What’s a Hypothesis?

Would you recommend to always to use and mention a hypothesis in a case? Are there some cases which are not suitable for a hypothesis or not suitable to mention?

I ask this because imho sometimes its not that easy to formulate a hypothesis, unless you are dealing with a case like the one below.

e.g. profits declined –> This is a classic problem. I typically start with a broad external + internal focused framework. Then I start externally first, then internally. If no data is indicating industry decline or competition issues, I will move to internal issues and move on to the revenues side first (e.g. change in customers) if no data move to costs (e.g. increase in costs).

H: Given that profits are determined by revenues – costs , I think this problem is caused by the revenues, there are either less consumers buying our product or we were forced to lower the prices due to buyer’s negotiation power

or H: Given that profits are determined by revenues – costs , I think this problem is caused by the costs side, I suspect there has been an increase in the costs

My Response:

In the 60 case interviews that I did, I never mentioned the word “hypothesis” once.  That being said, what I did do was “think out loud” and later realized what I was doing was being “hypothesis driven” without actually using the phrase.

That being said, I think it’s a very good idea to start your hypothesis explicitly.  I tended to do this implicitly and the people who interviewed me could clearly tell I was doing this, so I was never penalized for never saying the phrase, “My hypothesis is…”

In a “profits are declining” case, it’s very easy to create a hypothesis immediately.

If I’m the one being interviewed, I’ll usually say something like, “When profits decline, it’s either a revenue problem or a cost problem. Do we have any additional information as to which area we should analyze first (e.g., revenues or costs)?”

Usually the interview will smile politely and say, “no” (but nice try on getting me to do your case for you).

Then I usually draw out the following:

Profits = Revenues – Costs

“Well, let’s start by looking (arbitrarily) at revenues,” (implicitly I’m saying here that my hypothesis is that it’s a revenue problem, and I’m going to ask for specific data to see if hypothesis is correct).

The more explicit way to say this is, “Well to start things off, I’m going to start with a hypothesis that the decline in profits is due to a decline in revenue.  I’d like to test this hypothesis by seeing if revenues increased, decreased or stayed the same in the past few years.  Do we have any information on this?”

Note: As an interviewer, I would say, both answers are acceptable – though the latter is a bit clearer and preferred.

If the interviewer gives you data by saying, “revenues have increased by 20%”.

I would probably say, “Oh, that’s interesting,” (always a good stall tactic incidentally).

“If profits are declining but sales are increasing, it can’t be revenue problem.” (Implicitly saying: my hypothesis that this is a revenue problem is dead wrong.) “Let’s look at the costs side of the equation. Am I correct in assuming that costs have increased during this same time period?”

Or the explicit version would be:

“Clearly my hypothesis that this decline in profits is driven by a decline in revenues is wrong. I’m going to revise my hypothesis and say that the decline in profits is driven by an increase in cost.  I’d like to test this hypothesis further.  Am I correct in assuming that costs have increased during this same time period?”

Then I would use the rest of the profitability decline framework and continually revise my hypothesis (e.g., variable or unit costs have increased… or fixed costs have increased… or fixed costs from labor has increased due to a new contract with the union, etc… ) until you’ve got the whole problem isolated (which is what the case method is really about… problem isolation).