Publishers need not only data processing & collecting
Publishers need a solution that processes data about visitors, content, bids, ad placements, and networks/exchanges to understand what's getting the most revenue. These insights shed light on which content to write, what visitors to attract, how to set up placements, and what ad networks provide the most return. However, most of the information received from data processing isn't actionable.
First of all, certain variables are almost impossible to change. Knowing which browsers get you the most revenue is great, but difficult to incorporate into your monetization strategy. Besides paid advertising with advanced targeting, there are few ways to attract significant traffic from a particular browser to your website.
Secondly, ad auction parameters (browser, OS, country, time of the day, content type, ad placement and many more) influence final CPM in combination rather than separately. And it's easy to get overwhelmed when imagining the number of different possible parameter combinations.
In a nutshell, publishers need not only the data processing and collecting, but also the help of machines to automatically optimize ad auctions.