It’s not difficult to understand why marketers want to measure “influence.” At its most basic level, social media marketing is about consumers talking about your brand online — either positive or negatively — and potentially changing the opinion of someone who is somewhere in the long funnel between basic awareness and purchase. The value of that author’s message, as qualitative as it is, is just begging to be measured in some way. Surely a casual mention of your product in a tweet to five followers has less value than a spectacularly ringing endorsement in a blog read by a million people each month? And I agree that it does.
Where we start to run into trouble is when we make the seemingly innocent leap from a list of variables that may lead us to a more complete understanding of the value of that post — things like message intent, audience size, and more — to saying instead that we can combine those variables into an algorithm that can assign a numerical value to the “influence” of the author, and furthermore use that score to base business decisions in the future.
If I inspire you to do nothing else today, check out this interview with one of my heroes on the subject of influence, Duncan Watts. His POV really boils down to one thing — prove to me that author X has been more influential than author Y, and then prove to me that influence can be replicated again, and then you can start to define influence. The simple truth is, you can’t. We have been trying to nail this down for over a decade and still have debates about the value of a ridiculous service like Klout. I actually don’t know any serious marketers who think Klout has real value, unless your goal is to show how easy it is to manipulate services like Klout.
Put simply, it’s only influence if the author has proven to influence someone’s opinion, and can be relied upon to do it again. And if you want to apply a score to it, then that score must also mean something — a score of 60 should mean that author is twice as valuable (can drive twice as many sales and drive twice as many people to change their negative opinion of your brand) as an author who has a score of 30.
So if you want to quantify the difference between that tweet to five people and the huge blog endorsement, what are you to do? For now, keep it simple. Look at the reach of each message (because audience size matters) and the total number of messages (because volume of messages matter) and for the most part assume all other things are equal. They aren’t equal, but with enough scale it doesn’t matter. And the alternative just ends up throwing you down a rabbit hole anyway, debating about the relative value of 100 different qualitative measures of each message. You have better things to do.