You must have a good clue about both from the beginning.
It's not very successful to find a niche with little competition and alot of searches and build a website of that to then see that there are no profitable ads for this niche!
The other way round works a bit better. Say you first find a niche with ads that pays big time, much like you do with arbitrage, you build a website for that niche, and then you see that there were way too much competition for this niche. Well this works pretty well with arbitrage, but arbitrage isn't all about just having 1 site, it's much about alot of sites! But if you have been into the game for a long time like Jon, you could push a new site in a niche with alot of competition pretty well. But I'm pretty sure that Jon takes much things into the calculation before a new website or project; competition, amount of searches, profitable ads, possible ROI etc.
Hope this answered some question :338: