Cédric,

an important thing to keep in mind: There is no right answer for market sizing cases (I mean there is obviously the true number, which I doubt anyone knows for certain). But the number you come up with does not really matter as long as it is not obviously way, way off. If you'd say that the 300 mln Americans consume only 3 mln beers a year, then I would question your business sense and your judgment.

So the key to market sizing exercises is how you approach them. There is ONE THING that you need to show: Are you able to formulate a set of assumptions and hypotheses that get you a somewhat plausible answer?

The approach you took sounds pretty plausible:

1. Start with the entire population
2. Take out children
3. Segment into consumer groups (men, women, high / mid / low / zero consumption)
4. Assume a size for each group
5. Assume a level of consumption of six-packs for each group
6. Do the math

One word of warning, though. All these approaches have the tendency to get very complex very quickly - by step three you already had 8 consumer groups to keep track of.

It is ok to take some simplifying assumptions - like saying there is only high or low consumption, but nothing in between. You can usually see from the interviewer's questions or reactions if you're simplifying too much.

Also, make sure you are using round numbers as long as possible. Let's say you assume that 30% of the 300 mln Americans don't drink beer (minors, abstinents, people who don't like it, etc.) - that would get you a beer drinking population of 210 mln. It's totally ok to say in the interview that you continue the calculation with a simplified figure of 200 million. Makes things A LOT easier, as long as you make explicit what you're doing.

Last hint: Write shit down. Even if you can do the math easily in your head. And label the numbers!

Because:

A - helps you keep track of every assumption and interim result
B - easier to walk someone through your thinking
C - easier to spot mistakes
D - buys you some time

So our case might look something like this:

US population: 300 mln
Non-beer drinkers: 30%
Beer drinking population (BDP): 300 mln x (1 - 30%) = 210 mln (simplified: 200 mln)
Split male/female in BDP: 60 / 40
Male BDP: 200 mln x 60% = 120 mln

And so on...

(edited)