Advancing Insurance in the Digital Age - Data Automation and Risk Assessment
Banking Frontiers organized its annual InsureNext conclave in January in Mumbai. This lively panel discussion was the first panel discussion of the day explores insights into competitive differentiation, greater role of AI, data architecture, strategic in
Dipak Nair: You can create a new app, it gets copied in no time, right? Processes can get copied. Anything that is visible to the eye can easily be copied. So, in today’s world, how do you differentiate? What are the areas of improvement?
Siddharth Shah: I think any form of differentiation must focus on great customer experience, great automation, great product design and great actuarial assumptions going into your pricing, it is all contingent on data. Your ability to underwrite effectively and fast is contingent on data. So, having the right data and being able to provision it and make it available as needed will be critical for any insurance company to be successful. About the 4% of GDP insurance penetration in India, If the aspiration is to increase that and if the aspiration is to actually have insurance for all by 2047, it is almost critical that we start underwriting risk that is currently not covered, which could be both on the commercial /non-life side, as well as on the life side. The reality is that big risks of big companies, those have already been underwritten.
If you take a look at salaried employees in India, the perfect retail risk, they’re also broadly underwritten between group and individual policies across health and life. So, the ability to get to that next stage requires you to really underwrite the next set of population - both MSMEs as well as individuals - which could mean going to tier 2 & 3 markets, going to new to credit customers, and new to financial services customers. Data is increasingly important in being able to effectively and profitably underwrite that risk.
As we think about data pipelining, we need to aggregate the right kind of data. One would have a lot of historical data that resides in various systems within insurance companies, but very often it tends to be siloed. Next is to make that data available. For example, if you’re running real time analytics, how do you actually convert that funnel from every hit that comes onto the website and how do you actually provision that data and make it accessible and build use cases around it? If there’s a customer who is in the middle of a journey, how do you make sure that they don’t drop off from the journey? Or if they do drop off, how do you make sure that you take them to a logical conclusion?
Shashi Kant Dahuja: Though we are the most populous country in the world, in terms of overall premium, we are at number 10, and in terms of general insurance, we are at the 15th rank. In terms of life insurance, we are slightly better at the 7th rank. Despite facing numerous calamities like the Chennai floods, Gujarat floods, Odisha cyclone, etc., there is a significant protection gap in India. It is widely recognized that India has the highest protection gap globally, with only 54% of vehicles and 10% of houses being insured. Affordability remains a significant issue hindering insurance penetration.
In the insurance industry, data plays a crucial role in various aspects such as underwriting, claims processing, and risk assessment. However, there persists a perception that insurance companies often do not pay claims promptly or fairly. To address this perception, there is a need for leveraging data automation and advancing technology within the industry. A focus on improving the onboarding process, which is often considered a pain point, can significantly enhance customer experience and satisfaction. Insurance companies are increasingly
utilizing a combination of internal data sources, external data sources, and historical data to refine their operations. Moreover, initiatives like integrating data from platforms such as Vahan and leveraging information from CIBIL demonstrate progress in enhancing data utilization within the industry. Despite these advancements, there are still areas where further improvement is required to enhance transparency, efficiency, and customer trust in the insurance sector.
Dipak Nair: On one side and AI & ML and on the other side is the human intelligence. Currently, where do you see this needle between them? Do you see that this needle moving towards the machine side in the next 18-24 months?
Syed Meraj Naqvi: I don’t think that in next 12-18 months it is going to move towards AI. It is important to do data collection, data processing, developing a pattern and finally deliver something which is useful to the customers Human intervention is important because it is not the case that only the system will generate something useful for the customers.
Dipak Nair: How do you see fulfilling the gaps that are existing in the data architecture so that business can literally move this needle faster?
We are doing claims adjudication with limited data that is available. We are doing automated underwriting, but it has not really replaced underwriting. Even in claims adjudication, algorithms have not replaced the adjudicator. At best, they are giving indicators - red, amber, green.
Siddharth Shah: There are two points of view. Business cases for big ticket technology and data modernization exercises continue to be very difficult to realize, especially when average policy size is pretty small. And very often you’re looking at technology which is cloud based and it’s growing exponentially. So, the one thing we need to look at is that the solution must be about not just for financial benefits but also for employee e x per i e nce , c us t o mer e x per i e nce , reconciliation challenges, etc - things that we often overlook. Only when you make investments will the cost of incremental sale come down significantly.
From the supply standpoint, I think there are two major challenges. One is in terms of, you know, just adoption and business value realization. Even today, we see many instances where platforms get rolled out or data programs get rolled out, but the adoption and the efficacy of those continue to remain average.
Investment in data and AI is crucial, but how do we convince business leaders who might prioritize urgent needs over long-term strategic investments?
Shashi Kant Dahuja: As we explore the realm of data-driven initiatives, allow me to illustrate the significant return on investment (ROI) inherent in such endeavors. Consider this scenario: our company recently confronted a potential fraudulent claim totaling `4 crore. Leveraging the insights derived from FasTag data, we promptly identified and refuted the fraudulent attempt, thereby preventing a substantial financial loss. This tangible example underscores the transformative capacity of data in risk mitigation and safeguarding our financial interests.
Moreover, our f oray into data analytics has yielded valuable insights into customer behavior, particularly regarding women drivers. Through meticulous analysis of claim ratios, we have discerned that women demonstrate lower claim frequencies, especially within urban settings. This newfound understanding has enabled us to tailor personalized discounts e xclusively for female drivers, thereby enhancing customer satisfaction and refining our risk assessment capabilities.
Now, envision an insurance landscape where data serves not merely as a tool but as the cornerstone of innovation. Envisage an insurer leveraging traffic patterns and meteorological data to incentivize safe driving practices through discounts or offering customized coverage aligned with the specific risks prevalent in a customer’s geographical region or profession. These scenarios epitomize the boundless opportunities afforded by data, fostering a mutually beneficial relationship wherein insurers thrive alongside policyholders.
Miraj, as a broker, how do you navigate the challenges of working with multiple insurers across diverse product lines? Can you shed light on your data utilization in customer lifecycle management?
Syed Miraj: As brokers, our core mission is to guide clients towards the right insurance policies, ensuring they understand the options and make informed decisions. To achieve this, we leverage a robust CRM system that captures a wide range of customer data, from basic demographics to service types and supply chain details. However, a significant challenge lies in harmonizing this data from various insurers to identify patterns and recommend personalized add-ons or policies.
Think of it as a matchmaking process – we represent both t he customer and the insurer, striving to create a perfect match. This requires meticulous data management, ensuring seamless synchronization across diverse sources. The goal is to generate insights that ultimately benefit the customer, especially during the crucial claim settlement stage. Achieving this level
of customer satisfaction necessitates a s ubstantial investment i n data management strategies.
In essence, brokers play a vital role in bridging the gap between insurers and policyholders. By effectively harnessing and leveraging data, we can navigate the complexities of the insurance landscape, ensuring clients receive the coverage they need at the right time.
How does the risk assessment process at quotation align with manual methods versus the potential shift towards machine learning, data, and artificial intelligence?
Syed Miraj: Initially, we rely on manual processes, providing patterns to the system, which gradually learns and refines responses. The essence lies in managing historical data and claims experiences effectively within the system. Striking the right balance in response to clients with varying turnovers poses challenges, highlighting the distinct perspectives needed for small and large turnovers. The ultimate aim is to integrate machine learning and data analytics for more precise and reasonable risk assessments tailored to diverse customer profiles.
Data analytics and AI have vast potential, but they come with biases. Training systems can lead to wrong decisions, akin to ‘garbage in, garbage out.’ How do companies balance innovation and a futuristic outlook with the bias that systems can introduce, hindering adoption? Adoption is crucial, so how do you perceive this balancing act between innovation and bias?
Siddharth Shah: It’s interesting because the f undamentals of t he insurance industry are to manage risk. Here, we are managing the risk of bias creeping into our data, right? Absolutely. So, I think there is no silver bullet answer to it. I think there’s a spectrum where multiple organizations are trying their best to ensure that they can minimize any bias that creeps into the data.
So, I’ ll give an example. We recently worked with an organization on the topic of claims fraud, and I think that came up. So, we looked at the datasets and realized that we didn’t feel that the dataset was adequately representative of all scenarios. Incidentally, we used GenAI to generate synthetic data to strengthen the claims models, the fraud models. And then we were able to, at least in the proof of concept that we did, see the model’s efficacy go up significantly, and now that’s in the process of industrialization, right? So, this is a classic example of being innovative while simultaneously trying to manage risk or bias in the data.
Data privacy regulations like the DPDP Act add another layer of complexity. How can companies balance innovation with responsible data use?
Shashi Kant Dahuja: The DPDP Act emphasizes customer consent and transparency. While this is crucial, it shouldn’t hinder innovation. The key lies in building a robust consent architecture that clearly explains how data is used and empowers customers to make informed choices. Additionally, exploring alternative data sources, like anonymized browser data, could be a way to navigate privacy concerns while still enabling personalization.
Siddharth Shah: The Data Protection and Privacy Act (DPDP Act) heralds a significant opportunity for the insurance sector to champion responsible data utilization and foster enduring trust among customers. Central to this endeavor is the establishment of robust consent architectures and the meticulous documentation of data usage patterns over time. By adhering to these principles, insurers can not only ensure compliance with regulatory mandates but also cultivate a culture of transparency and accountability in their data practices.
In navigating the landscape of thirdparty data usage, insurers must explore innovative and alternative approaches to maximize the value of data assets while safeguarding consumer privacy rights. This entails scrutinizing existing frameworks and discerning novel methodologies that align with the evolving regulatory landscape. Embracing such adaptive strategies is imperative for staying ahead of the curve and fostering sustainable growth in an increasingly data-driven industry.
Moreover, the proactive pursuit of creative solutions underscores the industr y’s commitment to driving meaningful change and f os t e r i ng consumer confidence in data stewardship practices. By leveraging the DPDP Act as a catalyst for innovation, insurers can forge deeper connections with customers, thereby enhancing brand credibility and positioning themselves as trailblazers in ethical data management.
In essence, the DPDP Act presents not just a compliance requirement but a strategic imperative for insurers to reevaluate their data governance f ra meworks a nd r e c a l i brat e t hei r approach towards data utilization. Through steadfast adherence to best practices and a commitment to continuous improvement, insurers can navigate the complexities of data privacy legislation while simultaneously unlocking new avenues for growth and differentiation in the marketplace.