I have been neglecting my RSS feeds I’m afraid - quite missed this Survey Geek post and the various responses to it.
In a nutshell, Reg is pointing out how ‘performed’ social media discourse is, and Ray Poynter responds by saying, well, so is all discourse.
I’m highlighting this here because a) I’ve been banging on about this for ages myself, and b) none of the replies criticising Reg take into account a very important and obvious point, which is that social media monitoring firms have generally made a huge fuss about how they’re capturing natural data, raw opinion, the ‘authentic’ voice of the consumer etc. Ray’s reply is an excellent one but he’s missing out the background of these claims (or overclaims) which turn Reg’s point from an obvious one into a useful one.
The “authenticity” card has been super-useful for monitoring advocates because it’s a great counter-argument to the “representativity” one. “This stuff isn’t representative.” “Ah, but it’s authentic and unprompted!” And what if it isn’t? Well, you’re left with cheap and fast and sexy, which are GOOD THINGS of course.
But I think this is pretty much why it’s been marketing and PR, not research functions that have been ‘owning’ the monitoring job. They know they’re looking at communications, and communications have an intended audience and a why, and this is as true of the humble tweet as it is of the facebook update or the blog post or the advert or the public speech.
The difference between old research and “listening” tends to be phrased in terms of the difference between research where we control an unnatural context (like a survey or focus group or MROC) and research where we don’t. And put like that you think - if you’re like me - “Yes! Surrendering artificial control! Go for it!”. But how about if we phrase it like this: it’s the difference between research where we understand the context (cos we built it) and research where we don’t.
Suddenly that looks a lot scarier! But that’s the risk, isn’t it? A tweet is an utterance, said by a person to an audience, as the result of a stimulus. So is a comment in a focus group. The difference is that with the focus group we know who the person is, who the audience is, what the stimulus is. And with the former we don’t.
Ah! But we have millions of OTHER Tweets and we can get insight from those. Absolutely, but how much contextual meaning is being stripped out by these processes? In Ray’s comment to Reg’s post, he talks about contextual meaning and how if people trip up when walking, they say “ow!” and this instinctive response carries a huge amount of meaning. And it does! But imagine a “walk monitoring” service, which might say, well we analysed over 10 million walking events and less than 1% contained a talking-to-yourself event. That would give you quite different information from the meaning Ray’s talking about.
Hardly useless information, though. I am - as this post is doing an awful job of showing - a big fan of social media monitoring. The valuable work of crunching social media data and working out its relationship to behaviour is going on all around us (not always led by researchers, mind). It will increase our understanding of people enormously, it will make a lot of the “why” of interaction either apparent or irrelevant. But even the most powerful dataset isn’t necessarily going to capture the same insights that a ‘ground-level’ understanding of the context and texture of online life will. So social media monitoring is great - it just doesn’t always have to mean social media aggregation.
