Wednesday, January 21, 2009

How not to do a consulting case study

I'm trying not to pick on Kevin Gao of Management Consulted.  He was just an analyst, and seems like a well-meaning kid.  He's trying to help people break into management consulting, even though it's the worst time in the last 20 years to try and get into consulting.  So I applaud him for his entrepreneurism and his teaching attitude.

It's just that everything he's writing is riddled with errors.
Never trust an analyst's analysis.
For example, How many baseballs fit inside a 747?, Kevin says:
 If the question is “How many baseballs would fit into a Boeing 747?”, don’t say “I’d estimate the cubic volume of a 747 to be between 10,000 and 20,000 square feet”. Stick with one number. Ranges provide unnecessary complication and will double the number of calculations.
This is a really stupid fucking answer, for several reasons.

Reasons why this is a really stupid fucking answer

1:  Volume is measured in "cubic feet", not "square feet."

2:  Estimate something based upon base components that you can reasonable defend, not from a wild ass guess.  I would never accept his answer in a case interview from a candidate;  I would interrupt the candidate, and press them on why they think that's a reasonable assumption.  Here's the right way to answer the question:
"Well, I want to estimate the 747's interior volume by approximating it as a cylinder.  I know that the formula for a cylinder's volume is Volume = Pi * radius^2 * length.  

I'm going to assume that the length of a 747 cabin is 300 feet, since I've been in one and it seemed like the length of a football field.  Then, I want to think about the radius.  I remember the seating configuration in coach as 3 seats, 5 seats, and 3 seats, plus 2 aisles.  If I assume that each seat is 2' in width, and the aisles are 2' as well, that gives me 26 feet in diameter, so I'm assuming the radius is 13 feet.

Now, I have Volume = Pi * 13^2 * 300.  To make the math simpler, I'm going to round pi to 3.  Running the numbers... I get 152,000 cubic feet.  I want to round this up a little bit since I rounded pi down, so I'm going to estimate 160,000 cubic feet as the interior volume of a 747."

Every assumption is based upon a logical and discussable fact.

3:  His estimate is flat out wrong by an order of magnitude.

Smart, lazy people latch onto the 80/20 rule, because they think it means they can just make stuff up, and as long as they are directionally right, that's good enough.  I interview analysts, and the ones I hire are the ones who are detail oriented, and get the numbers right.  I ding the ones who speed through things quickly, trying to impress me, only to be wrong.
It's better to be right than to be fast.


Anonymous said...

Consultant Ninja - clearly this post requires a personal response. I know your post is well-intentioned, but unfortunately it is just flat out wrong. Not only does it miss the entire purpose of my post, but places quotes out of context and doesn't indicate even a cursory understanding of what I'm communicating.

I'll respond to each of your points in order:

#1. You're right - I meant "cubic feet" and this is a typo on my part. It's been updated. However, any reader can clearly see this mistake and it won't dramatically impact the lessons they take away

#2. You need to read my posts much more carefully. With step #5 of that post, the point is to avoid numerical ranges. The answer of "between 10,000 and 20,000 square feet" is not the right answer at all. Nowhere in my post do I state that it is.

Honestly, who is dumb enough to be asked how many cubic feet a 747 is and just throw out "10,000 to 20,000 feet"? The mere absurdity of that answer puts my respect for your judgment in question.

#3. I'm assuming it should be clear by now that I wasn't providing a real answer. Nor was I even attempting to do so - it was solely about the need to avoid numerical ranges. If I actually wanted to include the real answer to the cubic volume of a Boeing 747, that would take 30 seconds on a Google Search.

I applaud the fact that you're reading my posts closely. But please, if you'd like to pick apart my analysis, do so only after:

#1. You've read the entire post

#2. You've put some effort into understanding the content before grossly quoting me out of context

Thanks. I respect your writing, and have enjoyed the insight/attention that you put into your posts. But please, don't make yourself look bad by writing a long post deconstructing a problem that doesn't exist.


Study Management said...

Good written article. Thanks for sharing this informative article.

Wednesday, January 21, 2009