How do you solve market sizing cases with the appropriate level of detail?
My advice is you set the assumptions (define them explicitly) and generally you get about five minutes, up to ten minutes max to do this.
Further segmentation leads to greater accuracy, but sometimes you don't need greater accuracy -- in which case, it is an inefficient use of time.
If you are trying to get a very rough estimate +/- 25%... is the market big or is it tiny... is it $10 billion or $1 million... then you don't need very precise accuracy, and therefore do not need too many segments.
As a trick, I will often mention several segments, and then assume that two segments offset each other such that the number the two segments combined produce is the average for the market, and not bother splitting out those segments.
If the interviewer says, "Okay, now if I tell you your estimate is wrong... where do you think you went wrong (a.k.a., which assumptions are you least comfortable with)?", make sure you are prepared to answer that.