Just because you have numbers doesn't mean you have insight

Share |

One of the most seductive things about social media is the way it allows us to quantify things. I have more friends than she does – I must be more popular. That blog post got more hits than this one, so that one's more effective. We have more Twitter followers this month than last month, so we're on the right track.

Numbers are lovely that way. In a world where everything seems open to interpretation, numbers offer certainty. Five is bigger than three: end of argument.

Problem is, a beautiful number can hide an ugly bunch of oversimplification. Trying to quantify the complexities of human interaction in a multidimensional matrix of influence and activity in a few simple numbers is next to impossible (although potentially very attractive to venture capitalists).

Which is why, despite a valiant effort, social-media-analysts-turned-political-prognosticators fell so heavily on their virtual fannies in trying to use online metrics to predict last Tuesday's Iowa Republican caucus.

The good folks at Trilogy Interactive summed up how woefully short those predictions fell in a handy infographic. (Only one prognostication came close - eerily so - until a glitch in the data it was based on got corrected, and then it fell into line with the others.)

So why are retweets, likes, mentions and follows such poor predictors of electoral success? As Trilogy points out, it's partly because of the difficulty of focusing that information geographically. And it's partly the way those numbers confuse conversational buzz and notoriety with support. Micah Sifry puts it well:

Saying simple, stupid things that lots of people want to tell their peers about can get you tons of followers and retweets. But it doesn't mean anything definitive about grass-roots support. Otherwise, right now we'd be talking about Herman Cain's amazing victory in Iowa.

More fundamentally, the information that Twitter, Facebook and other platforms can offer us about our relationships to brands, candidates, ideas and each other is still pretty crude. And it would take a far more subtle, sophisticated and complex reading of the things we say to each other to infer anything very meaningful from those blunt-instrument statistics.

Which is worth remembering the next time you find yourself or your organization getting hung up on the number of followers, fans and subscribers you have. Those numbers can be useful... but they couldn't predict Newt Gingrich's future, and they shouldn't dictate yours.


Niall Cook says

January 7, 2012 - 1:48am

To be completely fair, we never tried to predict the outcome, and we've told the Trilogy Interactive guys that. We did find a strong correlation between our proprietary measure of social effectiveness (more usually used by brands) and one particular set of polls, but that was back in December. And of course, correlation does not equal causation. What is more interesting for us is exactly the point you make about fans and followers being false indicators of success. That's exactly why we created our PRINT methodology to look more at what brands do with their communities, not just how big they are. Niall Cook, co-founder, Sociagility

Heather says

January 7, 2012 - 9:07am

Good article!  And that's why we have qualitative research, as well as quantitative

Thanks for blogging, Rob!

Social Signal on...

RSS feedTwitterFacebookGoogle+

Work Smarter with Evernote

Get more out of Evernote with Alexandra Samuel's great new ebook, the first in the Harvard Business Press Work Smarter with Social Media series!

Available on Amazon, iTunes and HBR.

Join Newsletter

Rob on Twitter