158
c h a p t e r6 :SOCIAL ANALYTICS, METRICS, AND MEASUREMENT■Correlation has an application in connecting social media analytics and other
web and business analytics. First, it’s easy to do: Collect your listening results, clean
them to remove noise and irrelevant results, and then attempt to correlate the listening
results volume and sentiment with web traffic. If the patterns match—if the correlation
is high—you’ve got an indication that some deeper connection is at work. This is not
proof of such a connection! But it is an indication that looking further might be worth-
while, that what happens on the Social Web might be impacting what you observe at
your website. The connection in this case is fairly intuitive, to be sure, but consider
how this technique might be used to uncover relationships that are not immediately
obvious,
Try shifting the dates, too: Lag the website traffic by a day or two days or a
week and see if the relative measure of correlation improves. If so, you are seeing an
indication that what happens on the Social Web has a lifetime, or a transit time, across
the Social Web. This is certainly the case with brand advertising, for example, where
a build-up time in awareness is observed. This is very different from, say, TV-based
direct response or impulse buying (think QVC) where running a spot immediately trig-
gers a known (or certainly knowable) buying response. It makes sense that these same
conditions apply between the Social Web and your online purchase points or conver-
sion funnels: It’s worth your time to sort out the relationships.
If correlation is important, causation is the Holy Grail. When you nail down
causation, you’ve got real power as the implication is that you can actually drive a par-
ticular outcome. The test for causation is tougher, and rigorous, systematic A/B testing
should be at the core of your analytics practices because of it. This is as applicable to
your use of the Social Web as it is to your use of any other marketing technology that
can be optimized. By testing and comparing, you’ll separate cause from correlation and
identify the key activities and practices that will drive your business.
Take the time to examine the “usual suspects” in the context of web analytics—
bounce rate, page views, time spent, and unique users—and connect these to your
social media program by selectively changing elements of the social media program
and noting the results. Building on what you learn, add your conversion results to the
mix: Using Google Analytics, for example, set up conversion goals and then compare
the results of your A/B testing around your social programs with your conversions. The
result will be a quantitative understanding of your business, and the way in which your
social programs support your overall business objectives.Business Analytics
The business application of social media is—or should be—driven by its connection
to business. In the previous sections, the links between social media analytics and web
analytics were examined, with the result being a systematic testing process aimed at
finding the relationships (correlation) that drive results and then extracting the key