4 min. read
Listening to customers has always been the key to making sales. Most socially mature brands don’t need to be convinced of the value of social media listening, but many remain cynical or even unaware of social sentiment analysis.
Social sentiment analysis, also known as opinion mining, is simply the process of determining whether a mention you are tracking on a social channel is positive, negative or neutral. This is far from a simple task, however, for reasons we’ll go into.
Nevertheless, sentiment analysis is still a highly useful and an informative addition to your marketing and customer care. You should think of it as a simple way to gauge which way the wind is blowing.
Cut through the noise
The beauty of social media listening is that it does the legwork of sifting through millions of sources to draw out the mentions you have defined. These mentions could be anything from who is talking about your brand and in what tone, to how people feel about industrial or social trends.
We recently covered the ins and outs of how to best use social listening, so take a look.
In terms of marketing, the insights gained can be pure gold. You can save a lot of time and resources by acting on the data. For customer care and experience management, it is one of your strongest assets for nipping trouble in the bud.
The public nature of social media has endowed the customer complaint with unprecedented power. A negative post or message needs to be jumped on ASAP. Social media listening is how you do it; with social sentiment analysis helping to bump the most insightful or critical messages to the front of your to-do list.
We can only wonder the degree to which Pepsi and United Airlines employ social listening and sentiment analysis. But the year’s biggest PR gaffes so far provide textbook examples of why brands need to be monitoring their channels closely.
Listen more closely, engage more effectively
Our Falcon platform features a social sentiment indicator based on Lexalytics text analytics. Our Listen social monitoring tool employs it to assign positive, negative, neutral or unknown scores to social mentions. It is also a feature in our Engage inbox, where it applies the same classifications to incoming social messages.
Falcon’s unique inbox priority algorithm, noted by The Forrester Wave™: Social Relationship Platforms Q2 2015, prioritizes incoming messages partially based on their sentiment ranking.
A frivolous case in point…
When word hit our Copenhagen office recently that Spanish A-lister Antonio Banderas was, wait for it…”water skiing on a table” in a nearby canal, we opened up Listen and entered the following search queries: “Antonio Banderas” OR #antoniobanderas) (København OR Copenhagen OR Canal OR Kanal OR cph OR kbh OR Denmark OR table OR “table surf*”. We were quickly rewarded.
Not surprisingly, the random sight of Zorro (or Puss ‘n Boots if you insist) table-surfing in central Copenhagen is being well received on social.
The big challenge: us and all our quirks
Putting an accurate sentiment ranking to social speak is not an exact science.
The sheer volume of posts aside, sentiment analysis software must contend with the human factor. This ranges from dialect and slang to sloppy spelling and grammar.
Classic examples would be the slang usage of ‘sick’ or ‘ridiculous’ as positives. And what is software supposed to do with the statement, “I’d kill for a burger, YOLO!!!!”?
Added to that is the muddle of hashtags, abbreviations and acronyms along with the brevity that characterizes how we talk on social media.
Nevertheless, here are some simple tips for beating the slang, sarcasm and socialese:
•Define your own ‘word bags’: by this, we mean defining lists of words or terms that you are reasonably confident will be used in positive or negative contexts. Set listening projects based on these and keep an eye on the ratio.
•Apply machine learning: posts featuring slang terms may turn up negative or neutral when they are obviously not. With Falcon, you can change the sentiment rating on these. Over time the software will pick up on the change being applied to particular combinations and apply a more accurate rating.
•Benchmark: this is another trick to add context to chatter. For example, although a mention may seem negative towards your brand, wider matters may be afoot. With Falcon, you can see how your overall sentiment scores rank alongside other products or similar brands.
The next step in sentiment analysis: intent
This technology is still in its infancy, but its potential is exciting. It enables brands to pinpoint purchase intent simply by being able to pick up on phrases such as, “I need a new phone, any recommendations” or, conversely, when a customer’s comments suggest they may be leaving your brand. It will be interesting to see how this develops in the near future.
One more tool in the democratization of data
A long-standing estimate is that only 20% of the information available to a company is in a structured form that actionable data can be immediately drawn from.
An even older maxim is ‘knowledge is power’, so that remaining 80% is an enormous, largely untapped resource for companies to leverage. Once, tapping into that 80% was the preserve of companies with vast resources or budget enough for specialized research companies.
But these days, with the democratization of data in which social media is a crucial player, the field is being leveled.
Social listening, given extra context by social sentiment analysis, is how marketing teams with limited headcounts and resources can make substantial inroads into that untapped wealth of data.