Follow me

Just a few years ago, companies like Whirlpool or Honda used to analyze repeated words in warranty claims and service data and identify potential problems. Based on that identification, they used to take early action before the problem becomes big. Fast forward a few years, companies have access to a wealth of information about their customers through social media where people provide their opinions and views about brands.

Sentiment analysis or opinion mining involves building a system to collect and determine the emotional tone behind words and then gain an understanding of the attitudes, opinions and emotions of the people through that.

How Companies Can Benefit Through Sentiment Analysis

Owing to the popularity of social media, companies have started looking at sentiment analysis more seriously. Here are some of the ways in which companies can benefit from sentiment analysis –

Identify how people feel about the brand 

Track positive and negative conversations around the brand – identify the spikes, drill down to date, influencers, topics etc.

Identify the positive talking points around the brand and quickly tweak your digital campaigns to take advantage of those

Participate in the ongoing negative conversations to minimize the damage

Get inputs for development of new products/ services

Identify issues early on by building the systems to capture the initial warning signs

Know the public reactions towards specific events about the brand

Track the impact of marketing efforts

Understand the public sentiment about the brand as compared to the competition

How Sentiment Analytics Works

Quite understandably, sentiment analysis is a very complex and multi-step process. It involves Natural language processing and artificial intelligence. It takes the actual text element, transforms it into a format that machine can use, and then uses a lot of math to determine the actual sentiment.

The algorithmic approaches to sentiment scoring involve –

Vocabulary – Look for important keywords along with negation words

Rules – Look for presence of words in sentences and then use rules to categorize them by sentiment

Apply Machine Learning techniques – Develop a classification model which uses a dataset of positive, negative, and neutral content.

Challenges in Sentiment Analysis

Owing to the complexity of human language, tone, grammatical nuances, cultural variations, slangs etc., teaching the machine how the context can affect the tone is very difficult.  Sentiment analysis is, therefore, a very complex science. A sentiment like “I love my iPhone but hate my carrier” is very difficult for the software to decode.

Some words can be considered very positive in one situation and may be considered negative in another situation. Because of such differences, a system developed to gather inputs for one type of product or feature may not work well for another. For example, the use of word “long” can have a different meaning in two different situations. “My phone’s battery lasts long” is quite positive whereas; “my laptop takes long to boot up” is quite negative.

Named entity recognition, sarcasm, abbreviations, poor punctuations, poor grammar make sentiment analysis further more difficult.

Examples of How Brands Have Leveraged Sentiment Analysis

Expedia Canada launched an advertising jingle on television. But soon after its launch, the sentiment analysis showed a lot of negative sentiments around the music used in it. Acknowledging people’s criticism on social media, Expedia immediately launched anther commercial which showed the irritating violin being smashed. Expedia was able to respond to the negative sentiments in a playful way because it captured the sentiments quickly.

Barclays extensively used real-time sentiment analysis after the launch of its mobile banking application called PingIt. Within a few days of its launch, Barclays made significant changes to the app based on customer feedback.

The Obama administration had very effectively used sentiment analysis to gauge public opinion to policy announcements and campaign messages well ahead of the presidential election.

Delta Airlines continuously monitors tweets to find out how their customers feel about delays, upgrades, or in-flight entertainment. Any negative tweet (say about the lost baggage) is immediately forward to the support team which instantly takes care of it and this is helping in building a positive brand recognition.

Macy’s, one of the largest US-based retailers, uses sentiment analysis to apply predictive analytics and identifies unique trends that can impact their business. For example, tweets about a particular brand of jackets help Macy’s in deciding its future advertising campaigns and discount offers around that brand.

While sentiment analysis has its limitations, and cannot be 100% accurate, it certainly holds a lot of value. More and more organizations are becoming aware of the applications of sentiment analysis which can help them in fuelling the business growth.