Well, it's going to depend on some specifics and on how much data do you have (with the implications for the complexity of the model that you can afford), but the most basic approach that comes to my mind doesn't involve any regression at all.
Given your historical data ("I have a list of previous widgets, and how much they sold in each city") you can convert the sales per widget per city into percentages (e.g. widget A sold 27% in New York, 15% in Austin, etc.) and then look at the empirical distribution of these percentages by city.
The next step would be introducing some conditionality -- e.g. checking whether the sales percentage per city depends, for example, on the number of cities where the widget was sold.
Generally speaking, you want to find some structure in your percentages by city, but what kind of structure is there really depends on your particular data.
The problem - at least the one I'm currently focusing on, which might not be the one I need to focus on - is converting percentages-by-city on a collection of subsets, into percentages-by-city in general. I'm currently assuming that there's no structure beyond what I specified, partly because I'm not currently able to take advantage of it if there is.
A toy example, with no randomness, would be - widget A sold 2/3 in city X and 1/3 in city Y. Widget B sold 6/7 in city X and 1/7 in city Z. Widget C sold 3/4 in city Y and 1/4 in city Z. Widget D is to be sold...
If it's worth saying, but not worth its own post (even in Discussion), then it goes here.
Notes for future OT posters:
1. Please add the 'open_thread' tag.
2. Check if there is an active Open Thread before posting a new one. (Immediately before; refresh the list-of-threads page before posting.)
3. Open Threads should be posted in Discussion, and not Main.
4. Open Threads should start on Monday, and end on Sunday.