Companies use NLP-based sentiment analysis to source intelligence
Pandemic-led complications are prompting companies across sectors to use artificial intelligence (AI)-powered sentiment analysis.
For instance, after the Reserve Bank of India (RBI) lifted a moratorium on loan repayments, banks and non-banking financial companies (NBFCs) had to deal with a backlog of pending loans. To accelerate the process, many NBFCs and banks began to adopt natural language processing (NLP)-based solutions to assess a borrower’s sentiment from their conversations.
A case in point is Credgenics’ sentiment analysis tool that uses speech recognition and analysis of chats over automated voice bots and WhatsApp bots to generate insights about borrowers. These have allowed NBFCs and banks to identify the problems that borrowers are facing in paying loans. Credgencis claims that over 60 lending institutions including ICICI Bank, Axis Bank and IDFC First Bank are using its sentiment analysis-based software-as-a-service (SaaS) platform.
“This has allowed them to plan the communication strategy and channel for the lenders for optimum outcomes,” said Anand Agrawal, co-founder and chief technology officer at Credgenics. He said sentiment analysis helps extract subjective meaning from text to determine a borrower’s sentiment. It is an ideal tool for reviewing unstructured content about a particular borrower’s digital communication for insights.
According to Agrawal, sentiment analysis has enabled lending institutions to improve their debt collection rates by 15-20% and recover 70-95% of their bad debts.
Sentiment analysis is also helping companies stem attrition. Firms are using these tools to identify employees who might leave, and retain them with perks, salary hikes, and a better working environment. “We have seen customers able to retain 85% of their top talent (using sentiment analysis),” said Tanmaya Jain, founder and chief executive of inFeedo, a SaaS firm that provides sentiment analysis tools to over 200 companies, including Samsung, Airtel, Xiaomi and Lenovo in India.
Jain said one of the customers in India, a large unnamed enterprise with over 3,000 employees, was struggling to retain employees after a large merger with another company. After deploying a sentiment analysis chatbot, the company was in a better position to estimate employee sentiment and managed to increase its retention rate by over 10%.
InFeedo’s AI chatbot offers insights based on its interaction with employees. The bot uses NLP to understand the context and identify employees who seem disengaged and are more likely to leave.
The use of sentiment analysis is not entirely new. Earlier, the use of NLP for sentiment analysis was restricted to tech giants such as Google and Amazon, which had more data and AI and ML engineers to experiment with it.
Among Indian companies, e-commerce firms such as Flipkart were the first to adopt it to understand customer sentiment by analysing user reviews using NLP.
NLP, a subset of AI, allows a piece of software to read, understand and derive context in text and spoken words just like humans. It can be used in any field where human conversation is involved. Before NLP, most AI-based chatbots operated and responded within a set boundary of fixed set of questions and answers.
Sentiment analysis has been around for years, but the interest in it is growing among many firms now as the underlying NLP technology has become a lot more mature. “What has changed is that now the NLP and sentiment analysis is becoming a lot more mature in terms of accuracy, readiness,” said Jayanth Kolla, co-founder of market researcher Convergence Catalyst.
He added that the talent pool of people working on it has increased in the recent past, which in turn has led to more adoption.
According to Kolla, demand for sentiment analysis has grown since the pandemic. He noted that a lot of HR tech firms are using sentiment analysis to read the chatter on platforms like LinkedIn and Glassdoor to rank companies.
For inFeedo, the demand has grown 3x since the pandemic. “Earlier, with employees being on premises, it was easier to understand employee sentiment, but with hybrid and remote work, and with video conference fatigue, it is difficult for leaders to gauge their employee’s sentiment,” said Jain.