Metadata, PRISM and You
by Leala Abbott & Jon Gibs, Huge / October 19, 2013
Note: This blog entry is available in English only.
In the wake of disclosures about the N.S.A.’s PRISM program, the media and the world are learning something that information architects and data-driven marketers have known for a while: Metadata and analytics matter. While President Obama was quick to point out that the government hasn’t been listening to or reading the content of phone conversations and emails, the reality is that tracking the metadata associated with those communications alone can be much more revealing. In fact, the wealth of information that can be gleaned from metadata by itself is staggering. This analysis is called graph analysis.
The math behind it is complicated (geek out if you want here or here) but, put simply, graph analysis looks for patterns between points in a virtual space. For the NSA, that appears to mean focusing on a given individual’s network of contacts and its relation to “bad guys.” This is broadly analogous to how Facebook’s marketing team might segment a specific network of users to target appropriate ads.
This type of pattern recognition has been going on for quite some time. In addition, with PRISM, the tech companies that cooperate with the NSA append additional layers of information: so more than just a collection of nameless, faceless people—nodes on a sprawling map—there is a rich network of info. Thus modelers can look at the closeness of relationships between people but also their common media consumption, for example. Though there’s no evidence of this so far, theoretically the government would be able to draw conclusions like, say, people who listen to a certain type of music are more likely to be connected to terrorists.
Of course, these sorts of inferences make up a crucial part of the data-driven online advertising and e-commerce ecosystem. Take web analytics. Let’s say that pages on websites are nodes, and users are information passing between these nodes. To understand how users flow through a website (certainly something businesses want to do to better know their customers), site developers and analysts tag Internet pages for tracking.
This in turn lets us measure basic metrics like conversion and bounce rates and more exotic ones like purchase funnel efficiency. Further, we build solid metadata and taxonomy frameworks within content management systems (CMS), which enables us to uncover even richer information about a visitor’s journey through a website. This sophisticated architecture lets us see user behavioral patterns that would otherwise remain invisible.
Much as the NSA is looking to data owners to give their models more predictive value, we look to our client’s CMS and the metadata architecture we collaboratively develop to provide us with that richer information. Regardless of the serious political, legal and ethical issues involved in the PRISM program—and regardless of whether you shudder at the thought of marketers tracking your every click across a website or welcome the customer service improvements such tracking might bring—the bottom line remains that metadata and analytics matter.