YT YouTube2Text

YouTube2Text → Transcripts → Case Interview Fundamentals: Lesson 8 [Partner Estimation Drill: BMW Call Center]

Case Interview Fundamentals: Lesson 8 [Partner Estimation Drill: BMW Call Center] — Transcript

by CraftingCases · 14,243 words · 2,303 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00Hey everyone, how's it going today?
0:02Before I show you the next video in this
0:04module, I need to give you some context
0:06because there's a lot to be learned from
0:08this video, but if you don't have the
0:09right context, it's going to be more
0:11confusing than instructive. So, let's
0:13get right to it. Number one,
0:15estimations can be a part of a case or
0:18they can alone be the full case. The
0:21ones you've seen so far in this module
0:23would most likely be a part of a case
0:26and would be followed or follow
0:28brainstorming, framework, hypothesis
0:30testing questions. But this one that
0:32you're going to watch right now is the
0:34full case by itself. So, don't get
0:36surprised when you notice that the video
0:38is much longer than the other videos.
0:40But you're also going to notice other
0:42particular features of this video.
0:45Number one, you're going to notice that
0:47I interview Bruno in this video instead
0:49of doing a drill by myself.
0:51Number two, you're going to notice that
0:53there's pressure from the interviewer,
0:55which in in the case is me.
0:57And number three, you're going to notice
0:59that both the case and the follow-up
1:01questions are a lot harder than in the
1:03other videos within this module.
1:05And why is it so different than the
1:07others?
1:08Well, the answer to this question is
1:10piece of context number two, which is
1:13this video doesn't originally belong
1:15here. This might be a bit confusing, so
1:18let me take a step back. Right now, what
1:20you're watching is our free course, Case
1:22Interview Fundamentals.
1:24And we also have other paid courses and
1:27one of them is called Master Market
1:29Sizing, which is an advanced estimations
1:31course.
1:32In Master Market Sizing, we teach you
1:34how to go from knowing absolutely
1:37nothing about market sizing and
1:39estimation questions to being able to
1:41solve even the toughest ones in
1:43industries you've never heard of and the
1:45most absurd estimation questions you can
1:47imagine.
1:48And Master Market Sizing has two theory
1:51modules where we teach you our blueprint
1:53to estimation cases and the techniques
1:55to become a master in the skill, and
1:57three practice modules, which are
2:00typical estimation drills,
2:01non-conventional estimation drills, and
2:04partner estimation drills. This video
2:07was taken straight from the partner
2:08estimations module, and it was actually
2:11a case that Bruno got in one of his Bain
2:13final round interviews.
2:15Now, why is it here?
2:17The reason we put this video in Case
2:20Interview Fundamentals is because we
2:21felt this course had a few shortcomings
2:23up until now.
2:25A few things you couldn't grasp from the
2:27estimations module in our free course,
2:29now you can. Number one, what does it
2:32look like to be pressured from the
2:33interviewer? Previously, you couldn't
2:35see that because you could only watch me
2:37doing a drill without the interviewer.
2:40Number two, what does a hard estimation
2:42case actually look like? Cuz all the
2:44other videos in this module are built to
2:46teach you the basic principles, so we
2:48couldn't really make any of them too
2:50hard. Number three, what does an
2:52estimation look like when it's the full
2:54case? And number four, how different can
2:58a partner estimation be? And honestly,
3:01this is the least different partner
3:03estimation in the course, but we didn't
3:05want to make things too complicated
3:07here, as I wouldn't even have the space
3:09to teach you everything that you need to
3:11know to understand the most complex
3:13estimation cases out there. The idea
3:15here is just to show you that partner
3:17cases can get very specific, and that
3:20you're expected to be able to solve
3:22any kind of estimation, estimate really
3:24anything, not just a few simple markets.
3:27Now, I don't want to keep you here
3:29forever. I know you want to move on,
3:31watch the video, so let's get right to
3:32it. Hey everyone, what's up? Julio here.
3:35So, uh in this partner drill, we're
3:38going to do a drill that Bruno got in
3:41his final round at Bain. It was actually
3:44a case back then, not a drill. Yeah, it
3:46was an entire case, and it's going to be
3:48an entire case as well. And
3:51uh
3:51it was with a particularly tough partner
3:54whose role I will be playing in this
3:56drill.
3:58And I want to remind you of something
4:00before we move on with this drill. And
4:02that is you should know by now that in
4:05partnered cases the follow-up questions
4:08are a particularly important part of the
4:12case where the partner's looking for
4:13specific things with each question and
4:16the answers you give are very important.
4:19So in every single time that I speak and
4:24ask something of Bruno or make a comment
4:28or anything like that, I want you to
4:29pause this video, think for a while and
4:32come up with the come up with the answer
4:33that you would give the partner, write
4:35it down, perhaps say it out loud, but
4:38put it out to the world so that you can
4:40actually compare what your real answer
4:42is with Bruno's and then improve it so
4:44that you can get better and get to the
4:46level that you're going to be able to
4:48solve every case with the partner.
4:51Right? All right, Bruno, ready? Yeah,
4:53I'm ready.
4:54Okay, let's go then.
4:56So the question here is the following.
4:59So BMW
5:01is thinking of bringing in-house
5:04their call center
5:06and they want your help estimating what
5:09will be their total cost for their world
5:11worldwide call center. Okay. So just a
5:16couple clarifying questions. So right
5:18now they have an outsourced call center
5:21if I understand. And they want to bring
5:23it in-house. Any particular reasons for
5:26that?
5:27Well, they're a premium brand and
5:30they're not particularly happy with the
5:33service level that the actually not
5:36necessarily service level, but they
5:38can't even really measure the service
5:39level that they're getting and they want
5:41to make sure that they're delivering the
5:42best
5:43service for their clients. Okay, that
5:45makes sense.
5:47Also you mentioned that they will have a
5:49worldwide call center, but I understand
5:52that they will need to have several
5:55facilities because of different
5:56languages.
5:58Is that right?
6:00Yes, they might need to have more than
6:02one. Okay, cuz
6:04it's hard to put people
6:06with all language backgrounds in just
6:08one place.
6:08>> Yeah, you can't get an entire building
6:10in Germany or India or somewhere with
6:12people speaking every single language in
6:14the world. Okay, that makes sense. And
6:16the final question that I have is what
6:18types of calls
6:21do they have in this call center?
6:23Because I think that's a pretty
6:25important part of the
6:27type of service that they'll need to
6:28provide. What do you think?
6:31Okay.
6:32So,
6:34let me quickly draw an issue tree here.
6:38Uh
6:40I understand that they might have calls
6:42for before the sale and for after the
6:46sale. And for both of these groups,
6:51I understand that they will have inbound
6:53calls, so clients or potential clients
6:57calling them, and then outbound calls,
7:00which are calls that they make to
7:03clients or potential clients, right? So,
7:05in the pre-sale inbound bucket,
7:10I imagine that they might have calls
7:12regarding scheduling test drives, uh
7:16questions that potential clients might
7:19have about their cars or their services
7:21when they're considering to buy a BMW.
7:25And I
7:26I don't know if this would go to the
7:28call center, but perhaps media or PR
7:32firms or
7:34newspapers and magazines calling them.
7:37They wouldn't buy a BMW, so it's
7:40technically not pre-sale, but it's
7:42It's not their customers. They're They
7:44might call to inquire them about uh
7:48some
7:49I don't know, some specific car accident
7:52that happened somewhere or a new product
7:55launch. Now, I don't know for all of
7:57these if the call center would be the
8:01people who'd be getting these calls
8:02because for media and PR, for example,
8:05uh I would imagine that uh that would go
8:08to corporate. And then for scheduling
8:11test drives, questions about cars and
8:13services, this might go to the local
8:15dealership. Mhm. Right? Now, but but it
8:19but it could be part of their demand.
8:21Okay? On the outbound bucket, so BMW
8:24calling potential customers,
8:27I imagine there is
8:29there might be sales calls
8:31uh like the ones you get from, you know,
8:34telecom companies or banks. Although, I
8:37don't know if this fits with BMW
8:39positioning. Yeah, you're right. Yeah, I
8:41I'd find it really weird to get a call
8:43from a sales rep at BMW asking me if if
8:47if I want to buy a car.
8:49Uh
8:50it There could be calls uh moving
8:54There could be calls for which the role
8:56is to move the client through the sales
8:59funnel. So, for example, uh let's say
9:02that a client had a test drive and they
9:08and they didn't come back. Mhm. Uh
9:12They might want to call this person and
9:14ask if they're still interested, if they
9:16want to drive the car again, if they
9:18want to try a different model, if there
9:20were any concerns. Okay, that's a
9:22dealership role, but okay.
9:23>> Yeah, I would imagine so.
9:26Uh
9:27Okay, so let's
9:28You can just ignore the pre-sale calls.
9:31Yeah, I was going to mention. Let's move
9:32on to the after-sale. It looks more
9:34interesting.
9:35>> Yeah. Uh
9:37it seems like everything pre-sale is not
9:39on the call center responsibility,
9:42right?
9:42>> Okay. So, on the after sale, again, we
9:44have inbound calls. Uh, so, customers
9:47calling BMW
9:49Mhm. after they buy the car, and
9:51outbound calls, BMW calling their
9:54customers. So, for inbound, I imagine
9:57there there might be maintenance
9:59scheduling.
10:01Uh, so, there's frequent maintenance,
10:04and I want to schedule.
10:07Maybe I moved towns, I don't know where
10:09the dealership is, or something like
10:11that. And then, also questions about the
10:14car and the services. So, for example,
10:17let's say I'm driving my car, and a
10:20weird light lights up on the panel, or
10:23it starts making a strange noise. Mhm.
10:26Uh, I'll check the manual, but then
10:28there's a number there for me to call if
10:31I still have any questions, or if I get
10:33involved in an accident. Uh, I might
10:35want to call my insurance company, but
10:37also,
10:38uh, the car company, so, to order parts,
10:41or to see
10:44where's the best mechanic for that type
10:46of car. Perfect. And then, for outbound
10:48calls, one
10:50that, uh, makes a lot of sense to me is
10:53recalls. Uh-huh. So, if BMW finds a
10:57faulty
10:59part, or an issue with a specific batch
11:02of a model that they they created, they
11:04need to call all of their customers
11:07Uh-huh. uh, to
11:11to schedule this recall.
11:13And then, also research with their
11:15customers. So, to see if their customers
11:18of a particular model are satisfied, or
11:21to see how their experience with the
11:23dealership was uh, after they they've
11:27done a test drive, or
11:29uh, entered the sales funnel. And I
11:31suppose that this one you can't leave to
11:33the dealership to
11:35to do otherwise they'll probably screw
11:37the the results, right?
11:39>> Yeah. So, these are my ideas of what
11:41types of calls they might have.
11:42>> All right, that makes sense.
11:44Uh for now, let's focus on the
11:46after-sales inbound calls. Those are 99%
11:50of the calls. Okay.
11:52Uh makes sense.
11:54Uh although I'd imagine that
11:57there could be recalls very often.
12:00Uh not very often, uh pardon. There
12:02could be a lot of recall calls,
12:05but very sporadically. So, we're not
12:08considering those right now.
12:09>> Yeah, it's true. We're not We're not
12:10considering those.
12:11>> Okay, so I'll focus on inbound, and am I
12:14right to say that this means maintenance
12:18uh issues or scheduling and questions
12:21about the car and the services? Yeah.
12:23Okay, so I'll focus on that. So, here's
12:27what I'm thinking.
12:29Uh one way to estimate the cost would be
12:31to list all of the cost lines of a call
12:35center. So,
12:36uh employees, uh equipment, technology,
12:40then there's the telephone bills, and
12:44and then you have rent and everything,
12:47and estimate all of that. However, I
12:51think
12:52that
12:55I think that
12:58there's a common factor underlying all
13:00of these costs, which is the demand for
13:02the services. Okay. So, the way I'm
13:04thinking of structuring this is to
13:09is to do the num- to estimate first the
13:11number of minutes called,
13:14uh and then multiply that by the dollar
13:17cost per minute. Does that make sense to
13:19you? And then all of the costs that I
13:21mentioned would go within the dollar
13:23cost per minute. All right, I think it
13:25makes sense. Let's see how it goes. So,
13:27I'll I'll use that as the first layer of
13:29my structure. Number of minutes called
13:31times the dollar cost per minute.
13:33>> Okay. And the way I'm thinking of
13:36structuring number of minutes called is
13:40to find out how many BMW cars
13:43and should can I focus just on cars and
13:46just consider motorcycles
13:49and other types of vehicles.
13:51>> Yeah, do that. Okay. So, the number of
13:53BMW cars in circulation globally.
13:57So, not how many they sell per year, but
13:59how many are
14:00uh on the streets right now because they
14:03might call in the first year, they might
14:05call Yeah, perfect.
14:06>> in the fifth year that they own the car.
14:08And then multiply that by the number of
14:10minutes per car per year.
14:13Okay. Uh so, starting with how many BMWs
14:17are there in circulation globally.
14:19I would estimate the number of cars of
14:22any brand in circulation uh in the
14:25world. And then multiply that by the
14:28percentage of those that are BMWs. Okay.
14:31Okay. Mhm. So, number of cars in
14:33circulation is pretty simple. It's
14:37kind of easy to find. Uh all we need to
14:39do is to find the number of households
14:41in the world that have cars. All right.
14:44And then multiply that by the number of
14:46cars per household. Okay. And to find
14:49the number of households with cars
14:51uh I can estimate the number of
14:53households and then estimate the
14:55percentage of those that have cars.
14:58All right.
14:58>> And the main reason for a household not
15:00to have car a car is a lack of income,
15:03of course. Okay. And then to estimate
15:05the number of households, all we need is
15:06the global population and then divide
15:09that by the average household size. So,
15:12say the world has 7 billion people and
15:14there's like 3.5 people per household,
15:17that's 2 billion households just to give
15:19you an example.
15:20>> Mhm. Now, to estimate the percentage of
15:22cars that are BMW,
15:24do we have a value for that? Cuz this
15:26one is a little tricky.
15:28Uh no, we don't.
15:30Okay. Let's assume we don't, I guess. I
15:32guess BMW would have probably Since
15:34we're in this room,
15:36uh we we don't have access to that. So,
15:38the way I would do that is to estimate
15:40the percentage of cars that are premium
15:43cars, luxury brands, Mhm.
15:46uh and multiply that by the percentage
15:48of luxury cars that are BMW.
15:51All right.
15:52Might not be super precise,
15:55but I guess we're going to get closer
15:58than just guessing a random number for a
16:01BMW's, right? No, that makes sense.
16:03Okay. And then for the number of minutes
16:07per car per year of inbound calls to the
16:10call center, Mhm. I would estimate the
16:13number of calls that an average car or
16:16an average car owner
16:18uh calls BMW per year, Yeah. and then
16:21multiply that by the number of minutes
16:23per call. Okay. Okay.
16:25So, that will give us the number of
16:27minutes called or the total demand for
16:31uh calls in this call center. And then
16:33we need to multiply that by the dollar
16:36cost
16:37per minute.
16:38And the way I see it, uh the cost per
16:42minute of calls has two major
16:44components, okay? Mhm. The The first one
16:47is the total cost per employee minute or
16:51per minute of an employee. All right.
16:53And by total cost, I mean uh base the
16:57direct costs of the employee, so their
17:00wages, their benefits, etc.
17:03And also the indirect cost, so
17:05everything that they need to do the
17:06work. Okay. Rent of the facility, the
17:09technology, the
17:12the supervisors.
17:15All right.
17:16So, that's indirect.
17:18So, that's total cost per minute of an
17:20employee. And then the other component
17:22of the dollar cost per minute of a call
17:24is the phone network costs. So, how much
17:29we pay for the phone networks or the
17:31internet networks in case these calls
17:33are
17:34made via IP
17:36for a minute of a call. Okay? Okay. So,
17:39for the first bucket, total cost per
17:41employee minute, the way I would
17:44calculate it is to calculate cost yearly
17:48per employee.
17:50So, again, the direct and indirect
17:52portions of the cost. So, how much does
17:55a full-time employee cost per year? And
17:57then divide that by how many minutes
18:01will one of these people be in calls
18:04with customers?
18:06Which is
18:09which is not exactly how many minutes
18:11they work because a portion of those
18:13minutes they're going to be doing other
18:15tasks. Okay? Okay. All right. And then
18:18the phone network costs, I would break
18:21it down into how much we're charged per
18:24minute of a call,
18:26but then add an addition for the waiting
18:29time in line because I imagine that
18:31these customers are calling a a
18:34toll-free number, Yeah. which means that
18:36we're paying for the calls. Mhm. And
18:40so, let's say that they do a 2-minute
18:42call, Yeah.
18:43>> but they're waiting second for 30
18:45seconds. In our estimation of demands,
18:48that's going to appear as 2 minutes.
18:50Yeah. But then we we also have to pay
18:53for those 30 seconds while they're
18:56waiting for a representative to take
18:59their call. And hopefully, they're not
19:01waiting that long because again, we're a
19:03luxury brand, but
19:07but we need to take that into account.
19:09Does this make sense to you? Does this
19:11sound like a good plan? Yeah, sounds
19:13like a plan. Let's you we can some
19:14numbers in it though. Okay. Okay, so
19:16I'll start taking assumptions and and
19:18doing the calculations to figure out how
19:20much will this call center cost then.
19:22Perfect. So, because we have a huge
19:26estimation tree here, I think I'm going
19:28to start estimating the demand
19:31and then go into the cost per minute
19:34just so it's not overwhelming. Is that
19:36okay with you? Yeah, sure. Okay. So,
19:38starting the number of minutes called,
19:40the first assumption that we need to do
19:42is the population
19:45uh which I'm going to assume 8 billion
19:47people. Uh I know it was 7 billion like
19:515 years ago, so I'm guessing it's
19:53approaching eight. Okay. And uh then the
19:57average household size. This one is
20:00interesting
20:01uh because we're talking about BMWs and
20:05I assume this is bought this product the
20:08cars their cars are bought by wealthier
20:11families that tend to be in rich
20:14countries although there are a ton of
20:16wealthy families in poorer countries as
20:18well.
20:19Uh and the household size in these
20:22families is smaller. Mhm. Uh but it I
20:26will still try to use the global average
20:28because
20:29then we can just adjust for that in the
20:31percentage
20:32of how of cars that are BMWs. All right.
20:36All right. Mhm. Uh so I'm going to use
20:38uh four people per household here. Mhm.
20:41Uh I think it's closer to like 2.5 or
20:44three in wealthier countries and closer
20:46to five in lower income countries. All
20:49right, makes sense.
20:50>> So, four I think makes sense.
20:52Uh then the percentage of those
20:55households that own a car.
20:58What I'm thinking here is there are kind
21:00of two groups of people that don't own
21:03cars. Mhm. Besides those people that
21:05don't like cars which I think globally
21:09it's a minority. Yeah. Uh I think there
21:11are people who are very close to the
21:13poverty line, so either they're under
21:15the poverty line or they've just crossed
21:18it, and these people definitely can't
21:20afford a car. Yeah. And I would guess
21:22that's like 50% of the world's
21:24population. Makes sense. Considering,
21:27you know, a lot of people live in India,
21:30Bangladesh,
21:32uh the countryside of China, then
21:34there's Africa, Yeah. then all the
21:36poorer people in in developing
21:38countries. I guess 50%.
21:40>> All right. And then there's uh
21:42mid-to-low income
21:45uh families
21:47that either they can't afford
21:50uh
21:51or even though they have a nice life,
21:53they still can't afford a car, Mhm. or
21:56they choose to use public transportation
21:59because they need to
22:02uh
22:02really budget and
22:05adjust their budget to have a car and
22:07cut expenses on foods and living
22:09conditions and things like that. Okay.
22:11So, they just choose to use public
22:12transportation, and I'm assuming that's
22:14like 20% of the world's population. All
22:17right. So, you're left with 30%.
22:19>> Okay. of the people in the world who own
22:22a car. Okay. Okay. Then of those people,
22:25how many cars do they have?
22:27A lot of those people have one car,
22:30but now we're talking about wealthier
22:32people, so a lot of them will have two,
22:34three, even four cars. Mhm. So, I was
22:37thinking of just a assuming 1.5 cars per
22:41household. All right. Okay? Yeah, that
22:43makes sense. So, with that, we can
22:45calculate the number of cars in
22:46circulation. All right. I'll just do the
22:48calculations. So, 8 billion people
22:50divided by four people per household,
22:52that's 2 billion households,
22:54>> Mhm.
22:55uh
22:56of which 30% own a car, at least one
22:59car.
23:00Yeah.
23:00>> So, 10% of 2 billion is 200 million * 3
23:03that's 600 million households with at
23:06least one car. Okay. Times 1.5 cars per
23:09household gives us 900 million cars in
23:12in circulation globally. All right.
23:15Okay? Now, we need to find out how many
23:17of those are BMWs.
23:20And for that, I have the percent that
23:22are luxury cars and then the percentage
23:24of luxury cars that are BMWs. And as I
23:27said, this is tricky
23:30uh because it's hard to assume values
23:32for this. It's hard to have a basis,
23:34yeah? But what I was thinking is
23:37uh for the luxury cars, you know,
23:39there's that whole 1% thing of the
23:42population that they're wealthy and they
23:44can afford luxury brands and
23:46>> Yeah.
23:47and I was thinking that this can be a
23:49basis. So,
23:50uh if you look at other markets, for
23:52example, uh purses. Mhm. Most purses
23:56aren't luxury purses or
23:58uh watches. Most watches Well, people
24:01who actually use watches nowadays might
24:04be
24:05this could be a luxury uh watch because
24:08other people are using their smartphones
24:10instead. Yeah. But what I mean here is
24:12in most markets, fashion,
24:15uh purses, and
24:18uh
24:18I don't know.
24:20Uh
24:20home decoration. Yeah. Most people, the
24:23vast majority, don't own luxury brands.
24:25And the same is true for cars. Perfect.
24:27So, I would expect the percentage of
24:30cars that are luxury cars to be lower
24:31than 10%.
24:33>> Mhm.
24:34But also higher than 1% for two reasons.
24:37One is there's the 1% of the population
24:40that, you know, they they there's this
24:42whole 1% movement. Mhm. Uh which is no
24:45basis for anything, but since we have to
24:47assume a value, I'll use that as a
24:50parameter.
24:51But also these people with much more
24:53money, they can afford to have more cars
24:56than 1.5 per household. All right. Yeah,
24:58that makes sense. So, I'm I was thinking
25:00of assuming
25:023% of cars are luxury cars. Okay, that
25:05makes sense. Can we go with that? Yeah.
25:07Okay. And then there's the BMW market
25:10share within luxury cars.
25:12Which is tricky as well. And what I was
25:14thinking here is
25:17to ask you if we have the market share
25:19of BMW because uh
25:22if we have the annual sales uh in
25:25percentage of luxury cars, I I will
25:27assume that that didn't change that much
25:30in the past 10 years.
25:32So, we could equip make the market share
25:35equivalent to the percentage in
25:36circulation. All right, that makes
25:38sense. But, I do not know the BMW market
25:41share. Okay. Okay,
25:44let's try to figure this out then.
25:46Uh so, the way I will think about it is
25:51that
25:54So, there are two types of luxury cars.
25:56There's like affordable luxury
25:59like BMW. You don't have to be like a
26:02mega millionaire to own a BMW. And
26:05there's like
26:06very high-end luxury cars like Bugatti
26:09or, you know, Rolls-Royce. Yeah. Uh
26:13Lamborghinis. Yeah, Ferraris. Ferraris,
26:15yeah. And
26:17I will assume that
26:19the vast majority of luxury cars are
26:22affordable in that sense. Mhm. And then
26:25I can think of a ton of brands like BMW,
26:27Mercedes, Audis. Then there's Volvos,
26:30Land Rovers, Jaguar,
26:33Aston Martin. I don't know how
26:35affordable that is.
26:36>> the point.
26:37So, there's like 20 or 30 brands. So, if
26:40you split that evenly,
26:42you'd have like an average of 3 to 5%
26:45market share. Uh-huh. Uh per brand. And
26:48I think BMW is a more famous
26:52uh brand among these brands.
26:54>> Yeah. So, let's say that they have 10%
26:57of the luxury cars.
26:59>> All right, I think that makes sense.
27:00Okay, so let's use 10% then. So, that
27:03would mean that the percentage of cars
27:05that are BMW is 0.3%.
27:09Okay, cuz it's 3% luxury times 10%
27:13uh, luxury that are BMW. Exactly. So, of
27:16all cars in the circulation in the world
27:19900 million Mhm. 0.3%
27:22are BMWs, which kind of makes sense.
27:24It's like one in 300 cars. Yeah.
27:27Uh, if you looked in in the developed
27:29world, you'd say it's closer to one in
27:3250 perhaps, but then there's all the
27:34rest of the world Yeah. where you barely
27:36see one. Yeah. Uh,
27:38which would gives us a number of BMWs in
27:42circulation of 900 million times 0.3%.
27:47Mhm. So, 1% of 900 million is 9 million?
27:50Yeah. So, 0.1%
27:53is 0.9 million. Yeah. Times three,
27:56that's 2.7 million BMWs
28:00uh, in circulation globally.
28:01>> Okay. Okay, that's all right.
28:03Now, uh, with that to estimate the
28:06demand for the call center in minutes
28:07called, we need to estimate the number
28:09of minutes per car per year. Okay.
28:12So, we have number of calls per car per
28:14year
28:15and we have number of minutes per call.
28:17For the number of calls uh,
28:20I'm thinking my personal
28:24uh, usage Yeah. uh,
28:26I don't have a BMW, but uh, I barely
28:29call my car company. Yeah. I think I've
28:32called them like
28:34twice in
28:3510 years. Okay. Okay?
28:38But, uh,
28:40also I'm
28:42pretty young, uh, so we millennials
28:46don't call people that often. I imagine
28:50I don't drive a lot. I do Oh, okay. So
28:52that that probably changes
28:54your experience versus the average a
28:56bit. Yeah, and then I'm also thinking
28:59that the average BMW owner is older than
29:01me.
29:02And older people like to solve problems
29:04by calling companies. I do all that all
29:07on the internet. Okay. So I'm thinking
29:09maybe half a call per year.
29:12Okay. Because
29:14So one call every two years because
29:16BMW's they don't break
29:18down a lot.
29:21It's not like you don't have to do
29:24maintenance very often.
29:28So
29:29yeah, one is one call every two years
29:32all right for you? That sounds
29:33reasonable. Okay. So let's go with one
29:36call every two years and then there's
29:37the number of minutes per call. And
29:40the way I think about it is
29:43there are two types of calls. There are
29:44very quick calls like hey, I have a
29:47question.
29:49This light in my panel should I worry
29:51about it?
29:53And then they'll just give me an answer.
29:55So it's like
29:57four to five minutes. Okay. And there
29:59are very long calls
30:01when you have a real issue.
30:04Like where they have to troubleshoot and
30:07ask you to do things with the car and
30:10wait for your reaction
30:12to see what happens and then perhaps
30:14transfer you to call to another person.
30:16And there are very few calls in the
30:18middle. Like
30:19they're either very quick or very long.
30:22>> Yeah. Make sense? Yeah, it does. Can I
30:24assume that there's a 50/50 split among
30:28these two types of calls?
30:30Yeah, that sounds about right. Okay.
30:33So if we assume that the very quick
30:35calls are like four minutes long
30:40and the very long calls are like 30
30:42minutes long. Mhm. We would have an
30:45average of minutes per call of 17
30:47minutes.
30:49Because it's 50% of 4 minutes is 2
30:52minutes.
30:53And then 50% of 30 minutes is 15
30:55minutes. And then we add those two up,
30:57we have 17 minutes per call. Okay, so
30:59that's the weighted average. All right.
31:01Yeah, it's the weighted average, which
31:02is the same as a simple average because
31:05it's 50/50.
31:06>> Perfect.
31:06>> Okay? So,
31:08uh with that we have half a call per car
31:11per year, so that will give us 8.5
31:14minutes per car per year. So, 17
31:17uh times half or divided by two. Okay.
31:21Okay?
31:22Uh
31:23which kind of makes sense, 8 minutes per
31:25car.
31:26Yeah. Uh in demand. Yeah, if you own a
31:29car, then uh
31:31you'll spend 8 minutes in your year on
31:33the phone with the car company. That
31:34sounds okay. Yeah, it sounds all right,
31:36right? So, uh to estimate the demand of
31:40minutes per for this call center, we
31:42have 2.7 million BMWs calling an average
31:46of 8.5 minutes per year. Yeah.
31:49So, that would be about 3 million cars
31:53for 10 minutes, so 30 million minutes.
31:56Can I use this number? I I rounded both
32:00uh numbers up by a little bit.
32:03Uh
32:04Uh no, I'd rather you use the precise
32:06number. Just do the do the calculations.
32:08Okay, so that sounds fair. So, I can
32:11multiply 2.7 million times 8.5
32:15or I can adjust that 30 million figure
32:18because 8.5 is 15% less than 10 minutes,
32:22which is what I used.
32:23>> Okay. And uh 2.7 million is 10% less
32:27than 3 million, which is what I used.
32:29So, if I
32:31uh
32:31if I reduce 30 million by 15%
32:35and then the result of that by 10% I'll
32:37get exactly
32:39uh the number of minutes. So, I'll do
32:40the calculation this way, all right?
32:42>> All right. So, 30 million minus 15%
32:45is
32:47uh 15% of 30 million is 4.5 million.
32:51Mhm. So, 30 minus 4.5 is 25.5 million.
32:56Okay. Okay? And now we need to
32:59uh take 10% of 25.5 away.
33:03So, 10% of 25.5
33:06is 2.55 million. Okay, yeah. Okay? So,
33:10uh 25.5 minus 2.55
33:14equals
33:16uh
33:17If you were 2.5, then we'd have 23.
33:20Yeah.
33:21>> And then we have to take 50K away from
33:24it, so it's 22.95
33:27million minutes. Okay, just use the 23.
33:30Okay, so we have a demand for call
33:32center of 23 million
33:34minutes. Now, to calculate the call
33:37center cost, we just need to figure out
33:38the cost per minute. Mhm. So, if you
33:41were like a dollar per minute, it would
33:43be 23 million dollars in call center
33:45costs.
33:47But I I have no idea if it's a dollar or
33:48not. Let's figure that out now.
33:50>> So, as I said, the dollar cost per
33:52minute of a call will be the total cost
33:56per minute of an employee, including
33:59direct and indirect costs.
34:02And it will be the cost of the phone
34:05network per minute. Okay.
34:07>> So, I'll start with the total cost per
34:10employee minute.
34:12To calculate that, as I've mentioned
34:13before, I'm going to find the yearly
34:15cost per full-time employee. Mhm. Uh and
34:19then divide that by the number of
34:20minutes that they're on calls per year.
34:23All right. Right? So, I'll start with
34:26the total cost per employee per year.
34:30Uh so, we have direct and indirect
34:32costs.
34:33Uh, direct costs being wages, benefits,
34:38things that end up going to the
34:39employee.
34:41And then indirect costs are rent,
34:43equipment, uh, supervisors, single
34:45things like that.
34:47I was thinking of using a an assumption
34:49of $10 per hour
34:52uh, for employees on the direct costs
34:55and then another $10 per hour on
34:58indirect costs just to uh,
35:01to put them all all the indirect costs
35:03in a bundle and
35:05and call it like like like that. Now,
35:07the reason for $10 per hour
35:10for the direct costs is that uh, that's
35:13a little bit above minimum wage in most
35:16of Europe and in the US.
35:18And
35:20I imagine that BMW would pay like 12 or
35:24$13 per hour in those countries, but
35:27then you have lower cost countries which
35:30in which labor is much much cheaper,
35:33but you don't want to go too cheap
35:36because you want high quality employees
35:38to to talk to our luxury customers.
35:42Okay. I'm I don't know if that makes
35:44sense. I'm not saying it doesn't, I just
35:46don't know. I think we're going to need
35:47to get a lot more specific than that.
35:49What do you mean by that? Like
35:52Break it down further. What are the
35:53components of each one of those? Um,
35:56okay.
35:57Okay, so
35:59uh, so let's break it down. For the
36:00direct costs, there are wages,
36:04Okay. uh, there are benefits. So,
36:07health insurance, life life insurance,
36:10uh, in some countries you pay for their
36:13lunch, cafeteria,
36:16transportation benefits, things like
36:18that.
36:19>> All right. Uh, and then there are taxes.
36:23So,
36:24>> Okay. So, again, this varies by country
36:27and my guess is that in some countries
36:29you'd have
36:31no taxes from the company's part and
36:35just employee paying taxes. In other
36:38countries you have that
36:40but we need to consider that because we
36:42will have a global pool of call center
36:45operators, right?
36:46>> Okay.
36:47And then on the indirect costs, which I
36:50I had the intention to assume $10 per
36:52hour per employee, too. I guess the main
36:54costs are related to the work tools and
36:57environment Mhm. which is the rent
37:00of the facility or the office where
37:03they'll stay. And then you have tools
37:05like computers and so on and software
37:08which I've I've read once about call
37:12center software that they charge per
37:14position, so per employee in a sense
37:16Mhm. and that it's pretty expensive
37:19because it improves efficiency in the
37:21call center and so on.
37:22Then we have management.
37:24So, the supervisor of the call center
37:26operators because we're considering the
37:29cost per uh
37:32per employee that calls to the customer.
37:34Okay. But then we need a management
37:37structure on top of that and I want to
37:40include those those costs into the hour
37:43of the employee who calls the customer.
37:46All right. And then we have HR related
37:48costs such as hiring, firing, and
37:51training of these employees of these
37:54call center operators. So, that's how I
37:56would break down direct and indirect
37:59costs. All right, now it makes sense.
38:01Okay. Go on, you can place your numbers
38:02in it. Okay, so I need to assume numbers
38:04to each of these.
38:05>> Yeah. Okay, great. So, I'll start with
38:09wages. Mhm. Uh
38:12can I assume a 8-hour per day shift?
38:16Yeah, sure. I imagine that in some
38:18places you'd need more time than that,
38:20but just to make
38:21>> Yeah, you can assume 8 hours per day.
38:22Okay, so I'll assume eight an 8-hour per
38:24day shift and I'll assume that the
38:27employee works 250 days per year.
38:30Usually, it's 252
38:33that people consider considering Monday
38:35through Friday. Okay. Okay.
38:38So, I'll use 250
38:40just 1% difference or less than 1%
38:43difference. Right, that's okay. Uh and I
38:45would assume $10 per hour again.
38:48All right. Okay? Yeah, that makes Okay,
38:50so 8 hours per day times 250 days per
38:53year is 2,000 hours per year
38:56times $10 per hour would be $20,000
39:00in wages per year. Perfect.
39:03>> Okay? Then we have benefits
39:05and my guess would The way I would do it
39:08is to assume a percentage of of the
39:10salary. Uh
39:12so, something like 10% of the wages.
39:15Mhm. Uh if it makes sense to you. Yeah,
39:17that does make sense. That would be
39:18$2,000 per year. And then we have the
39:21taxes, which I really, honestly, don't
39:24know what to assume, but I'd guess most
39:27states would charge a tax for this kind
39:30of employment, something in the lines of
39:3210 to 20% of their wage. Yeah, you can
39:36use 20% as a global global average.
39:38Okay, so that's $4,000 per
39:42per year. So, summing 20 in wages plus
39:45two in benefits plus four, that would
39:47give us $26,000
39:50of cost
39:51uh directly to have an employee, a
39:54full-time employee working in this call
39:56center. And then we have the indirect
39:58costs. Okay. Okay. So,
40:02we have work environment, so we have
40:04rent, software, and tools. Mhm. Do we Do
40:06we have any assumptions for those? I
40:08mean, I could
40:10estimate rent from a generic renting
40:13area in terms of square meters and then
40:15multiply that by how many square meters
40:17are needed for a person to work in this
40:20kind of work, but
40:26I I don't really know the the average
40:28rent for this. I do have some numbers.
40:30So,
40:31uh for rent
40:33uh let's say
40:34it would cost
40:36$50 per square meter per year. Okay. For
40:40renting uh these facilities worldwide.
40:42Okay, on average. Mhm. Okay. So, I would
40:45assume something like three or four
40:47square meters per person. The
40:49call centers are pretty cramped.
40:52So,
40:52>> How did you get to that?
40:54The specific uh space that a person
40:56works
40:57would have like a meter by a meter. As I
40:59said, call centers tend to be cramped.
41:01Yeah. But then you need another square
41:03meter uh as
41:05uh hallways, like corridors for people
41:08to move around. Yeah. And then I just
41:10added an extra square meter for common
41:14areas
41:15that besides hallways.
41:18Okay.
41:19>> areas.
41:21All right. So, you have three square
41:22meters per per person.
41:24>> yeah.
41:24>> Yeah, that makes sense. Okay. So, that
41:26would gives us that would give us
41:29$150
41:30per year on rent per person.
41:32>> Mhm. And then we need software and
41:35tools, of which I think software is the
41:37most expensive one because
41:39computers and telephones can last quite
41:42a long time. Mhm. Uh
41:44do you have any range for software?
41:47Yeah, you can use $2,000 per year for
41:51software and tools together. Okay. Wow,
41:53that's quite a lot. It's 10% of their
41:55wage. That's
41:56>> Yeah, but it is really important for
41:58their For their work, right?
42:00>> Yeah, that makes sense. Okay, so $2,000.
42:03So, for work environment, we have
42:05$2,150.
42:07Yeah. Then we have management.
42:10I would assume again a percentage, so
42:12like 10% of their wage
42:16to spend with management. Does that make
42:18sense?
42:20Uh I'd like you to get more specific in
42:22that as well. Okay. I'm not trusting
42:24that number, do you?
42:25>> high level. Yeah. Okay. Okay, that makes
42:28sense. So,
42:29I would imagine something like
42:32three management
42:34levels in the structure.
42:36So, say for every 10 call center
42:40operators, you'd have a supervisor, so
42:42that's the first level. Mhm. Make sense?
42:44Yeah. Okay.
42:46And then we we'd have for every 10
42:49supervisors, perhaps a manager. Mhm. And
42:52then you'd have like a third level, so a
42:56director. And I'm thinking of this for
42:58each facility, but we can think of it in
43:01the aggregate. Yeah. Because I guess
43:04each facility would have the
43:07same level of management structures, and
43:09they'd be in enough scale for us to
43:12consider it as just one thing. Can we do
43:14that? Yeah, sure. Okay, so what I'm
43:17thinking is to use a multiple of the
43:20salary or the wage or the direct cost of
43:23this manager or the supervisor. Mhm. And
43:26then multiply by the or divide that by
43:30how much uh
43:33by how much how many people they can
43:35take care of. So, for example, if a
43:38first level supervisor can take care of
43:4010 people,
43:41and he earns twice as much as the
43:44average call center operator,
43:46that would add to each call center
43:49operator a cost on on top of the direct
43:53cost, a cost of 20%, so two times
43:56uh times 10% or 1/10. Make sense?
44:00>> Yeah, that makes sense. Okay, so that
44:01that would be for the supervisor would
44:03add 20% on top of the
44:06of the direct cost of a call center
44:09operator.
44:10Then on a second level manager, so the
44:13guy who or guy or girl who manages
44:15supervisors. Can we assume that they
44:17manage something like
44:2010 supervisors? Yeah, sure. Okay.
44:24So that would be 100
44:27field people. Yeah.
44:29>> Right? Okay. And could we assume that
44:32they earn like four times as much as a
44:34call center operator? Something like $40
44:37per hour? Yeah, that makes sense.
44:39>> Seems like a good wage for manager
44:42operation. Yeah, in a call center, yeah.
44:44Yeah. So okay, so that's so that's 4 *
44:48100, so 1%.
44:51So that would add 4% to to their cost.
44:54Yeah.
44:55And then the third level, the director
44:58So I'd imagine that this person
45:02makes like
45:04twice as much as a manager or eight
45:06times as much as a call center operator,
45:09but
45:10he take care of like 1,000 people or
45:13more. Yeah. So that would be like less
45:16than 1%. I'd say it's irrelevant. Like
45:19it would be
45:21Okay, you can ignore that.
45:22>> Okay. So we have 20% add-on cost
45:27from the direct cost and we need to
45:28consider not just the wages, but the
45:31benefits and taxes because these guys
45:32have benefits and taxes as well. Okay.
45:35For the supervisors plus 4% from the
45:37managers, so that's 24%
45:41of the management. So management would
45:43cost per call center operator a an extra
45:4824% of their direct cost.
45:50>> Okay. And Yeah, I was thinking
45:52>> to 25. Yeah, that's what I was thinking
45:54of rounding it up to 25 because of the
45:56director. Yeah. And because it's easier
45:58as well. Mhm. So 25% of 26,000, which is
46:03the direct cost that we found, is 26 /
46:064. Yeah. So, that's uh 26 / 2 is 13,000.
46:11And if we divide it by 2 again, that's
46:13$6,500
46:16of management.
46:18Okay, so about three times as much as
46:19the work environment, which was 2,150.
46:22>> All right, perfect.
46:23>> All right. And then we have HR.
46:26So, we have hiring, firing, and
46:28training. Mhm.
46:30Uh
46:31for hiring and firing, what I thought
46:34was to assume a turnover rate. Yeah. So,
46:38how often do we have to hire and fire
46:41employees so we can have the positions
46:43filled year year-round.
46:45>> Yeah. Uh and then multiply that by a
46:48cost of hiring and a cost of firing in
46:51terms of their wage.
46:54Not of their total cost
46:56uh because
46:57I don't think we need to pay benefits
46:59and taxes and that. But I've heard once
47:02that
47:03uh
47:03in general, to hire someone, that's
47:06about a month's wage or 10% of their
47:09annual salary. Uh that's about how much
47:13uh recruiting firms charge kind of
47:16regardless of uh the level of the
47:18person. Okay, regardless of their level,
47:20they'll charge 10% of the year's salary
47:23or something like that, yeah.
47:24>> Okay. So, I was thinking of using that
47:26figure.
47:27And for the turnover,
47:29I don't know the the value, but I I'd
47:32assume it's pretty high Mhm. because
47:35it's not a
47:36it's not the best job in the world to be
47:38a call center operator, right?
47:39>> You can assume that the average uh
47:41worldwide call center turnover is 1.5.
47:45Okay. So, is that for the whole
47:46industry, or would be for in-house For
47:49the whole industry. Okay. So, because
47:52cuz I was thinking this is a
47:54highly outsourced industry. Mhm. And I
47:57was thinking if you have an in-house
47:59uh
48:00call center operation of a good company
48:03like BMW, then turnover could be lower.
48:07That makes sense. Does it make sense?
48:09>> it to be feasible, yeah. Okay, so can I
48:11use one once a year? Like uh
48:14the average employee would stay with the
48:16job for like 12 months or so. Yeah.
48:18Okay. So
48:20and can I use the firing cost as the
48:22same as the hiring cost? So 10% of the
48:24wage. Yeah, that makes sense. Okay, so
48:26I'll do once a year, you have to pay 20%
48:30of the wage of the employee for hiring
48:32and firing. Okay. So that's $4,000 per
48:36employee. Okay, sounds good. And then we
48:38have training.
48:40Mhm.
48:41Which
48:43I'd assume it's less than 10% of their
48:45wage.
48:47I mean, I'm I'm thinking of a percentage
48:49of the wage because you want to train
48:51employees
48:53because
48:55it's better than not training them and
48:57them not doing their job well. Yeah. Uh
48:59but you'd also not pay like their full
49:02wage every month in training, otherwise
49:04it doesn't make sense. You just hire
49:06someone that's more skilled. Yeah, sure.
49:09So you can assume it would cost $1,000
49:13to train an employee.
49:14In training costs per year.
49:17Yeah, we're assuming that they're
49:18staying a year. Okay,
49:20that's true. And does that assume uh
49:24their time as well or just the cash
49:27cost?
49:27>> No, that's only the cash costs. Okay,
49:29great. So we have $1,000 in training
49:33which is 5% of their wage, makes sense,
49:35plus $4,000 in hiring and firing.
49:38So we have $5,000 HR costs.
49:42Now we can calculate the total cost per
49:44employee per year. So to do the
49:47indirect, we need to sum up management,
49:50HR, and work environment. Okay. Right?
49:53So, HR $5,000 plus 6.5 thousand in
49:57management is $11.5 thousand. Mhm. And
50:02we have
50:04plus the work environment. So, that's
50:0711.5
50:08plus 2,000 that's 13.5
50:11plus .15 that's 13.65
50:15thousand dollars. Mhm. Which is about
50:19half of the direct costs. Yeah. So,
50:24that's that makes sense.
50:26So, if we sum 13.65 plus 26,000 dollars
50:31of the direct cost,
50:33that would be 39.65
50:36thousand dollars.
50:38per employee or if we can round that up
50:41would be about 40,000 dollars per
50:43employee per year. Okay, yeah. You can
50:45use 40,000. Okay. So, total cost per
50:48employee is 40,000 dollars. Now, the
50:50next step to find the total cost
50:55per minute in employees is to find how
50:59many minutes in a year is each employee
51:02in a call. So, we can divide the total
51:05cost per employee by how many minutes
51:07and we'll find the cost per minute then.
51:08Okay? Okay. And to do that, I will break
51:11it down into how many minutes each
51:14employee works per year. Okay.
51:16>> And then multiply that by what
51:18percentage of time when they're at work
51:22that they're in calls because they might
51:24be working on something else rather than
51:27customer calls. All right. Okay. So,
51:29minutes worked per year as I said
51:338 hours per day times 250 days per year,
51:36that's 2,000 hours per year. So, we just
51:39need to multiply 2,000 hours by 60
51:41minutes per hour, which would give us
51:45Uh, 2 * 6 is 12 and then we need to add
51:47a zero. That's 120,000
51:50minutes
51:51per year, right?
51:53No. Actually, in many countries,
51:57uh, there are times there are days when
51:59you're paying your employees even though
52:01they're not going to work.
52:03Mhm. What do you mean by that?
52:06Uh,
52:07so in some countries you pay sick days
52:10for employees.
52:10>> All right, there's sick day Okay, so
52:12there's sick days.
52:13>> But not only.
52:15Not only, right. Uh, you have vacations,
52:18paid time off. Yeah, perfect.
52:20>> Right. What else?
52:22You have maternity leave. Mhm.
52:25You have, uh, paternity leave. Although
52:28that's probably very small because most
52:30countries don't have that.
52:31>> Yeah, it is just more. You can go with
52:33only those three. Okay, so Okay, so we
52:35need to take into account sick days,
52:37paid time off, and maternity leave.
52:38>> Yeah, perfect. And discount that from
52:40the 120,000 minutes. Yeah. Okay. So,
52:43I'll do that in percentages if that's
52:45okay with you. Yeah, that's fine. So So,
52:48it's easier to discount. So, sick days,
52:52I would say
52:54it's pretty high with this type of
52:56employment because they're not they
52:58don't have a a a
53:00a
53:01a prosperous career necessarily. Yeah,
53:03sure. So, uh, let's say a day per month
53:07out of the 20
53:08days that they Does that make sense?
53:10>> Yeah, that sounds about fine. Okay, so
53:115%
53:13but
53:14I'm thinking that not every country will
53:16make you pay them in sick days. Perfect.
53:20Yes. You can assume that half of the
53:22countries are
53:24proportionally half of our call centers
53:27will pay for sick days. Okay, so let's
53:29say 2.5%
53:32Yeah, perfect.
53:33less time because of sick days. Then we
53:36have paid time off. I would assume a
53:38month per year.
53:40I know that in the US it's less than
53:42that.
53:43Perhaps in China, not sure.
53:46But I'd say most countries will just do
53:471 month per year.
53:49>> Yeah, a month is fine. Okay, so that's
53:51about 8% a month out of 12. Mhm. So I'll
53:55use 8% and then maternity leave. This
53:58one's a tricky one. Do you Do you have
53:59any data for this?
54:02Mhm. What kind of data are you thinking
54:04of?
54:05I'm thinking of Well, first we need to
54:07figure out what percentage of our call
54:10center operators are women. Okay. Uh
54:13because if we're just considering
54:14paternity leave, then Yeah, sure.
54:16>> we'll just have this with women.
54:18>> So, you can assume half of our employees
54:20are women. Half of our employees, okay.
54:22And then we'd need to figure out how
54:25what percentage of women will uh get
54:28pregnant
54:30uh per year and then we need to figure
54:32out how much time they're off when they
54:34do get pregnant.
54:35>> Okay, perfect. Uh assume they're off for
54:383 months on average.
54:39>> 3 months on average, okay. And the
54:41percentage of time or the chances that
54:45they get pregnant.
54:47The way I would do it is to consider uh
54:50that a woman a woman is fertile for like
54:5220 years of their life. Okay. Uh
54:57and if they if they have uh
55:01a couple children uh during their life,
55:04that's a child every 10 years, Mhm.
55:07which is like a 10% of chances of
55:10pregnancy every year. Yeah, that makes
55:11sense. But then we may have women that
55:15are older
55:16uh working for for us. Mhm. So, we need
55:19to account for that. Or we can just use
55:2110% if we assume that most our employees
55:24are between 20 and 40 years old.
55:26>> go with 10%. Okay. So, we have 50% women
55:30uh having a 10% chance of getting
55:33pregnant, that's 5% Mhm. times 25% of
55:37the year, uh 3 months is 25% of the year
55:41of
55:42paid time off because of maternity
55:44leave. So, that's uh 5% * 25% is 5% / 4,
55:49which is 1.25%.
55:52Perfect. Okay? So, that gives us
55:56uh the minutes worked is 120,000
55:59minutes Mhm. minus
56:022.5% of sick days plus 8% of paid time
56:07off, that's 10.5%. Okay. Plus 1.25%
56:12of uh maternity leave, that leaves us
56:15with 11.75%.
56:18Yeah. Can I just do 12% just to make the
56:20math easier?
56:21>> Okay, so we have 120,000
56:25of that.
56:26Uh 12% of that should be uh 12 * 12 is
56:30144. So, that's 14.4
56:34Mhm. thousand minutes. Okay. Uh because
56:37we've rounded the percentage up, we can
56:41round this
56:43uh down. So, that would leave us with a
56:46nice 14,000 minutes instead of 14.4.
56:49Yeah, that's fine. Okay. Uh it's
56:52actually closer to the real number. So,
56:54so they'll be working 106,000 minutes
56:58per year.
57:00And then of that, a certain percentage
57:03of that time they're in calls, and I
57:07would think that's close to 80%,
57:10but by now I'm thinking that you want me
57:12to
57:14You are right. I want you to go further
57:16in this. This is I'm not trusting this
57:18number at all. Okay. So, let's say uh
57:20between one call and the other,
57:23uh they have some uh
57:26perhaps a few seconds or a minute
57:29uh with no calls, so nothing from the
57:33customer side calling them. Perfect. Uh
57:37they have lunch break. I don't know if
57:39we pay for that, but
57:41Yeah, I assume we do for Okay.
57:43>> for a short lunch break.
57:44>> Like 20 or 30 minutes?
57:45>> Yeah, 30 minutes. Okay, so 30 minutes
57:47out of an 8-hour shift would be
57:50something like
57:517% of their work day.
57:54>> Yeah, 6% yeah. Then they can go to the
57:57bathroom and for cigarettes,
58:01maybe another 20-30 minutes.
58:04>> Mhm.
58:05Uh
58:06and
58:07well, they have training and perhaps
58:10they have meetings as well.
58:12Performance meetings, for example, and
58:15we need to pay for that time. Yeah.
58:17Right? And I assume this is like 10% of
58:20their time. Yeah, just go with 80%. Go
58:23with your 80%.
58:24>> Okay, sounds good.
58:26You you've done the work. Okay, that
58:28sounds good. Uh
58:30so
58:32106
58:34thousand minutes per year times 80%
58:38So 10% of 106,000 minutes is 10.6
58:43thousand minutes. Okay. So 20% should be
58:46twice that. That would be
58:49uh 21.2 thousand minutes. Okay. So if we
58:53discount 106,000 minutes minus 21.2
58:58thousand uh we will have 100 100,000
59:02minutes minus 20,000 minutes is 80,000
59:05minutes. Mhm. And 6,000 minutes minus
59:071.2 thousand minutes is 4.8.
59:12So that is 84.8
59:15thousand minutes per employee in calls
59:19with customers. All right.
59:20>> And that's assuming that the demand of
59:23calls is stable.
59:26Uh
59:27which
59:29Yeah, which we are going to assume.
59:30Okay, sounds good. So, to figure out the
59:33total cost per employee
59:36uh minute or per minutes of an employee
59:39with customers
59:41that would be $40,000
59:44divided by 84.8
59:47thousand minutes per employee. Okay. Uh
59:50if this were 80,000 minutes per
59:52employee, it would be 40,000 divided by
59:5580,000, so 50 cents per minute.
59:58All right? I'm assuming that you want
1:00:00the precise number. I do. Okay. So,
1:00:03uh one way to do it is
1:00:06So, uh
1:00:084.8
1:00:11thousand minutes is So, 4,000 minutes
1:00:15would be uh
1:00:174,000 minutes would be 5% more than
1:00:2180,000
1:00:22and point eight is
1:00:26uh
1:00:28point eight is 1% of 80,000. Yeah. So,
1:00:324.8 is 6%. Mhm. So, because 84.8
1:00:37thousand is 6% higher than 80 Yeah. then
1:00:42the cost per minute is 6% lower.
1:00:46Very close to the truth.
1:00:49I don't think it's precise, but it's
1:00:51very close. So
1:00:52>> Yeah, this is close enough. I'm fine
1:00:54with this. So, I can get uh 50 cents per
1:00:56minute and take away 6% Yeah. which is 3
1:01:00cents.
1:01:01So, you're left with uh 47 cents per
1:01:05employee minute
1:01:07uh
1:01:08with a customer in line. Perfect.
1:01:11And the final component is the phone
1:01:14network cost per minute of a call. So,
1:01:17we need to pay the phone company either
1:01:19for internet or telephone. I don't know
1:01:22what's the usual way that these call
1:01:24centers work, if it's voice over IP or
1:01:27regular telephone uh calls, but what I
1:01:30do here is to uh find a cost per minute.
1:01:34Yeah. Uh
1:01:36and then there are of course local calls
1:01:39and international calls, depending on
1:01:41how we group these call centers. If we
1:01:44have one for every country, then they're
1:01:46all local, which probably doesn't make
1:01:48sense. Mhm. But then if we have like
1:01:51a split of
1:01:53uh a few call centers globally, then
1:01:56some of these calls will be local and
1:01:57some international. And then I'd add an
1:02:00addition for the waiting time in line.
1:02:02As as I've mentioned uh earlier today,
1:02:07uh
1:02:08the
1:02:10the minute of a call is the minute where
1:02:12the customer is actually talking to the
1:02:14call center rep, but uh they're waiting
1:02:18in line for
1:02:19somewhere between a few seconds and a
1:02:21few minutes,
1:02:22and we need to pay that as well.
1:02:24Perfect. Okay. All right. So, for the
1:02:27cost per minute, uh do we have any any
1:02:30idea of how much it costs? It's really
1:02:32hard for me to to think about it because
1:02:34in real What I would do is to find the
1:02:36retail rate and then discount because
1:02:38we're paying wholesale. Yeah. But I
1:02:42don't know my retail rate for phone
1:02:43calls because I just pay for a plan,
1:02:46right? Yeah, okay. Uh I can give you
1:02:48that. So, assume that local calls are 1
1:02:51cent per minute
1:02:52>> Mhm. and international calls are 3 cents
1:02:55per minute.
1:02:55>> Okay, so 1 cent per minute for local and
1:02:573 cents for international. And do we
1:02:59know the split?
1:03:02What do you think? Between local and
1:03:03international?
1:03:04What do you think? Well, it's a decision
1:03:06that they should do that they should
1:03:08make. Mhm.
1:03:09Uh
1:03:11Okay, so here's how I would think about
1:03:13it.
1:03:14Uh
1:03:17In a way, it would make sense for them
1:03:19to
1:03:21find the lowest cost country
1:03:23that they can have a big call center
1:03:26with massive economies of scale. Mhm.
1:03:28Because this difference between local
1:03:31international calls of 2 cents per
1:03:33minute Yeah. is less than 5% of the cost
1:03:38per employee per minute. Okay, that
1:03:40makes sense. The 47 cents. Yeah,
1:03:43the difference so that's 5% uh of the of
1:03:46that number. Yeah. And
1:03:49say if you find a low cost country and
1:03:51compare that with a German call center
1:03:53operator, for example,
1:03:55or an American call center operator,
1:03:58then the difference is going to be much
1:03:59more than that. Okay, that makes sense.
1:04:01Yeah. But
1:04:03you probably don't want someone from
1:04:05India or a low cost country having a
1:04:09German customer in line
1:04:11because first they don't speak German
1:04:13very well in in most low cost countries.
1:04:16And and second, we're a luxury brand. We
1:04:19need to take these calls
1:04:21uh well. So I would think that they
1:04:25would have not a call center in every
1:04:27country,
1:04:29but
1:04:31more than a few call centers across
1:04:33across the globe. So
1:04:35if I were BMW, I'd probably have a call
1:04:37center in the US with local people uh
1:04:41because of the accents and so on. I'd
1:04:43have one in Germany because they sell a
1:04:45lot of cars there. Then perhaps one in
1:04:48some European country where they can
1:04:50hire
1:04:51French native French speakers, native uh
1:04:55UK English speakers, Spanish, Italians,
1:04:59and so on. Yeah. Maybe have a a call
1:05:01center in Scandinavia with people from
1:05:03different countries there.
1:05:05One in Eastern Europe. Okay, that makes
1:05:07sense. Uh why don't you go with uh a
1:05:1050/50 split between local calls and
1:05:13international calls.
1:05:14>> yeah.
1:05:14>> so so we can take a simple average
1:05:16between 1 cent and 3 cents and have a 2
1:05:19cent cost per minute. Yeah. And then we
1:05:22have to add onto that
1:05:24uh something for our customers waiting
1:05:26in line. So, the way I would do is is to
1:05:29estimate the average waiting time Mhm.
1:05:32and divide that by the average call
1:05:33length, Okay.
1:05:35>> which is uh 17 minutes.
1:05:37>> Mhm. And then we know the how much
1:05:39percent we should add on top of the 2
1:05:41cents cost per minute.
1:05:44All right. Do we know the average
1:05:46waiting time
1:05:48for our call center? What do you think?
1:05:50Okay, that's a good question.
1:05:53Uh
1:05:54The way I see it, it's a decision Okay.
1:05:58>> company.
1:05:59And the reason for that is we can have a
1:06:02call center with more people Mhm. and
1:06:05have a lower waiting time. Yeah. Or a
1:06:08call center with less people and have
1:06:11customers waiting for like 10 minutes.
1:06:13Yeah. But I don't think it makes sense
1:06:15to have uh BMW
1:06:18customers waiting for a long time in
1:06:21line because if they're calling, like no
1:06:24one wants to call their car
1:06:25manufacturer. They've probably had some
1:06:27problem or many of them will have had
1:06:29some problem.
1:06:31And if you don't treat them well right
1:06:33now, they they're never going to buy a
1:06:35BMW again. Mhm. Uh so
1:06:39I'd say
1:06:40anything more than 2 minutes waiting
1:06:42time is too much. Okay.
1:06:45So, what do you think is the number?
1:06:47Uh
1:06:49if you're okay with it, I'd love to use
1:06:511.7 minutes, so I can divide by 17
1:06:54minutes of the average call length and
1:06:56have a nice 10% there. That is fine. We
1:06:59can do that. Okay, great. So, that would
1:07:01give us a
1:07:03phone network cost per minute of call of
1:07:06uh 2.2 cents or 0.022
1:07:10dollars.
1:07:11>> Can you walk me through that?
1:07:12Oh yeah, sure. It's zero it's 2 cents
1:07:14per minute.
1:07:15>> Mhm. Plus 10% which is 0.2 cents.
1:07:19Oh, okay. Because it's 1.7 minutes
1:07:21divided by 17 which is the average
1:07:23length of the call.
1:07:25Yes.
1:07:25>> Okay, okay, I got it.
1:07:26>> So we have a 10% addition on the cost
1:07:30per minute because of the waiting time.
1:07:32>> Okay, perfect.
1:07:32>> So we're assuming that the waiting time
1:07:34is 10% of the call.
1:07:35>> Mhm. So and we're assuming that we're
1:07:38paying the same fee Yeah. to the Telco.
1:07:41Okay, makes sense. Yeah. So we have 2.2
1:07:43cents.
1:07:45Uh which which will give us a total
1:07:47dollar cost per minute of calls of 47
1:07:52cents
1:07:53uh in in total employee costs
1:07:56Mhm. and 2.2 cents in
1:07:59uh phone network costs which is 49.2
1:08:02cents.
1:08:04Or
1:08:05we can round it up to 50 cents.
1:08:09Uh
1:08:12which happens to be the name of the
1:08:13rapper as well.
1:08:14Uh
1:08:15so that's 50 cents per minute of calls.
1:08:19Or if you want to be specific, it's
1:08:2149.2.
1:08:23Yeah, we can go with 50 cents. Okay.
1:08:25Uh so that is interesting
1:08:28because at 8.5 minutes per car per year
1:08:34we have a call center cost of about $4
1:08:37per car. Okay. So it's not too much if
1:08:40you consider uh
1:08:41how much a car costs, right? Yeah. Now
1:08:44let's see the uh total cost of call
1:08:48centers which is the initial question.
1:08:50And to do that, all I have to do is to
1:08:52multiply the 23 million minutes in
1:08:55demand Yeah. of call center operators
1:08:58and multiply that by 50 cents. So we
1:09:01have to divide 23 million by two Yeah.
1:09:03which will give us a 11.5
1:09:07million dollars of annual call center
1:09:10costs for BMW. Okay. What do you think
1:09:13of that?
1:09:15Mhm.
1:09:16That's a good question. So,
1:09:18uh one way to think of this number is
1:09:20how many cars I'd need to sell to get
1:09:22that revenue or that profit back. So,
1:09:24Okay. if we think like uh the the
1:09:26average BMW costs
1:09:29uh $50,000 per car, and I'm assuming
1:09:32that the average BMW is a Series 3, not
1:09:36a Series 7 or Yeah, sure.
1:09:38>> expensive cars.
1:09:39Uh so, that would be about
1:09:43uh just to to do uh round numbers here,
1:09:46a $10 million cost at $50,000 per car
1:09:50would be uh 200 cars sold in revenue.
1:09:54Or if we assume a 30% profit margin,
1:09:57that would be 600 cars sold per year
1:10:02uh to pay for a call center, which
1:10:07which is not a lot. Uh see, we we we had
1:10:10assumed that we have 2.7 million BMWs
1:10:13circulating globally. Yeah. If we assume
1:10:15that a BMW lasts for 10 years, Mhm. that
1:10:18means uh assuming that uh growth is flat
1:10:22and everything, Yeah. that that means
1:10:24270,000
1:10:26cars sold per year. Mhm. Right? So, 600
1:10:29cars out of 270,000,
1:10:32if we round that up to 300,000 cars per
1:10:35year,
1:10:35uh 600 out of 300,000, it means that uh
1:10:40we
1:10:42for every 500 cars that we sell, Mhm.
1:10:45one
1:10:47uh is there to pay the call center if we
1:10:49assume a 30% profit margin.
1:10:52So, it's not a it's not one of the
1:10:54largest costs, for sure. Mhm.
1:10:58It's not so expensive for this company.
1:11:01Okay, and if you were the CEO, how would
1:11:03you think about this cost? What would
1:11:05you think of it? If I were the CEO of
1:11:07BMW
1:11:08>> Yeah. Mhm. Good question.
1:11:12I think there's there are two ways to
1:11:14think about it. The way that
1:11:16people usually think of call centers
1:11:19and mainly telecom companies, those
1:11:22companies that call you all the time is
1:11:23that they think of it as a cost. Uh-huh.
1:11:25That they need to cut, right? Now, if I
1:11:27were the CEO of BMW, let's say I have
1:11:30this $11.5 million
1:11:33per year cost. Mhm. Maybe I could cut 10
1:11:36or 20% of it if I made the call center
1:11:39very efficient and maybe reduce the
1:11:42service levels. Yeah, maybe. So, that's
1:11:44a $1.5 to $3 million saving per year.
1:11:48Mhm.
1:11:49Uh but
1:11:51as I hinted in the waiting time part of
1:11:53the case uh
1:11:56if you have a BMW and you need to call
1:11:58them you don't really want to do that.
1:12:00You probably have a problem.
1:12:02Uh and
1:12:05it's one of those interactions where you
1:12:07might lose the customer forever.
1:12:09Uh-huh. Or if you treat them really,
1:12:11really well Mhm. uh you might
1:12:15get their loyalty and then they'll tell
1:12:17all their wealthy friends about it and
1:12:20they'll buy a new BMW when the time
1:12:22comes.
1:12:24So, if I were the CEO, uh I could think
1:12:27of this as a marketing expense.
1:12:30And hey, tell me more about that. Yeah,
1:12:31I'm thinking of it that way like
1:12:35I wouldn't say it would be a bad
1:12:37business move to
1:12:39even invest more Mhm. to increase the
1:12:42quality of this call center
1:12:45and maybe
1:12:46uh maybe if we
1:12:49let's say we had a $20 million cost per
1:12:52year instead of 11.5, we'd probably have
1:12:54a really really good call center with
1:12:57really good service
1:12:59and
1:13:01and the customers that had to call and
1:13:04they do call once every 2 years as we
1:13:06have as we have assumed, they'd be
1:13:08delighted.
1:13:10And
1:13:12we wouldn't need to grow our sales too
1:13:15much to justify that.
1:13:17All right, sure that makes sense, but
1:13:19can you give me some examples of how
1:13:20they would be delighted? Like you're
1:13:22talking about a $10 million $10 million
1:13:24increase Yeah. In cost? Yeah, it's just
1:13:28an assumption. It's hard to measure.
1:13:30>> I get that, but can you give me some
1:13:31examples? Like what what would this
1:13:34super call center do differently? Mhm,
1:13:37good question.
1:13:39Let me think of a few examples, right?
1:13:41So right now we're paying $10
1:13:44per
1:13:46per hour for this for this person. Okay.
1:13:49So they're not super qualified, mhm,
1:13:52especially in in a developed country.
1:13:54Yeah.
1:13:55If we if we paid them 12 or $13 per hour
1:13:58or even $15 per hour
1:14:01and instead of doing like $1,000 in
1:14:03training, we had like $2,000 in
1:14:05training,
1:14:07perhaps we could have
1:14:09super polite people who know how to
1:14:13treat the customer well and
1:14:16they're basically more polite. So that's
1:14:18one way to think about it. Okay. Another
1:14:20way to think about it is to have I don't
1:14:23know, like let's say that about 10% of
1:14:26the calls
1:14:28had an issue on the road and they need
1:14:30to troubleshoot it.
1:14:32It would be much better if they could be
1:14:35uh
1:14:37on the call with a trained mechanic.
1:14:40Okay. Which could cost like 30 to $40
1:14:43per hour if it's a super skilled
1:14:45mechanic that can troubleshoot a car
1:14:48through the phone. Yeah.
1:14:50Uh but it wouldn't add the cost too much
1:14:53because we're talking about a three-fold
1:14:56increase in salary, but just on 10% of
1:14:59the operators in the call center. So,
1:15:01it's a $3 per hour increase on average.
1:15:05Okay. So, and that would make the
1:15:07experience much better because if your
1:15:09car broke in the middle of the road,
1:15:11you're probably not never going to buy
1:15:13that brand anymore unless
1:15:16you're on a call with a mechanic that
1:15:18can help you out and they can maybe even
1:15:22FaceTime you and see what lights in the
1:15:25panel showed up or Uh-huh. something
1:15:27like that. Okay, that makes sense. Uh
1:15:29another example. We We currently pay
1:15:31$2,000 per year in software and tools.
1:15:34Mhm.
1:15:35Maybe there are pieces of software that
1:15:37we can add that integrate with uh our
1:15:40sales process, for example. Mhm. And
1:15:44that would mean that the call center
1:15:45operator uh when they talk to the
1:15:47customer, they could say things like,
1:15:49"Well,
1:15:50uh hey John, it's nice to
1:15:53to talk to you uh and they know your
1:15:55name because of your phone number."
1:15:58Yeah. And they say, "Well, I see it's
1:16:00your third car with us and really
1:16:03delighted to to
1:16:05to have you on call today. What can I
1:16:06help you with?" And they have some basic
1:16:08information about you. Like, you've been
1:16:11a customer before. What's your name?
1:16:14Where do you live? Yeah, so it can feel
1:16:15like you're talking to someone who knows
1:16:17you. Instead of a random person that's
1:16:20Uh-huh. that doesn't care about you,
1:16:22yeah? So, I I don't really know if we'd
1:16:24need to bump this up to $20 million. I
1:16:27don't know I don't know if it makes
1:16:28sense. Yeah. But I'm thinking that maybe
1:16:32uh we can increase the service level
1:16:34through creative ways and maybe even we
1:16:37as a consulting firm
1:16:39can think of these ways and propose to
1:16:41the client uh how to do that.
1:16:45And
1:16:46the cost easily justifies itself because
1:16:51we're talking about a $50,000 product
1:16:54and currently a $4 per customer annual
1:16:57cost.
1:16:59In an important interaction that they
1:17:01have with us that albeit infrequent
1:17:04infrequent
1:17:05uh when it happens it's an important
1:17:08interaction. Okay.
1:17:09>> All right, that makes sense. All right,
1:17:11guys. So, this is it for this case. Now,
1:17:15I want to go through some key insights
1:17:19that you can learn from this case.
1:17:21They're probably the most important
1:17:23lessons in it.
1:17:25All right, so the first thing is you
1:17:29might have noticed that
1:17:32in this case we went and Bruno went into
1:17:36extreme depth within the estimation.
1:17:39You've probably not even done an
1:17:43estimation with that level of depth of
1:17:46structure so far, right? And the
1:17:49question you might be asking yourself is
1:17:52how do you even know if you're going to
1:17:54have to go to that level level of depth?
1:17:56Because if you don't go to that level of
1:17:59depth and interviewer wants you to go to
1:18:01that level of depth, then well, that's
1:18:04going to be a problem, right? And if you
1:18:06do go into but the interviewer didn't
1:18:08want you to do that, then you've got a
1:18:10problem as well cuz you're going to take
1:18:12forever to do a simple estimation that
1:18:15interviewer wanted to be super simple.
1:18:17Now, the answer to this question is
1:18:20simpler than it sound than you might
1:18:22think.
1:18:24It's
1:18:26you never have to do an estimation this
1:18:30deep without the the partner asking you
1:18:32to do that.
1:18:33But you have to be prepared to go into
1:18:36this level of depth in any estimation,
1:18:39in every single estimation. Because at
1:18:41any point, the partner can ask you to go
1:18:44deeper. So, in this specific case, it is
1:18:48super important that you are efficient
1:18:51because the case is so long. So, if
1:18:53you're not efficient, then the partner
1:18:56wouldn't even have time to ask you to go
1:18:58deeper where he wants you to go deeper
1:19:00the most.
1:19:01But, you could never know that the case
1:19:04would be so long beforehand, right? You
1:19:06can't start the case and then before you
1:19:08start structuring, you ask the partner,
1:19:10"Also, how long is this going to go?
1:19:11Just so I know how efficient I'm going
1:19:13to be."
1:19:14So, you need to be efficient in every
1:19:17estimation that you do.
1:19:19And in this one specific case, the main
1:19:22source of efficiency was the approach
1:19:26that was used.
1:19:28Which brings me to the second key
1:19:30insight. The approach that Bruno used to
1:19:34break down this estimation. You might
1:19:36have noticed that the idea of breaking
1:19:40uh the costs down right away into number
1:19:44of minutes demanded and number of
1:19:46minutes of calls and cost per minute is
1:19:50not the first choice for a lot of
1:19:53people. But, well, it should be. So, the
1:19:56question that I'm going to answer now
1:19:57is, "Why did Bruno break the structure
1:20:00down into minutes and cost per minute
1:20:04than rather than the the actual
1:20:06cost components, which is the most
1:20:08common choice if you are doing
1:20:13uh a cost estimation?" So, let me show
1:20:16you first what this would look like if
1:20:22what this the structure would look like
1:20:24if we were actually doing uh breaking
1:20:27breaking the cost down into the cost
1:20:28components. And then I'm going to show
1:20:30you why it can't translate all into
1:20:33minutes and cost per minute, and then
1:20:35I'm going to explain you why this choice
1:20:38is better than
1:20:41going straight into the cost components.
1:20:43So, if we were to break down
1:20:46the worldwide BMW call center costs into
1:20:49the cost components, this is what we
1:20:51would have.
1:20:52Uh employees, rent, software tools,
1:20:55phone network cost. This is more or less
1:20:57what it would be like. And then, let's
1:20:59say you break down the employees into
1:21:02employees that make the that make calls,
1:21:04management, and hiring and training cuz
1:21:07that's also within employees costs,
1:21:09right?
1:21:10Uh
1:21:12if
1:21:13then we should try to get to
1:21:16uh the number of employees that make
1:21:17calls, or rather the cost of employees
1:21:21that make calls, which should be the
1:21:22number of employees times their
1:21:25um
1:21:26their salaries, right? Uh and benefits,
1:21:29of course. So,
1:21:31to get to that, to get to the number of
1:21:33employees, we would need to find the
1:21:36demand in number of minutes of calls,
1:21:38right? And then we would cross that
1:21:41uh against the
1:21:44supply of minutes per employee. So, how
1:21:48many minutes in a year an employee will
1:21:51talk an employee will talk on the phone
1:21:53with customers.
1:21:54So,
1:21:55the way to get to employees that make
1:21:57calls is by finding the demand in
1:22:00minutes of calls, and then the supply of
1:22:03minutes per employee.
1:22:05And then to find um
1:22:07the management cost,
1:22:10we would need to find the number of
1:22:11employees, cuz the number of managers in
1:22:14every single level depends on the number
1:22:16of employees in the base of the
1:22:20of the front line of the organization,
1:22:22right? So, it depends on the number of
1:22:24employees, which again depends on the
1:22:26demand in minutes, of course.
1:22:29And then the hiring and training costs,
1:22:31that depends directly in the number of
1:22:34on the number of employees, right? Which
1:22:36depends again on the demand in minutes,
1:22:39of course.
1:22:40And then if we're talking about rent,
1:22:43right? What is What does the rent cost
1:22:47of a global
1:22:49uh call center operation depend on? It
1:22:51depends on how many people there are
1:22:53going to be. Cuz that's the only thing
1:22:55that's inside a call center operation
1:22:58plant, right? It's only people. So, it
1:23:01depends on how many people there are
1:23:02going to be, and of course,
1:23:05uh how much space they're going to take,
1:23:07and how much that space is going to
1:23:09cost. But basically, number of employees
1:23:11is
1:23:12uh the most important driver.
1:23:14Then you're going to cross with the
1:23:16others to get to the actual costs. And
1:23:19the number of employees again depends in
1:23:21the number of minutes uh of calls.
1:23:25And then software, same thing, number of
1:23:27employees, cuz not surprisingly, the
1:23:30software companies charge by number of
1:23:33users. All right? And then tools, the
1:23:36way we're going to get to the tools is
1:23:39number of employees, because who's going
1:23:41to use those tools? The people who are
1:23:43making the calls. Otherwise, you don't
1:23:44need the tools, right? And then
1:23:47uh phone network costs, well, finally
1:23:49something that doesn't depend on the
1:23:51number of employees, but still it
1:23:54depends directly on the minutes of
1:23:56calls, cuz that's exactly what you're
1:23:58paying for, right? Uh you're paying for
1:24:00the minutes of calls to the phone
1:24:02network.
1:24:03And if you do all this, well, since
1:24:07everything depends on the demand in
1:24:10minutes of calls, then it's not a driver
1:24:12of every single cost component, that's
1:24:15the driver of the entire cost, right?
1:24:18Instead of saying demand times uh the
1:24:22cost per minute per for employees, the
1:24:26minutes times the cost per minute of for
1:24:30rent,
1:24:31right? Every blah blah blah blah blah.
1:24:34You can just
1:24:35direct directly structure this into
1:24:38uh number of minutes of calls and cost
1:24:41per minute of calls. And then you break
1:24:42down cost per minute of calls into
1:24:45everything else.
1:24:46So, now, why is this better than uh just
1:24:52breaking down to the cost components?
1:24:54Well, for one thing, it allows you to do
1:24:57the estimation quicker. It's going to be
1:24:58a lot quicker to do it this way.
1:25:01Probably around three times faster.
1:25:04Right? It's going to allow you to have a
1:25:07smaller, simpler model, which is easy
1:25:10which is easier to look at, easier to
1:25:14think about, and easier to talk about.
1:25:16So, it's a lot easier to show to your
1:25:18client, but it's also a lot easier to
1:25:23think about what would change. So, let's
1:25:24say uh you're using this model to
1:25:27discuss with your client and with the
1:25:30partner uh a few changes that you could
1:25:32do
1:25:33in the operation of the call center.
1:25:35Maybe you want to think about what it
1:25:38would mean
1:25:40to the total costs if you were to
1:25:44uh
1:25:46change how long the calls would take,
1:25:49the average call would take. Because you
1:25:51want to get better results, so you want
1:25:53to increase the the minutes of each
1:25:56call, the average time of each call by
1:25:5750%. What would this mean to the total
1:26:00costs? You can tell with Bruno's
1:26:02structure, but with this alternative
1:26:04structure that I showed you, you can't
1:26:07really tell uh immediately. It's going
1:26:09to take some work, which is going to
1:26:11make this discussion that could be one
1:26:14meeting last probably two or three
1:26:16weeks.
1:26:18Or maybe if you want to change the
1:26:21the time that employees are off calls.
1:26:23You want to increase that so that you
1:26:25can reduce waiting time of your clients.
1:26:28So,
1:26:29if you think try to think about this
1:26:32with the alternative structure, again,
1:26:33this could take weeks. But if you have a
1:26:36simpler, more elegant structure than the
1:26:39one that Bruno used, then this could be
1:26:41done uh in just a meeting, right? And
1:26:44what this means is that this just makes
1:26:47up a more elegant solution. Uh because
1:26:51it finds the actual drivers and put them
1:26:54in evidence rather than bury them in the
1:26:57end of the tree and behind every single
1:27:01cost component. And this kind of
1:27:04elegance just shows mental clarity.
1:27:08Because the relevant factors are clean,
1:27:11they're clear from the noise, and the
1:27:13less relevant ones like cost per square
1:27:16meter, they're not parallel to the
1:27:18number of minutes. They're relegated to
1:27:20the ends of the branches, which is a lot
1:27:22better. Now, if you estimated exactly
1:27:26the same way, the same value, and the
1:27:29same assumptions as Bruno did, the lack
1:27:31of elegance could bro could probably
1:27:34bother the partner up to the point of
1:27:36rejecting you.
1:27:38Uh it could also just
1:27:41show it could also just take so long
1:27:44that the partner wouldn't have the
1:27:46chance to challenge you in the things
1:27:49that he wanted. And so he would say,
1:27:50"Well, I can't I couldn't test whether I
1:27:52want this guy or not. I can't hire him."
1:27:54And now let's move on to the next key
1:27:58insight. Why did the interviewer want to
1:28:01go into so much depth? Or uh in the
1:28:04words that you're probably thinking, why
1:28:06was the interviewer, in this case me,
1:28:09being such an ass, right? I was
1:28:11challenging everything that Bruno said.
1:28:13I was challenging him to go deeper. And
1:28:16in most times, his numbers were good. He
1:28:19The numbers that he got by going so much
1:28:21deeper were not so much better. So,
1:28:25what did the partner want? So, he really
1:28:28wanted four things.
1:28:30Number one, he wanted to see if you
1:28:33could build a really, really fragmented
1:28:35model cuz he knows that the structure
1:28:38that you created the first time was good
1:28:40enough as a general rule for estimation
1:28:42cases at least. But he wanted to see
1:28:45that if you needed to go deeper, you
1:28:48would be able to do so. So that when
1:28:50you're doing real cases, real projects,
1:28:54if you need, you can build a really deep
1:28:57model.
1:28:58And now, he also wanted to see if you
1:29:00could discuss each tiny item in the in
1:29:06the model. So,
1:29:08he wanted to know that that you would
1:29:10see every single nuance if you needed
1:29:13to. So that if you're doing a real case,
1:29:15again, you wouldn't forget something
1:29:17just because it's so tiny. So that let's
1:29:20say you're doing a project where you
1:29:22need to do this estimation, right? And
1:29:24then you need to do it in a day, and so
1:29:27you do it the simple way.
1:29:29But then suddenly you need more depth,
1:29:31and then you have a week to do it. He
1:29:33wants to know that if you have that
1:29:35week, you're not going to do it in a day
1:29:36again and then send it to them and then
1:29:39they're going to or in that case your
1:29:41manager or partner is going to say this
1:29:43is not good enough. I need deeper, and
1:29:45then they'll throw it back to you, and
1:29:47then you're doing it in a day again and
1:29:48go a little bit deeper, and then you're
1:29:50going to do it again and again and
1:29:51again. He doesn't want that. They want
1:29:53someone who can actively be go in depth,
1:29:57and that's why he's testing that in the
1:29:58interview. Of course, you don't need to
1:30:00be
1:30:01to go in that level of depth at first,
1:30:04but he wants to know that once you know
1:30:06that you need depth, you can do it.
1:30:10Right? And the next thing is they want
1:30:13to be sure that they are avoiding the
1:30:15frameworky candidates.
1:30:17What I mean is the candidates that come
1:30:19with answers ready in their mind. So,
1:30:21let's say you've already seen a case
1:30:24where you need to estimate costs of a
1:30:25call center.
1:30:27Uh you already have the answer, right?
1:30:29But then if you go into that level of
1:30:31depth, it's impossible to use
1:30:33frameworks. It's impossible to use
1:30:36references of cases you did in the past.
1:30:38You're left with only your skills, and
1:30:40either you can create a solution for
1:30:42each tiny problem, or you can't. Or you
1:30:45have just so many uh
1:30:48frameworks and ready answers in your
1:30:50mind that you can do anything, but
1:30:51that's fine as well. Like if you have
1:30:53that many
1:30:54ready answers in your mind that you can
1:30:56use on the spot, well, you're good
1:30:58enough, right? And the fourth thing that
1:31:01the partner or the interviewer in that
1:31:04in that case, well, this is a partner
1:31:05case, so it would be the partner, and uh
1:31:08it's me, the interviewer. I wanted to
1:31:10see is if Bruno would break.
1:31:13I know that it's incredible incredibly
1:31:16frustrating to be challenged like that.
1:31:19And the interviewer knows as well in the
1:31:22actual consulting interview.
1:31:23They know that your ego takes a blow
1:31:26each time, and it feels like the
1:31:27interviewer is saying that you're wrong,
1:31:29that your answer is too shallow, uh in
1:31:31the end that you're not good enough.
1:31:33And a lot of people get afraid by
1:31:35because of that. They get nervous, they
1:31:37make mistakes in the rest of the case.
1:31:39Do not get scared by this. They know
1:31:42what they're doing. They're They know
1:31:44what's going on, and it's nothing
1:31:46personal, and they don't mean that you
1:31:48did something wrong.
1:31:50Now, here's what's going on.
1:31:52You know, in consulting, many, many
1:31:54times the job is frustrating like this.
1:31:57Many, many times the interaction with
1:31:59clients is frustrating like this. And
1:32:01you're going to have to make points to
1:32:03your clients or to some people among
1:32:05your your clients' organizations. And
1:32:08those people are not going to agree with
1:32:09those points. And they're going to come
1:32:11up with every single
1:32:14uh objection and questioning. And you
1:32:18intuitively you will know that it will
1:32:20not make a difference, but you're still
1:32:21going to have to prove it. And the way
1:32:23to prove that if you're a consultant is
1:32:26with logic and with more in-depth
1:32:28breakdown. And if those attacks break
1:32:31you, then projects that happen like that
1:32:35are going to become a huge problem. And
1:32:37they can't afford that. They can't
1:32:38afford you having problems with that
1:32:41kind of project cuz
1:32:43I'm not sure if they're the majority of
1:32:45the problem the projects, but they are a
1:32:48fair sum of them. But I would That's
1:32:51possible that they're the majority of
1:32:52the projects.
1:32:53Consulting is frustrating like this so
1:32:56many times, and this is exactly what
1:33:00they're testing for.
1:33:02All right. Well, guys, this is it for
1:33:04this case. This is it for the insights
1:33:07of this case.
1:33:09Uh I hope it was helpful to you, and
1:33:11I'll see you in the next video.

About this transcript

This page contains the full transcript of Case Interview Fundamentals: Lesson 8 [Partner Estimation Drill: BMW Call Center] by CraftingCases, generated from the public captions YouTube serves with the video. The transcript has 14,243 words across 2,303 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.