I recently set up blog as a support site for students I teach. It's not password protected or truly private, but I don't advertise it's existence and it isn't indexed by search engines, so the only way students arrive at the site is if I send them there, which I do via dark social channels (VLE, email, etc). This means that the server logs tell an interesting story about the technology our students are using to access information.
% arrive via Windows, many of these from university computers in open access labs. The preponderance of Windows doesn't surprise me as this is "work" and we know that students often prefer not to pollute their personal devices with such content. 10% arrive on Macintoshes, although if you add in iOS that climbs to 17%. 12% use mobile devices to access the content. (I suspect the iPad is me as I haven't seen any evidence of our students using them.) And yes, they really, really do like Chrome, confirming my informal observations.
There's something else I can do with this - use the server stats to probe the Dark Social referrers. In spite of apparent passivity on Google+, 41% of incoming traffic actually comes from there (very high lurker rate - they observe but will not contribute). The rest comes from my mailshots: 35% from GMail, 21% from Yahoo and 3% from Hotmail.
I like this data. In spite of the fact I hate myself for resorting to DarkSocial, this is just another way it is working for me currently.