Please help - Partner round market sizing: number of cars sold last year

Question

Dear Community,

I would like to seek your advice on market sizing. Recently, during an interview with a partner from an MBB firm, I was asked to estimate the number of private cars sold in my country last year. Below are my specific questions:

1. Growth Market Considerations:
I included a “growth market” component in my analysis, considering factors like dropping car prices and increased EV adoption. However, since the question focused on last year’s sales volume, is it relevant to include this section in the analysis?

2. Acceptable Variance:
My estimation was five times lower than the actual number. Is such a significant discrepancy a deal breaker?

3. Asking for Assumptions:
I didn’t ask for specific numbers during the interview, assuming the interviewer wanted me to solve the problem independently. This led to an undervaluation of the “penetration rate.” Would it be beneficial to ask the interviewer if any assumptions seem unreasonable? Do interviewers typically provide guidance when asked?

4. Penetration Rate:
I assumed that 10% of households would be willing to buy a car, which may have been biased due to my experience living in a mega city with effective public transportation. After finding that my estimated figure was only 20% of the real number, I suggested reevaluating this factor due to its high impact. How can I develop a more accurate assumption for this factor?

5. Sanity Check:
I calculated that, on average, 0.3% of the population buys a car each year, translating to 1% of households. It's hard to tell whether this 1% makes sense or not. How would you make a judgement after getting this 1% household buys a car every year result? Also, what would be better approaches to conduct sanity check for this question?

I would greatly appreciate your feedback!

Answer 1

Hi there,

I will try to tackle your questions one by one:

1. I'm not entirely sure what the actuals question from the partner was but based on your guidance, it seems to me that the growth component is not to be taken into account. These numbers are more relevant if the question had concerned the split between types of cars sold. An easier way to go about this question would be to calculate the number of cars in your country (probably based on number of households) and make some kind of educated guess of the lifetime of a car and the renewal rate. Important to understand whether it only concerns new car sales or also used car sales

2. While perfect accuracy isn't expected, a factor of 5 is quite a bit off. But in the end it is more about your approach and rigor than the actual number.

3. Usually, the guesstimates are not supported by assumptions provided by the interviewer. The purpose of these exercises is whether you can sense check numbers and come up with an approach. It is okay to ask some clarifying questions such as ‘With car sales, do you mean new car sales or both new car and used car sales?’

4. I suggest by finding a way around using complex numbers such as the penetration rate and find numbers you are more comfortable with such as the number of people in your country, the number of households, the frequency at which people (your parents for instance) buy a new car, etc.

5. Try to benchmark this number. Consider average car lifespan (e.g., if cars last 10 years on average, you'd expect about 10% replacement annually). Or try to benchmark against market size - if 0.3% of the population buys a car each year, that amounts to X cars, with an average value of €Y - does that seem like a reasonable market size?

Remember, the interviewer is often more interested in your problem-solving approach and ability to structure complex problems than in the exact final number. Demonstrating logical thinking, market awareness, and the ability to course-correct are crucial skills.

Best,

Oliver

Answer 2

Hi there,

Funny - that's the exact same market sizing question I discuss in my book The 1%: Conquer Your Consulting Case Interview. :-)

My approach

I'd approach it in a simpler way, in a nutshell, for Germany with rounding and simplifications:

1. Population: 80m

2. Household size on average: 2 people per household → 40m households

3. Average cars per household: 0.5 (in cities many have 0, overall some have 2 or 3 → defend these assumptions well) → stock of cars = 20m cars (private ownership)

4. Replacement rate/avg lifespan of a car: 8 years → every year 1/8th of the stock is replenished = 2.5m cars are sold every year

5. Finetune the result, e.g., add cars from businesses, taxis, etc = 10% → 2.75m per year

(The actual number in 2022 was 2.65m)

Regarding your questions

1. If you want, I'd always add the potential growth at the end.

2. It should be within the right ballpark, so 5x off would be a lot, unfortunately.

3. You can always ask for data in market sizing. If they have it, they will share it. If they don't have it, make assumptions and test them with the interviewer.

4. Segment and think about how big each segment is/how much it influences the total. It's good to use your experience as a starting position but this also poses the risk of blindspots.

5. Translate the magnitude into 1 out of 100 - that way it's easier to grasp it and indeed come to the conclusion that it's quite low (depending on the country of course). Other than that, with the right approach the sanity checks already happen during the approach and calculation stage (e.g., is 8 year lifespan realistic, etc.). You could also think about the far-fetched sanity checks like looking at life events that require a car purchase or income levels (is it realistic that people can afford cars, etc.).

Hope this was helpful!

Cheers,

Florian

Answer 3

You have some great advice here. One thing I will say is you need to build up some intuition/knowledge about industries. This will come in handy across the board and especially in questions like this one. Start by spending an hour a day reading relevant business and economics articles from the FT, Wall Street Journal, Bloomberg and The Economist.

How would this have helped? For example, say the US has 150M households and you assume 0.3% buys a car each year, that is less than 500K cars a year but the market is closer to 15M vehicles a year sold annually.

Answer 4

Hi,

I will try to answer one questions at a time.

1. Growth Market Considerations:
I included a “growth market” component in my analysis, considering factors like dropping car prices and increased EV adoption. However, since the question focused on last year’s sales volume, is it relevant to include this section in the analysis?

>> Not required since asking for last year's sales.

2. Acceptable Variance:
My estimation was five times lower than the actual number. Is such a significant discrepancy a deal breaker?

>> Could be. However, if you walk the interviewer through the reasoning behind you assumptions, they will have the opportunity to guide if they want to. Speaking out loud your approach and thinking behind estimates is very important.

3. Asking for Assumptions:
I didn’t ask for specific numbers during the interview, assuming the interviewer wanted me to solve the problem independently. This led to an undervaluation of the “penetration rate.” Would it be beneficial to ask the interviewer if any assumptions seem unreasonable? Do interviewers typically provide guidance when asked?

Again, if you walk your interviewer step by step through your thought process and assumptions, it may be okay to ask if it seems reasonable. But be careful to not ask too many questions - you should have the confidence to back your assumptions. Also, interviewers will only help you succeed if you are able to build a good impression. Asking reasonable succinct questions if okay but asking too many questions will weaken your candidacy.

4. Penetration Rate:
I assumed that 10% of households would be willing to buy a car, which may have been biased due to my experience living in a mega city with effective public transportation. After finding that my estimated figure was only 20% of the real number, I suggested reevaluating this factor due to its high impact. How can I develop a more accurate assumption for this factor?

>> You have to think about what factors impact penetration rate. % of people in urban areas, affordability (purchasing power, parking space access) , alternative options of transport. Additionally to calculate # of cars sold, assuming these are new cars, you have to assume the average numbers of years people keep a new car, what percentage of population buys new cars vs. used cars etc.

Also, be careful about the geographical scope of the question. If it is about the total sales in country, you have to look at both urban and rural population, not just cities.

5. Sanity Check:
I calculated that, on average, 0.3% of the population buys a car each year, translating to 1% of households. It's hard to tell whether this 1% makes sense or not. How would you make a judgement after getting this 1% household buys a car every year result? Also, what would be better approaches to conduct sanity check for this question?

0.3% seems very low if you are looking at a developed country where a bug % of the population can afford cars. Whereas if it is an underdeveloped country, maybe that may be reasonable.

Since the question was to estimate for your country, you should be able to use metrics such as population, # of households, % middle class and rich etc. to do a sanity check.

One way is to multiple the # of cars estimated with the average price of cars in your country to see if the total sales number is too high or low. For example a 10M total car sales for an average sized country. You would have to segment the market by type of cars - luxury, standard, economy in an actual project - but for the purposes of this case, you can verbalize this and ask if it is okay to just assume one segment for simplicity.

Another way to do sanity check is compare with the sales of a country with similar demographic and size and see if it makes sense. If you don't have any idea about the market size of cars in any country, then you cannot use the analog approach but verbalize it.

Hope this helps.

Thanks,
Sohini