Benedict Anderson once said that “communities are to be distinguished, not by
their falsity or genuineness, but by the style in which they are imagined.” His writings on Imagined Communities swam back to me as I read the latest in big data news. SwiftKey – the company behind keyboard app of the same name – have looked at the global data of these doodles and have released this report into which emoji symbols are used most, country by country.
As Anderson famously pointed out, how we choose to communicate and particularly the symbols which we adopt as a group are often the keys to understanding the psychology of a group as a whole, so naturally, I read the report with interest…
Initially, I was excited because I desired specifically to look at Ireland under this new form of ‘big data’ and the lovely people at SwiftKey were kind enough to pass on the raw data and full reports to me. In it there are some very interesting observations, so I highly recommend checking it out in the above link.
Unfortunately however, little old Ireland was not covered off in the research – as it was a limited list. Never mind, we’ll be in plenty of reports once our secret plan for world domination in marketing is a success. Oh wait, shh, forget I said anything. Anyway, the research and reporting was sound as far as I could see. There’s the whole issue of the what that data is that you have given to them – essentially everything you type – and people’s usual laziness about privacy until something like this pops up (which I’ve talked about many times before) but aside from that… a real piece of new, useful content.
t as I watched the various carbon copy articles from the various ‘reputable’ news sources spew out in reaction, I was compelled to make a different observation than my original (and somewhat admittedly, fluffy) intention to illustrate how Ireland too, was ‘feeling all the feels’.
Specifically, my point is this:
All the data and resources in the world is of little use to anyone, without a crucial addition of a few brain cells.
Just because Italians use love heart symbols less than the French does not make them less in love. Nor does it mean the Germans are on par with Italians when it comes to romance.
Just because Canadians use a lot of gun emojis does not mean they are more prone to violence than Americans.
Just because someone has categorised an emoji as ‘happy’, ‘sad’, ‘violent’, ‘family’ or even ‘raunchy’ doesn’t mean the rest of the world is going to agree – especially on a country by country level.
Just because Arabic speakers are more likely to use a bikini emoji, does not mean they are more interested in clothes. It may be diet. Or that they take more holidays. Or in reference to a particular place, person, emotion or thing.
Just a quick survey of the Simply Zesty office alone tells me that the symbols attributed in this research as LBGT (which of course they could be used for) are automatically attributed in our office to ‘friendship’ – two same sex people holding hands.
But that’s just a sample of this little office. Perhaps it’s different where you are?
Because emoji symbols are just that, symbols.
And whether it’s a smiley face or a status luxury car, symbols are complex, highly contextual and constanstantly changing systems of communicating meaning.
This is a problem I keep encountering in the world of ‘big data’ (cringe) and the absolute worst sinners are us marketers.
You can use all the best software and latest tools and have the fanciest lookinginfographics you want, but at the end of the day, it is like putting a Hennessey Venom GT in the hands of a blind monkey with a drinking problem. And beyond that, even if you are an expert, your audience may not be. So, while the SwiftKey research appears robust and does not – to great extents – infer too much unjustified ‘insight’ from the data shared, once it is in the hands of marketers and journalists, it has become a different story.
To be clear I adore services such as BrandWatch and SurveyMonkey – who make research and ‘big data’ more accessible to everyone. In such new ways of thinking about problem solving, lies the future of research and real insight that has the power to change the bottom line of a business.
But just because you have an easier, fancier method of gathering data does notmean you throw methodology out of the window too.
Nor does it mean that the people using these tools no longer have the need for something as mundane basic maths, or analytical and statistical skills… never mind any training in how to choose, design, represent and structure research in the first place.
Unfortunately, I would struggle to come up with more than one piece of recent research I’ve stumbled upon on social platforms in the last few months that had a truly sound methodology – or in many cases – any at all.
In the past two weeks alone I have (politely and privately) highlighted to no less than 4 researchers – on live research – that their methodologies or survey structures had problems, to the point the results would not be usable. This is because I want research in general, to have a good name, and modern methods in particular. Otherwise, what the point of it, or strategic planners, for that matter? I also feel obligated to lend a hand to people, if it will make a significant difference to the investment of time and money they and their clients have already put in, and highlight a need for training. But… the general response is nearly always a polite ‘thanks but oh well’ and only one person I’ve ever contacted has actually partially adjusted a live survey. One of those contacted was in relation to a government financed survey that had no less than four serious issues that meant only one ‘public opinion’ would come out on top, exclusively. These then – with even the best intentions (and I do believe they were done in the best intentions) – go on become published pieces to again, do the rounds and be reiterated and deformed by media into misinterpreted monstrosities.
Or worse still, used to make important business decisions.
Incidentally, I have also been approached twice in the last month, to direct people in other businesses to ‘quick, free solutions’ to solve their client’s research ‘problem’ because (and I quote in one case) ‘It’s my responsibility but I don’t really have the time to learn about this stuff… and couldn’t be bothered anyway‘.
On that note, I want to use the opportunity presented by this current story to state categorically the following:
Accurate sentiment analysis through the automated monitoring of the use of single words or even phrases is not possible. FACT.
The comparative measurement of the use of a word, a phrase or a smiley does not an accurate – or even semi-accurate – measurement of sentiment make. Ask any trained linguist or AI developer.
Believe me… you don’t want to be in the world where computers could do that anyway.
Swiftkey are aware of this in their research for sure. And I applaud them for it. Yet daily, I come across dozens of ad agencies and 3rd party software companies out there claiming the opposite. And using it as a key selling point for software.
If that’s why you invested in social listening software, directly or through an agency, you need to stop and reassess – disregard those automated sentiment reports and demand better.
Listening tools are invaluable. I believe, absolutely everyone should be using them. And marketing or business decisions without research or strategy will only get you so far (i.e. not far). But just cut the sentimental waffle and lets focus on solving real business problems.
If you would like to find out how to currently use listening tools or how to improve your research strategy to give you some real, practical and usable insight – that spans all aspects of business – drop us a line.
– See more posts like this by swinging the work and personal blogs at:
http://www.simplyzesty.com/Blog – lots of awesome posts from smart people on tech, trends and practicle advice for your business in a digital world.
https://furcoatsnoknickers.com/ – my own little trends blog.