The question is what is the simplest or most basic product that people want to buy. If you don't merge all of the feature requests, but pick the smallest subset that prospects tell you that they will pay for, you are less likely to get overwhelmed/diluted. Once you know what the minimum feature set is you are more likely to be able to develop it, and if you have pre-sold it there is much less risk than you have developed something that you love that no one wants.

"Frankly, I don't see a point in pre-selling, do you?" If you have ever developed something that no one (or too few) people wanted to pay for you may get more cautious about developing something without assessing customer buying interest first.