the
34:23number as long as you're coming up with
34:26a logic to test whether this number
34:28makes sense or not. You want to come up
34:30with other numbers you can compare this
34:32with or whatever consequence or
34:34component of this number two so that you
34:37can get to um
34:40an idea of whether it's
34:42too high or too low or not too high or
34:45not too low. Let me show you how I did
34:47this in this case and then
34:49uh you get a better sense of what I'm
34:50talking about. So
34:53Now, you can get creative with your
34:54number.
34:55So,
34:56uh
34:57when
34:58I said each American spends around $30
35:01per year on bubble gum, that's between
35:03one and three meals.
35:05And I probably spend around the
35:06equivalent to five meals in gum per
35:08year, so that sounds reasonable. What's
35:10this saying? This is saying
35:12I'm comparing the American expenditure
35:15per year what everyone's spending
35:18on bubble gum
35:19to the price of meals so that I can
35:21compare that to my
35:23expenditure in bubble gum. So, I'm
35:25comparing it to something. I'm comparing
35:26it to my expenditure in bubble gum.
35:29I put meals in the middle of those
35:31because I don't spend in dollars, I
35:34spend in Brazilian reais. So, I want to
35:37compare it to something comparable and I
35:38used food for that.
35:40Uh
35:41when I was doing this case, I actually
35:44made a math mistake. I missed a zero and
35:47I caught this during this reality check.
35:49I caught uh I got to 920 million dollars
35:54and I got to $3 per American.
35:58And that didn't make any sense at all
36:01and so I started looking in my model to
36:03see if I could have gotten
36:05any assumption really, really wrong or
36:07if I did a math mistake and in the end I
36:09made I made a math mistake.
36:11Okay?
36:12So, the next lesson here is
36:15if you can find a way to say what you
36:18would like to do. So, sometimes you
36:19won't be able to find a number.
36:21Sometimes you're going to be dealing
36:22with something way off, way far away
36:25from your reality.
36:26Uh that's not going to be quite that
36:28common in real case interviews, but it
36:30is going to be common in your practicing
36:31cuz you're going to get uh if you're
36:34from a country in Europe, you might get
36:36exercises from uh Brazil or the United
36:39States or from China and
36:41all vice versa. And
36:44that should be the last last last
36:46resource ever. If you can do anything as
36:49a real reality check, that's better than
36:52doing what I'm going to show you, which
36:53is say what you'd like to do. But, if
36:56you can't find a number, that's a
36:57resource better than saying nothing
36:59about it.
37:01And
37:02you're also going to practice so that
37:03this never happens, but it might. So,
37:07you still want to show this ability to
37:09the interviewer even if you don't have a
37:11number to do something with. So, you
37:14want him to know that you're going to be
37:15a consultant who actively checks their
37:17numbers to see to see if everything's
37:19going in as expected with your model,
37:22with your numbers, with your slides, and
37:24the project that you're following or
37:26anything like that.
37:27And let me give you some examples of how
37:29this could have happened in this case,
37:32and that's going to help you understand
37:34what I'm talking about. So, uh here
37:38let's suppose that uh I didn't
37:40I couldn't compare this to how much
37:42gummy I eat per year for whatever reason
37:45or the interviewer said, "Well, I don't
37:47trust that you spend that much income or
37:49that you know how much people spend in
37:51gum. What would you compare it to?" So,
37:53I could compare this value to the
37:57gummy bear market in the United States,
38:00to the bubble gum market in Canada.
38:02Uh of course, I need to do some
38:06uh
38:06malabarism with those numbers. I need to
38:08change them a bit. So, if I were to do
38:10the gummy bear market in the USA, okay,
38:13I could compare it straight away. Or
38:15maybe I could think that people spend a
38:18little bit less with gummy bears than
38:19with uh bubble gum, and also less people
38:23consume that. Uh
38:25if I were to compare this with the
38:26bubble gum market in Canada, I'd have to
38:30uh
38:31change this a little because populations
38:33are very different. The
38:35the size of population is different. If
38:36I were to compare this to the bubble gum
38:38market in the UK, I'd have to
38:41uh
38:41weight this based on population. I could
38:44as I could uh
38:47I could compare this to the GDP uh for
38:51candy in the USA. And this would only
38:53help me make sure that this isn't too
38:54large, but it could also help me a bit
38:57to know that this isn't too little.
38:59Uh I could compare my value for the
39:03average American consumption of bubble
39:05gum,
39:06uh which is the variable that's most
39:08likely to be wrong.
39:10Uh I could compare this to Mars
39:12revenues, the
39:14candy company.
39:16Uh I could compare it to any of those
39:17values. And I would say to the
39:18interviewer something like, "Well, I I
39:20really can't find something to benchmark
39:23this with, but I'd like to check if it's
39:25if it really makes sense.
39:27Do you have any of those numbers?" And I
39:28would say some of those. And if they say
39:30no, I'd say, "Well,
39:32uh
39:33that I'm not sure. It looks okay to me,
39:36but if I had some time, I would find any
39:39of these to compare with my number and
39:42get an idea of whether it's
39:44way off the ballpark or not."
39:46Now,
39:47the next lesson is if you have to adjust
39:50your value, go for variables that can
39:52change more. So,
39:54it's quite common for interviewers to
39:56end the interview, the the estimation
39:58question by saying,
40:00"All right. Well, let's say that your
40:01number is two times off." Then they'll
40:03they'll say something that's twice
40:06as much as the value you said or half
40:08the value you said or five times or
40:10anything like that.
40:11And they're going to say, "Well, that's
40:13wrong. This is the real value. What do
40:15you think happened?"
40:16When that happens, you go straight for
40:19the variables that have two conditions.
40:21One, high chances of being wrong, and
40:24two, could be different by a lot. I
40:26already told you which one of those
40:29uh were important for me. Here, I said
40:34um if there's anything way off, it has
40:36to be in price or in average
40:37consumption. I've explained you why,
40:39because
40:41uh
40:41if I were a bit naive, I could believe
40:43that
40:44uh bubble gum were priced in 1 cent a
40:47piece or even 50 cents a piece, and that
40:51would sound a little bit off, but not
40:52that much off. And I I could believe
40:54that the average American consumer would
40:57consume 1,000 pieces of gum in a year or
41:00even 40. I I just have to be a little
41:02bit naive for that. But those can change
41:05much much more than of course the
41:07American population, but also than the
41:08penetration of bubble gum. Like if it
41:11were to change much and still sound off
41:13to me, it would go down to 50% or up to
41:16100, and it would change if it were down
41:18to 50% 30% my result. It would not get
41:21close to changing 2x or even 5x. And if
41:25it were to change up, it would also
41:27change in like um
41:2950%. So, wouldn't really make a
41:32difference. So, if you have to adjust
41:35your value, go straight for the
41:37variables that can change more. All
41:39right?
41:40Also, when I was doing this drill, I was
41:42proactive with this. The while I was
41:45answering, I said, "If there's anything
41:46way off, it has got to be in price or
41:49average consumption because this and
41:50this and that." You don't have to be
41:52proactive like that. Uh of course it's
41:54good, but it won't change so much. So,
41:57if the interviewer asks you what could
41:59be wrong, then you definitely have to
42:01point it out, and you have to point it
42:02out this way. You go for the variables
42:04that have more chance of being wrong,
42:07and that can change more in a reasonable
42:10span of value. Okay?
42:13And next,
42:15don't You don't need to wait for the
42:17interviewer to ask to to ask you this.
42:20Uh you can uh to ask you the
42:23uh the reality check question. To ask
42:26you, "So, do you think this is correct?"
42:28You can be proactive and you can after
42:30you finish your answer, say, "Okay, I
42:32got to
42:33uh 9.2 billion. Let me just check if
42:36this makes sense." And then you do
42:38whatever you want as a reality check.
42:41And last, don't just check your math.
42:43Please, please, please, don't just check
42:45your math. This is very, very common.
42:48When the interviewer asks candidates, uh
42:50"What do you think of this number?" and
42:52they don't know that they're supposed to
42:53do reality checks, their instinct is to
42:56check their math and say, "Yeah, my math
42:59is all right. I trust my assumptions, so
43:01yeah, this is correct." And sometimes
43:03people think that that's what the
43:04interviewer means, that's what the
43:06interviewer wants them to do. And
43:07sometimes they just want to show some
43:10confidence. And
43:12uh it happened all all all the time when
43:15I was coaching candidates before they
43:18knew that they had to do reality checks.
43:20That's not what the interviewer is
43:21looking for.
43:22Uh they're looking for
43:24seeing you use your creativity to check
43:28if a given number makes sense and to see
43:31if you're able to do that because as I
43:33said, you're going to be doing that
43:34every day as a consultant and you want
43:36to show your interviewer that you're
43:39going to be able to do this in the
43:40day-to-day.
43:41So,
43:43in that moment, what they want to see is
43:45a reality check. Okay? So, this was our
43:49five-step approach to estimation
43:51questions and with this and some
43:53practice, you're going to be able to ace
43:55any estimation case you get.
43:58Uh I hope this helped you and I'll see
44:00you in the next video.

About this transcript

This page contains the full transcript of Case Interview Fundamentals: Lesson 5 [Estimation: The 5-step approach] by CraftingCases, generated from the public captions YouTube serves with the video. The transcript has 7,447 words across 1,154 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.