Alexis C. Madrigal is a Senior Editor of The Atlantic magazine, covering Technology.
He was poking around the analytics for The Atlantic's site and he found somerthing strange--large chunks of traffic were described as coming "direct" to the site--meaning, people were typing the URL in directly or had bookmarked the page.
The problem? The URLs for the pages were 500+ characters long--in other words, there was no way these were people typing in directly.
His theory is that there is a big chunk of Web traffic which analytics packages miss which is social--it's me and you trading links through email or IM. You send me a link through IM, I click on it, and to the analytics software it appears that I've come "direct."
Madrigal calls this phenomenon "dark social," meaning a dark space of traffic we have no visibility into. He recounts his discovery here, in his article "Dark Social: We Have The Whole History of The Web Wrong," and it's well worth reading.
Analytics packages get attributions from the large players like Facebook and Twitter, but that's clicks directly from those sites. What if I click on the link in Facebook, go to that story in the Atantic, then copy that URL and send it to you via IM and you click on it. Your click looks like you came "direct social" or just "direct," or even "unknown."
Dark social doesn't exist in all analytics but it does in enough major sites that the social attribution of traffic could be more than 50% of all visitors coming through social networks.
This points out a vast new problem area in social networks that will be hard to crack: how can we as social marketers understand the hand to hand passing of links through programs like email, IM and mobile when we don't have visibility to those social transactions?
Some of the big data start ups are experimenting with sucking in unstructured social data and crunching it through natural language processing and other big algorithym science. But if my email to you if private and can't be captured, that's still a type of social connection into which no one has visibility.
There's no easy answers here in the problem that Madrigal has pointed out. But it will be interesting to hear from the analytics community in the coming weeks about dark social.
Who can shine a light into dark social?