Data science is a stupid buzzword. The ideal candidate knows enough about IT to massage data, the more the knows about the domain to investigate the better, and for sure some statistics. Most of all always do sanity checks .. does it make sense? Can it be? Is the data correct?
It is an art. Like writing awesome code, etc. practice, practice, and working with experienced people is key.
The smarter players measure and target accordingly. The rise of RTB and 3rd party companies or your own JS solutions let you cherry pick what you want, or you simply bid much lower (if at all) on traffic you consider not valuable.
This is mostly a problem for unsophisticated players and exaggerated hype originating from players that support tracking view ability, etc.
As a statistican doing machine learning I think there is much more to it; You can do statistics with models. There is progress in statistics, like in all other sciences.
A catchy title, not explaining well what is really going on.
Very difficult; I would look at open source projects somehow related to what you need. There are developers that are 10-times as productive as others, or 100-times. These you need to get motivated; Sharing part of your company is tough if they do not hang around for the long term, so if you give them options, have it mature over a few years.
Go for cheap, newbies hardly ever pays off. Imho domain knowledge matters, too.
Iceland is a country worth investigating when it comes to hosting.