Thanks to the material proposed on your website as well as to a lot of practice sessions with my friends (I don’t have a business background), I managed to successfully pass my first round interview with BCG.
One of the feedback elements concerned the lack of creativity in my approach. Could you please comment on how to be creative in a process that I understand as highly structured and formatted?
My post on Intuition and Structured Problem Solving covers many of these points, but I will elaborate a little further. Creativity in a case interview comes from one of three places:
1) Creativity in intuitively figuring out a hypothesis early in a case
I have covered this elsewhere, so I won’t cover it here.
2) Creativity in how to analyze a case
For example: You have a case where you have two options to solve the case, and both options seem like they could work. You can often use a creative analysis to figure out which one is better.
So let’s say you have a situation to increase sales and you know you can do it by raising prices or by increasing sales volume or market share. Mathematically, both could work. And in your business analysis, there is room to raise prices or room in the market to grow market share.
So which one should the client do?
Well, one way is to creatively use an estimation question to size how much of a price increase would be needed to achieve the goal or how much of volume increase would be needed, and figure out which number seems more achievable in the context of the case.
Creatively use estimating and “sizing” to figure out how much impact a particular option may have, and then incorporate the answer into the case.
3) The final way that comes to mind in being creative is in how to get data.
For example, I had a case from Bain where the year was supposed to be 1980, the client was Motorola, and they want to know if there is going to be a market for the cell phone. So you can’t use any personal experience, you have to use the data available in 1980 to figure out and prove whether a market could exist for cell phones over the next 50 years.
So at one point in the case, the interviewer said, “Let’s fast forward three years. The company decides to proceed with this mobile phone technology, and they have sales data now for three years. The results are not very impressive. They are thinking of killing the project due to slow sales. How would you prove that this was a bad idea?”
So fundamentally this is a “how do you get data/when data is hard to get” type question. By the way, after you get a job offer, it is this aspect of consulting where you spend an awful lot of your time — how to get data when the most obvious data does not exist.
So in my head, I was thinking, “So we have three years of sales data. The client thinks sales are lousy because it is lower than what they were hoping for… but how do we objectively know if sales were good or bad? I mean, I don’t have anything to compare against.
“I can compare it against the last three years, so % growth seems reasonable. But the sales are nowhere near the volume needed to make the massive investment needed worthwhile. So geez, is this just a perception that sales are poor or are sales genuinely poor? How do I tell?
“I need something to compare it against to objectively determine this. Well, what data do I have access to that can tell me whether the third year of sales for a new technology called a cell phone is objectively good or bad?”
The “creative” solution I considered was looking at unit sales for cell phones in the first three years vs. the unit sales of other new technologies in their first three years — radio, television sets, fax machine, microwave, etc.
I could compare unit sales in the first few years of these new technologies, or I could compare % of total U.S. population adopting the new technology in the first three years.
So it turns out this was exactly right, and based on the data the interviewer provided, cell phone sales were actually quite good. Out of ten technologies, cell phone sales in the first three years (expressed as a % of the U.S. population) exceeded seven out of ten other technologies that ended up being very successful over time.
This adoption curve data then allowed me to figure out in which year cell phone sales would reach sufficient volume to repay the massive investment required to bring it to market.
From that, I was able to figure out likely manufacturing costs relative to unit shipments (using the similar unit cost vs. unit volume ratios for other technologies), and forecast likely consumer price points, given those unit costs.
Actually, the person who gave me that case in my final round at Bain (which I did pass, incidentally) was Theresia GouwRanzetta, who is today one of the 50 most powerful women in Silicon Valley.
So that’s probably the most creative example I can think of. Analytically, I knew I needed something to compare the data against. (See my case interview frameworks handout and see the tips on the last page where I emphasize taking apiece of data and comparing it to something… always.) But nothing directly related was available to compare against… so I went with something that was indirectly related, which was a point of creativity on my part… and as I was scrambling for something that was close to or approximated the real data I wanted, I came up with the adoption curves of other consumer technologies as a proxy.