It came when I happened to receive interview invitations at two financial service companies in Vietnam, whose main business is about consumer lending. They were looking for an IT Business Analyst position, and I was one of the candidates. Both interviews went well, sort of. I could answer most of the questions, some of them are actually tricky ones. What amazed me is that there was a type of question that both interviewers asked me, which is essentially about Guesstimating.

What is that?

Guesstimating = Guess + Estimating

Guesstimating questions are also known as sizing questions. Some people prefer calling them Fermi question. Either way, they are tough. They can be ambiguous, complicated, and you have to deliver a reasonable answer with minimal data. Theoretically, you can come up with a ‘good-enough’ value, but still, it needs to be reasonable. Trust me, even Google could not help you completely in this case. The following are examples of sizing questions:

• “How many piano tuners are there in the USA?”
• “How many baseballs would fit into a Boeing 747?”

How to solve?

The most important thing with this type of question is not about giving the final number with sheer accuracy. Even the interviewer may not have the exact number in mind. It is rather about your approach on how to solve the problem, as the interviewer wants to assess your logic and thinking process, whether you know breaking the problem into smaller pieces or not. That is how we, human being, usually deal with problems.
Pratically, you should go through these steps:

1. Clarify the question
2. Break down the problem into smaller pieces
3. Solve each piece
4. Consolidate!

Example

Here is what I was challenged during the interview:

How many hair salons are there in Ho Chi Minh city, Vietnam?

As it was the first time dealing with this type of question, heck I had no idea where should I start. I had thought of either giving a rough number, which would be a dumb thing to do, or simply saying ‘I don’t know’, which would be even worse. Both answers could not WOW the interviewer, as far as I believe. Taking a short break in mind, I decided to go with this approach:

1. There could be around 20 hair salons in my area, which is District 3, Ward 4
2. Then I assume there will be likely 10 wards in each district. I don’t know exactly the number, but that’s my assumed estimation and it’s okay to give assumption like that with this type of question.
3. HCM city has 19 districts, including: D1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, Tan Binh, Binh Thanh, Go Vap, Binh Tan, Phu Nhuan, Thu Duc, Tan Phu
4. Knowing all of these values, we can then conclude the total number of hair salons in Ho Chi Minh with a simple math: 20 * 10 * 19 = 3,8000

Below is the screenshot of a model that I created on Guesstimate.com, a tool that helps us to calculate and visualize the estimation process easier:

I’d rather say this is an easy question, in relative to others. As I’m already familiar with my living area, so I can give an estimated number of Hair Salons there with confidence. This is a Bottom-Up approach. However, this number is still subjective, as the interviewer could not double check that information. As a matter of fact, question that leads to ‘confirmation bias’ situation like that can be tuned to a better version, for example:

• How many hair salons are there in the USA?

Because neither I have any information about the number of Hair Salon in a specific local area in the US, nor do I know how many districts there could be, so I have to take another way: the Top-Down approach, to start with general information that everybody can know. Here’s my reasoning:

Number of hair salon = Number of barbers in the US / Number of barbers who work in each hair salon

1. So to begin with, I will come up with the US population: 318 millions. For ease of calculation, I round it to 320 millions. Let’s call this value A.
2. Then I need to calculate the number of people who are in the working age range (from 14 to 65). I recall the number from reading a statistical newspaper last week: 67% of US population. Again, I round it to 70% to calculate easier. Let’s call this value B.
3. I also need to take the US’s Employment Rate into account. It is around 95% (obviously, you should be aware of this number if you care about economic). Let’s call this value C.
4. Then I assume, out of 100, there would be on average 5 people whose profession is barber, therefore 5%. Let’s call this value D.
5. Again, I assume there would be likely 10 barbers on average work in the same Hair Salon. Let’s call this value E.
6. Thus, the number of barbers in the US = A * B * C * D = 10,640,000. Let’s call this value F.
7. So, according to the formula given above, the number of hair salons in the USA will be (F / E) = 1,064,000. So let’s round it to 1,000,000, which is our final result.

Again, you can have a more detail view of the corresponding model that I created on Guesstimate.com

Books about guesstimating

I strongly recommend these ones:

• Cracking the PM interview  — Gayle Laakmann McDowell, Jackie Bavaro
• Superforecasting: The Art and Science of Prediction — Philip E. Tetlock