Sentiment analysis: ‘Social’ savvy banks can gauge flip-flops of customer service in quick time
By : Flytxt R&D Team
Social media is expected to be a vital cog to the new-age banking with its potential as a media to engage with digital customers at scale as well as its ability to spread sentiment and messages faster. It offers financial institutions an ideal opportunity to share not only promotional offers and information but also address customer issues and concerns before it aggravates further to result in mass distress and churn.
Indian consumers are getting increasingly digital savvy with around 90% increase in internet user base and more than 150 million active social media users with majority of them on Facebook. It explains why most of the banks have their own Facebook profiles and use Twitter not only for marketing and customer service, but also to offer new innovative services. The recent ICICI and Twitter tie up – ‘Tweet to Transact’ is an example of many initiatives from banks in going ‘digital’ and ‘social’.
Flytxt carried out a sentiment analysis of 4 of the top notch banks in India to analyse how their digital presence and initiatives are shaping up customer opinion and market sentiment. We picked both nationalised as well as private institutions to have a better comparative picture. Data for the past 4 months was extracted from relevant Facebook and Twitter pages maintained by these banks for analysis.
The text data (tweets, feeds, comments) was preprocessed to remove unwanted text such as special characters and stop words, followed by tokenisation where a sentence was split into words or tokens. This was followed by the process of lemmatisation where the tokens are converted into groups of different inflected forms of a word, called lemma. After this, a bag-of-words model was created which is essentially a list of all tokens.
After this, the sentiment analysis was carried out based on opinion lexicons which are basically sentiment dictionaries containing lists of English positive and negative sentiment words (around 6800 words in total). An intersection of the bag-of-words model with the positive sentiment dictionary and negative sentiment dictionary created the lists of positive sentiment and negative sentiment words, respectively. Top 100 positive and negative sentiments were visualized as positive sentiment and negative sentiment word clouds, respectively.
A sentiment score was also calculated based on the number of positive and negative sentiments which determined the overall sentiment polarity levels.
Let us have a look at some of the results of sentiment analysis done for each bank:
Bank 1: Rewards won followers; but there’s no replacement for good service
One of the largest privately owned banks gained a lot of support on social media for its cash back reward schemes. Offers like winning a free Kindle and special ones for women customers earned appreciation on social media. Invoking positive sentiments by spreading awareness of such offers and reward schemes can be one fairly successful social media strategy for banks.
However, an analysis of keywords that resulted in negative sentiment clearly reflected the fact that customers are not happy with their grievances redressal system. Instances of passing the buck among customer service executives and relationship managers as well as lack of communication have been major pain points raised by customers.
Bank 2: Digital customers applaud and curse in same breath
Analysis of social media comments from another large private bank also proved that offers and incentives go a long way in receiving positive vibes and cheers from customers. For example, there were lots of positive comments on offers like free online shopping vouchers on first transaction using internet banking or a free e-book reader with mobile banking services. It is a closed loop with the digital customers appreciating the digital initiatives of banks on digital channels.
However, the bank also received quite a bit of negative feedback and sentiment on their services like online payment transfer, credit card services and mobile banking. Many customers cited poor customer service as a reason for their dissatisfaction. The analysis clearly showed that social media is a good platform to gauge positive and negative feedback on services, more so for digital ones.
Bank 3: Even failure of one product line can lead to negative sentiment overflow
For this Indian private sector bank, assured, efficient delivery of bank products and free reward points proved to be a winning factor to gain trust, acceptance and appreciation from the social media peers.
However, loan customers expressed displeasure with loan accounts not being closed on time despite numerous follow-ups by the customers. Lack of communication and ownership by the managers added to the fury. Many criticised the service, calling it ‘worst’ and ‘pathetic’.
Bank 4: Loved by the masses; lethargic approach is a dampener
This Indian multinational, public sector bank has always been popular among the Indian middle-class community. It has always been associated with trust and security. This image is further exemplified with a promise for a more personal experience. Even the offers have been directed towards appeasing the masses like free bus tickets, which have also have been popular.
Despite the popularity, the lethargic and bureaucratic approach associated with a typical public sector organisation has been a dampener even among the social media masses. Little or no vocal communication due to a flaky helpline number was also a deal breaker as was use of archaic vocabulary in the fine print.
Overall, Bank 2 and Bank 4 had a greater social presence. While the former earned 90968 number of tweets/feeds/comments, the latter too enjoyed nearly as many responses at 90662. Bank 1 and Bank 3, respectively, earned 49635 and 52040 responses on social media.
Bank 2 had the highest positive polarity followed Bank 1 and Bank 4. Bank 2 was promoting various interesting offers from the bank through social media which made their sentiment score higher.
In a nutshell, incentives and rewards invoked very high positive sentiment. Similarly, good services and quick and active responsiveness to customer problems earned a positive feedback for these banks.
Contrariwise, the presence of negative sentiment was around a number issues and complaints pertaining to the customer service. While there is no replacement for good services, working on the negative sentiments could improve the customer experience. This will, in turn, improve the popularity and profit for these banks.
Listening to social media enables a deeper understanding of customer’s feedback and emotions for banks and if they can act on time it can help them in not only retaining customers but also enhancing brand loyalty and relationship value.
-By Dr. Prashanth R., R&D Manager – Data Sciences and Deepak K., Associate Architect – Big Data Applications, Flytxt