The thing is, "access" to data isn't enough because these companies each have their information organized in ways that have evolved with their companies in their own unique ways. One company's data set isn't compatible with another's typically. That's why I have to laugh every time I hear the NSA has server access. All of that data is fairly useless unless it is formatted or translated into a common, usable form for the NSAs big data analysis. So, this info is definitely culled and organized by the companies beforehand, IMO.