# Get Active in Our Amazing Community of Over 455,000 Peers!

Schedule mock interviews on the Meeting Board, join the latest community discussions in our Consulting Q&A and find like-minded Case Partners to connect and practice with!

# What are the major categories of guess-estimation questions

Guessestimation McKinsey
Recent activity on Jul 09, 2018
3.3 k Views

Hi folks,

I'm trying to understand how to categorize guess-estimation questions. So far I came to below understanding:

1. Market size: calculated by either % of visit x \$/visit,

or calculate from supply restraints.

2. Comparison (e.g. length of road at certain city): mainly pick a proxy I'm familiar with and run the calculation.

I don't think my understanding is thorough, can you please let me know whehter there are any other approach to such questions?

• Date ascending
• Date descending

Hi there!

I always found that the estimation questions that I faced in interviews were intentionally more abstract than the formulaic "How many _____ are bought in ______ each year?" problems you tend to find online. For that reason I think the best practice is to categorise the different techniques and ways of approaching estimation questions rather than categorising the questions themselves.

I would look to approach these questions in a number of ways in my head before deciding on the easiest method (or the one that involves the least blind guesses).

1. Consider Demand: This is often the most simple way to break down an estimation problem involving the quantity of products/services sold in a country. If asked how many men's haircuts are sold each week in the USA, for example, you could assess how frequently you require a haircut and extrapolate this to the population of the USA.
2. Consider Supply: Conversely, you may want to approach this from the hairdresser's perspective. In a typical barbershop you have an average of X people working at any given time, you know that an average haircut takes X minutes etc. Factoring in things like peak working hours, hours worked in a day, days worked in a week... will give you a surprisingly accurate estimate for the number of haircuts sold by a barbershop in a week. This can then be expanded for the whole of the USA by using step 3...
3. Use Proxies: As you mentioned, proxies are very useful to have banked in your head. Ultimately the interviewer wants to see a logical breakdown of your estimation and so personal experience is an excellent tool to have at your disposal. I like to have memorised the populations of major countries/capital cities along with an example town/city that I am very familiar with. Using the same example I would know that in an area I live with 20k people, there are X barbershops. I could then extrapolate this for the whole population of the US.
4. Replacement Concept: Often for estimating the number of products sold in a market annually you can use the life expectancy of that product. E.g. If there is 7.4B people on the planet and average life expectancy = 70 years - we know that (with 0% growth) in 70 years there will be another 7.4B people. i.e. 105M born a year. We can also then factor in growth rates for growing markets.
5. Spacial Awareness: Some of the most abstract (and rare) problems I have faced involved testing a candidate's spacial awareness. E.g. "How many tennis balls can I fit inside a Boeing 747?". I think the best techniques here are to use the pen and paper you are given and take each dimension one at a time. You may find it easier to estimate the distance between rows of seats and, drawing on experience, estimate the number of rows in a plane rather than guessing the length of the plane.

You can see how some of these techniques overlap and draw on each other. Luckily these types of problems can be easily practiced in your own time and remember that a logical thought process is much more important than an exact answer.

Hope this helped!

Hi,

Agree with Jack - it's less about different types of the problems but rather about the tools that you use:

1) First of all, there are 2 ways to structure market sizing:

• Formula - basically a math formula to come up with a solution. The problem with the formula is that it is easy to forget something or get lost.
• Tree - same as with regular cases you build a tree. A very simple example: you need to calculate the number of dogs on manhattan. A number of dogs = share of households having a dog * # of households. # of households = population / average household size. In the end, you'll have a pyramid where you have to fill the numbers on the base of the pyramid. This approach is much easier and help you track all the numbers

2) You should learn the key market sizing techniques:

• Making assumptions based on personal experiences (Use the example of your house where out of 100 apt-s 10 have dogs)
• Adjusting numbers (NY is a busy city thus fewer people have dogs)
• Sanity check - try to apply your calculations to the real environment
• etc.

3) You should learn the key tools:

• Using age even age split (suppose life expectancy is 80 years. Assuming even age split we have 4 mln people in US of each age)
• Using 80/20 split (suppose 20% people earn 80% wealth and the average salary is xx...)
• Using approximations (Length of NY-SF flight and plane speed to calculate US length)
• etc.

4) Learn key numbers: populations, gas price, gas consumption, Boeing speed and nmber of seats, average salary, # of gates in the airport, GDP growth rate, inflation, etc.

5) Practice 10-15 cases and you'll be fine

Feel free to PM for clarifications

Good Luck!