Bleep You Data! Why Numbers Suck Unless You Give Them Meaning
In my work, metrics are a big deal. If you don’t measure the success of your social media program, or measure what’s going on before you get started, you can’t see where you’re going, where you have been, or where you are now. It doesn’t matter what you measure – well, actually it does, but different programs, companies and people might measure different things – but it’s important to quantify your assets and results.
That said, I go back to my favorite made-up “wise saying” about metrics: data lies, trends don’t. Ok, so quantify, but you really want to qualify.
How can data lie, you might ask? four is four, one million is one million. But here is where trends, and just as important, context, come in: perhaps those four are the only four you need, perhaps that one million has…four good ones, whether they be site visitors, social media followers…start to get the picture?
Data is meaningless without context and interpretation – without divining trends and meaning.
How does data lie? Take the hackneyed “best time to Tweet” data, helpfully posed by numerous social media nerds. Yes, it is helpful to think about. Did someone say 5pm? It’s always 5pm somewhere (Happy Hour!). What day at 5pm? What kind of Tweets? What do people do with these Tweets? Is it a good time to get ReTweets? Responses? Click-throughs? Do certain topics or types of content get better response than others? The superficial reports are great for discussion but are useless for action. Christopher Penn, never prone to silliness, put the lie to the shallow end of science by posting recently, at 8:42 am, but discovering that people who responded were from all over the world (well past Happy Hour in Australia, yes?)
What about social media growth numbers? First, those can be anything – growth in followers, likes or subscribers, rate of engagement (any actions people take on your content, such as comments or replies). Second, they are dependent. There is no such thing as a straight line. Perhaps a campaign meant you had great numbers in October. Is that great or expected? Does the inevitable letdown mean bad news, or have you set expectations? Does your data settle higher than it was before the spike in numbers? What is your six-month trend line?
There are a lot of question marks in this post, but that is the point: you should always be asking questions – and answering them, rather than letting the number speak for themselves.
Numbers are stupid. And they lie. Give them a voice, and give them meaning. If you are in social media marketing, that’s your job.
Yes, Bleep You Data! (Ok, I don’t swear much here, but this clip has a NSFW word in it – couldn’t resist)
ETA: A Facebook conversation with Matt Ridings showed another side of this concept; he remarked on what he can tell about people “based upon whether you text ‘haha’ or ‘Ha Ha’.” In the course of the conversation, it became clear that factors outside of the conversation – age, dempgraphics, tech-savvy – have a lot of weight in deciding how true that is. You just can’t escape contextual analysis of data (or facts).
ETA2: I omitted one of my favorite examples of not letting data get in the way of the truth. Nate Silver, whose Five Thirty-Eight blog runs in the New York Times, is one of my early influences in this line of thought. His insistence on looking down on individual poll results in favor of aggregating polls to tell the greater, more accurate (but still with reservations) story is a model example.