Social media analysis has, for (better or) worse, caused an avalanche of abridged opinion that now influences how all products are marketed, reputations managed, initiatives directed and opportunities identified. Gleaning public opinion has always been an important aspect of market research but this computational reliance is growing and soon, it seems, no company will know how else to determine public postulation than by allowing machines to collate it all into the easily-digested categories of positive, neutral and negative.
These enumerated thoughts of any online populace now provide what is often an instrumental basis for strategic movements made within any company, which sometimes works great. Other times, however, it collapses – usually crushing a sheepish analyst or two in the fall.
It is fine to numerate your markets, but you must understand them. And in attempting to do that, we now see the global focus upon automated sentiment analysis, natural language processing and aggregated opinion banks – endless gizmos delving into content and deciding the thoughts of posters.
You have seen companies make drastic misjudgements in adapting to social media, with ideas that seem progressive and edgy when conceived resulting in crassness and offence. Given this difficulty found in posting the simplest tweet, it is logical to expect that a massive level of misreading is due when it comes to businesses actually analysing their social media feedback.
We cannot rely upon numbers spat out by data-mining algorithms, for numbers is all they are. Analyse data with programs and you make programs of those who input the data. The thoughts of a community cannot be categorised so simply – people are not binary, as they learned in the election just gone. By filtering out extraneous chatter in a quest for pure data, we risk forgetting that it is customers who make that chatter, and that to filter them is to ignore them.
At SQN, we look at data from many metrics, but know from experience that there are too many variables for any next step to be decided by a computer. If a tweet declares that a vacuum sucks – is that good or bad? The computer doesn’t know – flipping a coin would be more accurate. Our language is variable and tough to elucidate, even by us. The computer certainly cannot deduce each nuance of a human perspective relayed through a textual medium evolving just as fast as the computer itself. Perhaps one day, but not soon. Just ask Google Translate.
At SQN we champion the human element and cultivate chatter, as it is in these conversations that we come to understand a client and engage with their ideals. We do use analytic programs but the resultant data is always analysed by sentient eye. There is no better way to stay connected to the human mindset than by using your own and, even when machines really can pull all this off, I think SQN will stick to tradition. The human element is simply indispensable – and that is exactly how we intend to remain.
This piece was written by SQN’s newest team member, PR and Communications Coordinator Matthew Perryman @mflperryman. He comes to us with a background in writing and publishing and loves everything bicycle-related.