Wouldn’t it be nice to know what our buyer personas are thinking? To know exactly what they want? To get a firm grasp of how they feel about your product or your brand? This is why social sentiment analysis has risen in prevalence. But it has its critics. There are some who think it is a waste of time.
Mike Sutton is one of these social sentiment naysayers, who felt “fairly disappointed” by where “Sentiment Analysis is today”. Equally scathing, influencer Mark Schaefer publishes a brief article titled: Why you can’t believe social media sentiment analysis (“it’s crap”). The title leaves little room for interpretation. However, there are those who advocate as vehemently as Schaefer condemns it.
Meg Carpenter believes that sentiment analysis “provides an excellent source of data”. Brittany Berger, similarly, argues that it “is like a metric that adds context to other metrics”.
So, who do we believe?
And, why is it such a contentious subject in the first place?
What is all the fuss about?
Social sentiment analysis is a specific form of social listening. It attempts to analyse and describe how people are reacting. By monitoring the interactions on social posts, the idea is to get a feel for how people are responding to your brand.
At first glance, the idea seems to be anything but a waste of time.
Since 2010, this process has started to pick up traction Moreover, the discussions make it less a marketing tool, and more a field of study. The machinery of marketing can collect data and make calculated inferences on its tone. But can it give us more than a grade somewhere between positive and negative?
As Ron Sela points out, social sentiment analysis at it’s most basic, is a tool that counts positive and negative keywords. If there are “more negative keywords” then the overall tone of the piece is negative.
In essence, this seems like a solid approach to it. But, this is where my own scepticism starts to itch. To gauge the reception of your content means having to decipher one of the most difficult and unreliable codes ever invented.
The lexicon problem
Whenever anyone tries to tell me that a software can understand the authorial intent, I have to roll my eyes. Maybe it is the healthy cynicism my English Masters degree installed in me, but I think that it’s the most naïve things anyone can argue.
Meg’s article did suggest a list of tools that would help you with your social sentiment analysis. Each of them yields a bunch of useful tricks. However, none of them really go as far as decoding the language of a thousand voices in unison. It seems to me that tools such as Talkwalker Quick Search and Hootsuite Insights have handy features. But, to consider using any software to determine sentiment is ultimately a waste of time.
It is foolishly dangerous to assume that clever algorithms and coding can effectively psychoanalyse scrappy text. It is an even crazier notion to assume that these algorithms can pick through thousands of dialects and acronyms to get an accurate depiction of how a brand is being received. The problem is, thousands of people might use negative and positive words. But these words don’t necessarily mean the same thing to everyone.
“Yeah right” and other sentiments
Perhaps the simplest demonstration of this argument is also the most reductive. However, consider the words “yeah right”. In most of our dialects and every day speech, we understand this to be translated into a facetious statement. You might as well say “whatever”, or “keep dreaming”. However, consider the two elements of the phrase “yeah right”.
Yeah: A derivative of the word yes. A statement or exclamation of agreement.
Right: In this sense, it is closest to the word meaning “correct”. Another positive word.
But to us, “yeah right” doesn’t mean “yes that is correct” as it might suggest.
This is the most basic example. However, the way language is evolving is more complex. Words like “peak” are being used in colloquial areas to describe something negative. Words that are used in Northampton seem to have an entirely different meaning when trying to converse with someone in Birmingham, Liverpool or Glasgow.
You could be forgiven for thinking that the way someone uses words in speech might not be relevant here. But then, this is social media. Users are writing about their day. And, they aren’t reaching for dictionaries as they do it. More and more, users are writing what they think – in the way they would say it.
And then there is this little thing called sarcasm.
Social Sentiment Analysis and Sarcasm
It was comedian Micky Flanagan that first made me realise the one inimitable thing about growing up in the UK. When it comes to sarcasm, we are trained from a young age. In fact, it is how we greet our best friends. Personally, I don’t think I have had a conversation with my own parents in the last twenty-five years that wasn’t a ripping dialogue peppered with torment.
And machines just don’t get it.
The Tay Tweets debacle proved that.
When machines try to emulate humans, or mimic their interactions online, it all goes well until the most important obstacle of all. Humour. Machines don’t understand irony. Even the most sophisticated AI built struggles with the concept of friendly insults and mockery. Yet, this is what we do on social media.
Read through your feeds, or your notifications.
Do your own quick social sentiment analysis.
If software scanned through the comments and posts on your feed without a fluent sense of irony, how would they fare? How would the message be received and understood? Would it be a waste of time? Is everything as clear cut as it sounds, or has sarcasm defied the machine again?
So, you’re saying it’s a waste of time then
Something needs clearing up. Despite the less than favourable scrutiny this article has levelled at social sentiment analysis so far, it isn’t the intention that causes concern. It’s the method.
The idea of social sentiment analysis needs nurturing and understanding. As a process, it is still in its infancy, but the idea behind it is sound.
Trying to discern what our customers desire has never been a waste of time.
In fact, more time should be devoted to the practice. Not because we can improve the software enough to be able to sift through the comments and give us an accurate picture of what our customers collectively feel. Because to adequately describe any form of human sentiment, I believe we must strip the machine right back.
Superhuman Social Sentiment Analysis
To put social sentiment analysis in the hands of the robots seems to go against everything that inbound marketing attempts to achieve. Inbound marketing does use a lot of software tools to simplify the processes, however, there is one thing that sets inbound aside from what has come before it.
The prospects and customers are centre stage.
Your marketing efforts divert much of your team’s energies into seeking approval from your buyer personas. The software used to attempt to discern how your prospects are feeling, presents its findings in the form of graphs, percentages and other numbers. The thing with mathematical presentations of anything, is that numbers rarely have nuance.
If we are to see and treat our buyer personas as human beings and individuals, then representing them as statistics is a complete waste of time.
The old-fashioned way is no waste of time
Humans understand humans.
And, whilst the tools do put all of the feeds and information in one place, they are better served as a support mechanism. To get a better feel for how people are responding, you need the analytical skills of a marketer who can combine their social listening and monitoring skills to join conversations and shed the ambiguity.
Social media management tools aren’t adequate to perform your social sentiment analysis. But having them will make the analysts job far simpler.
Social sentiment analysis will prove useful if the results are nuanced enough for you to be able to use them when creating your next batch of content. If you can translate the sentiments into something tangible, then your next lead generation effort should be much easier.
So, I suppose, the answer is no.
Social sentiment analysis is not a waste of time.
The debate is wrought from sides expecting too much from bots and the cyborgs that skitter through the system. But the premise is correct. In some ways, it feels like this analytics method is more like a form of cyber-psychology.
How do you perform your social sentiment analysis? What tools do you use? Do you agree with the overarching argument that technology is an obstacle to getting adequate result? Your thoughts and debates are interesting. I would love to read them.