Weird Interview Questions

Via Tyler Cowen, here are some odd questions with snarky answers.

There are some themes here.  Several are sort algorithms (the horses and the balls) and a number are probability and distribution questions (e.g. the stairs and the stools). Several are clearly sales and customer service situations (e.g. the invisible pen).

And several are estimation problems (e.g. how many airplanes are in the air right now).  The latter type question was very popular when I was at McKinsey & Co.  Many interviews actually gave the victim interviewee some kind of business case.  The point was to see how well the person broke down the problem, considered facts they would need to obtain, etc.

A subset of these was the ever-popular market estimation game, such as "how many home windows are bought each year in Mexico?"  As an interviewer, one wants to see the person think "OK, there is new construction and replacement.  For the new construction market, we need the size of the home construction market, number of windows per home...."  That sort of thing.

We would also generally ask them to guess at numbers for all these and actually come up with a number.  This is not some test of trivia -- being able to look at numbers and reality check them is an important skill, so having a reasonable intuition about the proper scale of business and economic statistics is useful.  In fact, if there was one skill as a consulting manager I was constantly trying to hammer into younger consultants it was to look at the numbers coming out of their spreadsheets and ask them if they really make sense.


  1. Max Lybbert:

    > being able to look at numbers and reality check them is an important skill

    This reminds me of various "we're all gonna die!" posts you've made over the years when a journalist repeats obviously inflated numbers, usually w/r/t Global Warming.

  2. perlhaqr:

    Would it be impolite to answer the "telescope in 1750" question with "Kopernicus published De revolutionibus orbium coelestium in 1543"?

    Or maybe the test is if you're capable of correcting your boss without making him feel ignorant.

  3. eddie:

    The snarkist failed badly in answering the question about counting bottles in a liquor store. "What in the name of God would I be doing counting unsold bottles in a liquor store?" Well, considering that the question was asked by Diageo North America, a division of the world's leading spirits, wine and beer company (or so Google tells me), there's probably a pretty good reason that you might be counting bottles in a liquor store if you worked for them. I imagine they have people who do this all the time, and they probably have sales clerks screaming at them to get out all the time.

  4. Foxfier:

    The ability to ask "is this realistic?" is being greatly eroded-- my rotational peeve being medically, especially in respect to pregnancy. My doctor instructed me to have no more than one, 4oz serving of tuna a MONTH. At some point, I started looking around-- turns out the gov't guideline was 6oz a week of any fish while pregnant. Then I did some more looking around, and found out that they'd taken the lowest dose that ANYONE had suggested might be safe, and divided by ten... at that point I stopped, less I find the original research, somehow, and find out that THEY had halved or similarly reduced the amount where they had possible bad results, and/or that it was based on the assumption that someone was eating only sharks that swam in mercury, or something.

  5. perlhaqr:

    Foxfier: I bet those are some shiny sharks.

  6. Foxfier:

    The better to spot them for jumping.

  7. jeffmeh:

    My favorite was always, "How many gas stations are there in the U.S.?" I have a management consulting background also. One really can learn quite a bit about a candidate with this type of question. While there may not be a single "right" answer, there are certainly "wrong" answers. A snarky retort, a panic, a wildly incorrect assumption (e.g., "Let's see, there are about 200,000 people in the U.S...."), an overly complex or clearly flawed estimation model, an inability to furnish or explain an estimation model, a refusal to give an estimate because the information is incomplete, or a general lack of poise under pressure likely indicate that the candidate is not well-suited to be a management consultant.

    I was also frequently amazed at how close one could get with an estimate if one kept it simple and had only decent assumptions. If the candidate proposed a reasonable estimation model, explained it well, and was within an order of magnitude of the right answer, that was "right" enough. That type of question can lead into a much more interesting discussion than just the litany of past jobs, but it only makes sense to ask based upon the context of the job requirements.