Companies are continuously seeking innovative and effective ways, to understanding the sentiments or socialized feelings consumers have towards her brand. There is a plethora of social media monitoring and engagement platforms out there like: Linkfluence, Sprout social, Hoootsuite UberVU, brandwatch, synthesio, sysomos, Tweetdeck and a host of others. It is a congested market place as each company strives to attain a higher accuracy level on social sentimental analysis.
A few days ago, I signed up for a trial version with one of these social medial monitoring and engagement platforms. I was quite curious to see how insightful and in-depth her sentimental analysis tool will be. I typed in a random company and it returned with an over 55% negative sentimental ratio. I clicked with an intention of generating more insights but to no avail. All it provided was geographical location of the customers or users. Next to the sentiment tab, it had an icon known as ‘terms.’ I clicked on it with a hope of generating common terms associated with each sentimental metric, and all it provided were abstract and very limited information.
Ben Donkor, a social media analyst wrote a good piece about how sentimental analysis transcends a tripartite metric of negative, positive and neutral. In this piece he stated that the conventional yardstick for determining a metric is this: “If a piece of content has more positive keywords than negative keyword. It’s a positive content; if it has more negative keywords than positive keywords, it’s a negative content.” Now using this parameter to evaluate the sentimental metric of an online comment, review or tweet, is very misleading and shallow. Sentiment analysis should go beyond the basic level of keywords and phrase; deeper elements need to be captured for richer and well informed insights.
Elements in Social sentimental analysis
In having a thorough understanding of the several elements in sentimental analysis, I will extract a review on a company called Lyft, from the Trustpilot review platform. We will dissect the different elements that make up social sentimental analysis, with an aim of discovering how negative, positive and neutral metrics are at best surface indicators- that should usher deeper analysis. The review goes thus:
“I would love it if Lyft were successful, because I think competition is good. Problem is, I think they will fail as Uber provides a much higher level of customer service and is far more professional. Lyft has many issues, but one of them is their ‘Primetime’ pricing which is calculated in a much more customer unfriendly way than Uber. As an example, if Uber says ‘Surge pricing’ is 150% it means that if the normal rate is $10, you will pay $15. That makes sense to me. With Lyft, when they say ‘Primetime pricing’ is 150%, it means that if the normal rate is $10, you will pay $25, that is, nearly double what you would pay on Uber. And in my market, UberX is 20% cheaper than. Add in all the attitude, with Lyft (fist bumps, I am supposed to be friends with my driver) — higher prices for an inferior product. Finally, customer service at Lyft is surly and basically non-existent.”
Topic: This is the main subject of the discussion, review, tweet or complaint. In the above scenario, the topic of the review is on Lyft. Now to a certain degree the topic also focuses on Uber, albeit in a comparative manner. The reviewer is certainly talking about the two companies, with a prime focus on Lyft.
Aspects: This focuses on the sub topics and in this scenario it includes things like customer service, primetime pricing, quality of products and attitude. In having a better sentimental analysis for this review, each aspect needs to be drilled into, to ascertain objectivity or subjectivity.
Sentiment: In the above scenario, two forms of sentiments are displayed. A positive and negative sentiment towards lyft and a positive sentiment towards Uber. Now from a Lyft standpoint, there are plethora of negative sentiments that appear to be more objective than subjective, due to the deep level of facts and data presented by the customer.
Location: This review was made in the United States and the location of the reviewer is also an important element of sentimental analysis. My next article will look in-depth into the relationship between geographical location and social sentiments. The individual’s location in the United States also provides him with an opportunity to compare the service between lyft and Uber, as these companies only operate in selected countries.
Holder: In this situation, it is a one-person account or review. In certain situations you may have two individuals view about a product or brand. An example could be: “I got the laptop bag delivered on time, I like the colour but my mum thinks it is dull.” In this scenario, there are two holders or reviewers, the individual and his/her mum.
Profile: This is an area that has not been given any much thought by social media intelligence and analytical firms. I have been researching and working on using profiling to have a deeper and more informed understanding of sentiment analysis. From the above customer review, the dominant profile is that of advocacy. An advocacy profile is one that categorises individuals that can easily recommend a product or service. These individuals easily recommend a product or brand; they are also viewed as brand advocates. They believe in the good of a company and can easily recommend such a company to friends, family and the online community.
The two important phrases in the above review that indicates an advocacy nature of the reviewer are: “I will love it if Lyft were successful” and “I think they will fail as Uber provides a much higher level of customer service.” They would love it if Lyft were successful, which is a positive sentiment synonymous to advocates. The second statement shows how the reviewer subtly recommends or advocates for the service of Uber.
Leon Chaddock believes that, sentimental analysis is fast and close to real time. This is very obvious, but social media monitoring and engagement platforms should strive to provide a platform that shows more indicators than a mere negative, positive and neutral metric. This will be continued in subsequent articles.
