Yashish Dahiya and Alok Bansal founded insurance aggregator PolicyBazaar.com is not just in the news for its upcoming IPO slated for 2018 and year-on-year growth, the company has also surged ahead of the competition thanks to its strong tech backbone. Analytics India Magazine caught up with the Gurgaon-headquartered PolicyBazaar’s CTO & CPO Ashish Gupta to understand how the insurance aggregator is unlocking the value of big data and leveraging data analytics to make insurance trendy and accessible for its customers.
Over the years, PolicyBazaar.com has become the “Google of all Insurance policies”, a marketplace for insurances, pitching the right product to the right customer. From simplifying the buying process to facilitating product comparisons, the online insurance aggregator has created a niche in the market for its enhanced customer experience.
According to Gupta, the company that competes with other heavyweights such as BankBazaar.com, CoverFox and recent upstart EasyPolicy has cornered “90% of the market share, has interactions with millions of customers and an access to massive dataset of customer interactions”. “Big data is a key component and we leverage it to continually enhance the customers’ experience throughout the customer journey,” he said.
So, how is the company making insurance trendy again? Gupta cited a use case about segmenting customers and matching them with the right agent. “This leads to a far superior experience for the customer and much better efficiency for our operations. Next, based on historical trends we recommend the products and add-ons that a customer should consider. We then also recommend the right sum insured that would help increase the chance of the customer getting the policy while catering to inflation at the same time. Big data on the website also enables us to identify kinks in our systems, ironing out which enhances customer experience and conversions,” he shared.
Unlike other marketplaces, buying an insurance policy is not an impulsive decision. Which means selling insurance online is very different from online retail and data plays an important role in “pitching the right product to the right customer”. Predictive analysis is used to understand when to reach out to customers which is key to online insurance. Gupta, who leads a team of 10 engineers reveals segmentation is done on multiple levels, based on the information given in the profiles. “We first identify the customers based on their demographic profile and then recommend products based on that. We further analyze their behavior on the website – like if they are searching for something specific or just looking at various plans etc. We also segment customers based on CRM activities like email and SMSes, to identify the best possible cohort. Further, big data is used for upselling and cross-selling the products on the website,” he said.
On the other hand, the company is also hailed by insurance providers for providing a deeper understanding of the customer and the category. Gupta reveals how the company helps in predicting claims ratio and accordingly profile customers, thereby enabling efficient pricing of the end-product. “We are also able to identify opportunities that the current product range does not cater to and help insurers define new products that would suit specific customer segment. Once a customer purchases a product, we further use big data to identify the next most appropriate product that the customer should purchase. Analytics also goes into deciding which customer to reach out to and when for the best conversions,” he said.
Data Capturing & Storage
Over the last few years, Policybazaar set up a stable architecture for the last few years, so the data is very well mapped out and ready for consumption. The in-house data bank is where the data is stored and customer profiles are analyzed for allocations and segmentation. There are multiple tools and technologies used for big data. “Typically, the algorithms are based on Python and ‘R’. For visualization and consumption across all our staff, we use visualization tools such as Sisense to provide rich data in an easy to understand format,” Gupta shared.
Talking about sentiment analysis and how social media platforms are leveraged, Gupta stated that the company is very proactive in tracking and responding to grievances on various social media platforms like Twitter, Facebook. The data analytics team also keeps a track of trends that come out of the engagements. This, in turn, helps the company understand the customer sentiment and how are they reacting to Policybazaar and its subsidiaries.
Say Hello to Pbee – PolicyBazaar’s AI-enabled chatbot
Today, automation has become a must-have and is being layered across all functions in enterprises. To this end, PolicyBazaar introduced India’s first AI-powered chatbot PBee to sell insurance online. Besides being the first-to-market, the chatbot has been remarkable in other ways. Since the introduction of PBee, the chat volumes have gone up to 5x of what they were before the launch. The time to respond has gone down to 25% and now over 60% of all messages are handled by the chatbot. “For us, PBee is a remarkable success story and it has not only increased the productivity of our customer advisors but has also managed the sheer volume of queries received every day on our online platform. Our advisors receive nearly 20,000 questions per day – across all insurance products, and with PBee the response time has been reduced by eliminating human errors and seamless customer experience. PBee,” shard Gupta, talking about the AI agent. The chatbot is now being extended to health insurance and investment customers as well.
Why Is PolicyBazaar Betting Big On Voice Analytics
Gupta believes one of the biggest opportunities is voice analytics. He asserts voice is a goldmine of customer insights which gets lost in recordings and a digital solution could transform the way we the consumer and guide our product managers and customer representatives on how to service them better. The other opportunity lies in automating processes or making them self-serve. Citing an example, he said how the company has launched a video inspection that converts a three-day surveyor lead process into a 30-minute self-serve process. “The same would be extended to claims enabling instant approvals and saving a huge amount of hassle for the customer. More and more such initiatives are required to make buying insurance a truly user-friendly experience,” he said, signing off.
Try deep learning using MATLAB