Tools for Social Media Measurement

I was whining last week on my other blog about the lack of integrated solutions for social media analytics. No sooner had I written that, of course, than up popped several interesting solutions to prove me wrong. I plan to write soon about a couple of specific products, but will use this post to set a framework for evaluation.

I suppose I should start with a definition of “social media”. By this I simply mean any communication method that allow users to interact directly with each other, as opposed to a broadcast medium where only a few people can send messages. I’m not intending to be especially restrictive here – I’d include blogs, public forums, Facebook , Myspace, YouTube , Flickr , Twitter, LinkedIn, Plaxo and many others. These all provide a huge stream of public chatter that marketers can tap into both to monitor what is being said about their products and to proactively spread their preferred messages.

From a measurement perspective, I see several distinct functions. Today, these are largely served by separate point solutions. Integrated systems are beginning to emerge that combine at least a few. The ultimate integrated system would service them all. The functions are:

- tracking mentions. This is the simplest goal; it simply means uncovering and reporting on social media events that relate to your product, brand or company. The fundamental tool here is the keyword search. Many systems do these, and some even combine different social network sources to provide a consolidated report. Google Alerts is probably the best known, although it doesn’t do much with social media aside from blogs. BoardTracker and Linqia are more focused on social communities.

- identifying mentioners. Most social media comments are signed with a user ID of some sort, but the identity of the person behind that ID is often not clear. I haven’t actually seen tools that address this, but they probably exist. What’s needed is to look at whatever public profile is available, use that to find out other information about the person, and then in turn see if you can find that person in other social media. As a not-too-scary example, I recently saw a Twitter post that mentioned a vendor I follow. Checking out the poster's profile to see if she was worth “following”, I saw that she was from a small town where I used to live. Curious, I then found her in Linked In and discovered the company she worked for. Yes, this sounds uncomfortably like stalking, but it’s old news that the Internet is really good for that. What’s interesting here is the potential to help understand background of an individual and her social media profile. The steps that I took could easily be automated; indeed, products like ZoomInfo do something similar, although so far as I can tell they don't include social media other than blogs.

- measuring influence. Influence has two overlapping dimensions: the influence of an individual mentioner, and the influence of a particular event. The mentioner’s influence is related to the profile I just mentioned, but also to blog readership, “friends” and network members in various social platforms, authority as measured by links and recommendations, etc. Again, these statistics are available in a scattered fashion for individual social media, and it wouldn’t be hard to build a system to pull them together once you had linked the user IDs. Surely someone is out there doing this but I haven’t tripped over them. Maybe if I spent more time at the gym?

Measuring the influence of a particular event is actually easier. It is a matter of links, views, downloads, recommendations, ratings, etc. The statistics are often published along with the item itself. One possible tool is TrackUr, a low-cost product (from $18 to $197 per month) that scores Web sites based on “the number of backlinks pointing to a web site, the number of blog discussions, an estimate of traffic, and even the number of times the web site has discussed the phrase in the past.” Another that I suspect costs much more is Radian6, which “tracks comments, viewership, user engagement and other metrics, 24/7, so that you can clearly see the reach and affect[sic] each post has on the community.” It also can “uncover the influencers online by topic, based on user-determined formula weightings.”

- understanding sentiment. This is the domain of semantic analysis (that’s a pun, kind of), which is a long-established field with many players. One specialist applying its technology to real-time Web content is Hapax. Solutions integrated more closely with social media search include Crimson Hexagon and newly-launched Scout Labs Scout Labs is also a low-cost option, with plans starting at $99 per month and currently offering 30-day free trial.

- measuring impact. Ah, the bottom line: what did people exposed to the social media event actually do? Even the Web hasn’t yet reached the stage of universal behavior tracking that would really let you answer this, and I personally hope it never does. But one product that gets close is Tealium Social Media, which builds a list of Web URLs (both social media and regular online media) related to your product, checks which of those your Web site visitors had seen previously, and pops the results into Google Analytics so you can treat the Web events like any other visitor source. (See my earlier blog post on Tealium for details.) At the other end of the process, Vizu lets marketers embed a question in Web ads that asks about the brand attitudes, and compares this against answers of people who didn’t see the ad, thereby measuring the net impact of the ad itself. The vendor has embedded its questions in social media applications from vendors including Lotame (ads in social networks), AdNectar (social ‘gifting’) and Buddy Media (custom social applications). See their press release for details.