That really depends on the product. If the aim of your product is to work in third world countries with poor access to technology and internet then you should build something as basic and all supporting as possible. If you're building something without a social edge and it's about monetising from 'richer' consumers then wherever they are they probably have relatively recent browsers.
My last startup was focused on high value customers and we only built it to support latest chrome, last two versions of i.e. (which at the time was 9/10) and safari 8/9.
This site provides a good set of stats:
If you're going for the high end customer make sure you have good support for safari because mac users may be a minority but they spend the most money :)
PS. You should also consider a mobile / responsive version.
Get data! Let your customers tell you through the data what you need to build. I would start by building the one browser you know for sure you're going to support. Roll it out and review your analytics to see what browsers they are using. If you have a browser that you don't support that is providing a decent % of your traffic you can prioritize that as your next browser to be sure you're compatible with. Much cheaper and quicker to let the data determine your development path than spend too much time on assumptions. At the end of the day it ultimately is just an assumption until you have data from your audience (not competitors or similar products and services).