Why those Facebook word clouds are hugely misleading

Like me, you have probably seen a proliferation of word clouds on Facebook recently. All good fun (turns out I use ‘Christmas’ a lot!), but they’re extremely misleading and here’s why.

Meaning is lost

The image above is my friend Simon’s Facebook word cloud. As you can see, he writes ‘people’ most of all. Quite a lot, I’m guessing, implied by the size of the word. Sort of interesting, but what’s the context to that? What is he saying about people? Is it around a particular area of interest? Is he talking historically? What does he actually mean? According to the Oxford English Dictionary the word ‘people’ has ten different meanings. Which one is Simon trying to convey? This type of visualisation simply counts words in isolation, which has the dramatic effect of losing valuable context to Simon’s posts.

Counting words is next to useless

Language in isolation is rarely useful. It becomes much more useful when we compare it to other examples. So what if Simon talks about people a lot? Maybe everyone on Facebook talks about people a lot, or maybe it’s a factor of his particular group of friends. Maybe he talked about people a lot last year and that’s skewing the results. It’s only by comparing his use of language in these additional dimensions that we properly understand what he’s trying to say.

Context is king

About that context again — it’s EVERYTHING in language (before the emails hit my inbox, I know it’s not actually everything btw, I just love the hyperbole and it is pretty important). In Simon’s example, how do we know he’s not being hateful when he mentions people? Only because he’s a friend do I know that he’s quite interested in politics and the impact of political decisions. That’s why he talks about people, in both a positive and negative way (let’s just say he’s not a huge fan of right wing politicians!), but nowhere does the word cloud highlight this important aspect to his posts.

This applies even more to brands

Whilst this is all quite interesting when looking at a friend’s posts on Facebook it becomes even more important when analysing a brand’s language on a proliferation of channels, for example measuring the success of marketing campaigns. If you’re using simple count-based word clouds then you have a huge context-shaped hole in your analysis. You’re missing the comparison to competitors and the industry as a whole. Crucially you’re missing the consumer reaction and context, and even more importantly how this changes before, during, and after the campaign.

If you want to chat about how you can do this and more at scale, faster and more objectively than ever before, please get in touch.

Rich Wilson is co-creator of his two sons, Relative Insight, and many other tech startups.