I would always structure such a case completely top down. E.g. in your case, start with the population, then % of people needing glasses, % of those in reach of your optician,...)
The channels only become relevant in this example on the lower levels of the structure (% in reach). As glasses are not a "nice to have" item for most people, the total market will not change depending on the channels used. This might be different for e.g. a restaurant (where a delivery option might create new customers, that would otherwise cook their own food).
And yes, you are right, that you are going very deep into one branch in your example. This is a good thing, as long as you point out that there are many other branches. In a real case setting you likely won't have the time, so you have to prioritize harder. E.g. make the assumption that 60% of all adults (14+) will need glasses and buy 1 pair every 2 years (If you ask the interviewer if this seems to be valid, you will likely get an OK)
For staying more structured, draw out the structure of your approach explicitely (this will also help your interviewer to follow along). In this example, a tree structure will work well.
Then be sure to frequently check back on your initial strucuture and you won't get lost in detail.
Hope this helps. Feel free to ask more questions.