I suspect most approaches (barring methods so time consuming your spending will be out of control before you even start) aren't going to give you something some

I suspect most approaches (barring methods so time consuming your spending will be out of control before you even start) aren't going to give you something something remotely accurate (or useful for your question). Having said that, there might be unexpected side-benefits to research though, like finding out exactly which features people are key to people downloading some competitor programs and neglecting others. It sounds as though you want to estimate how much un-addressed market potential there is (otherwise you would be taking the alternative approach of estimating the resources needed so you make a product that's flat out better or better value for money than your rivals, thus competing for market share in a tapped-out market). But don't count on it or waste too much time - why not figure out what maximum resources you would put into this based on a tiny chance of success, release that first attempt, and improve on it as an when either it starts gaining traction (suggesting real demand and possible success in future if you go through with this) or when new resources open up to you that you don't mind risking on this to take it a little further?