Advanced Analytics in Insurance

Insurance is an industry where innovation is vital to customer-centricity, and particularly in customer service. According to an Accenture study, nearly 61% of customers trust digital tools that monitor their application status.

One key area of focus for innovation in this industry is the capability in information extraction from data. This helps to predict claim occurrence and price for the premium on providing effective service delivery to policyholders. According to McKinsey, nine out of ten insurance companies identified legacy software and infrastructure as barriers for digitalization. As a result, this industry — which, in the US alone, accounts for $1.2 trillion — needs to change the way it manages business development urgently.

Innovation through the adoption of analytics can be classified into three areas:

a. Data sourcing from multiple sources and data handling for information extraction

b.Risk monitoring in premium pricing and fraud detection in claim handling

c. Gamification in behavioral understanding and effective service delivery

Let’s take a look at data sourcing and handling in detail:

1. New Data Sourcing:

The extraction of data from an individual’s or group’s online activities (digital footprint) helps insurance industries to continuously perform behavioral analysis.

Wearable devices help in gathering health-related information of an individual which is helpful in medical analytics.

Telemetric helps to track real-time information about properties, especially Automobiles. For example, rapid acceleration while pulling away from traffic lights increases the likelihood of an accident, as does sudden braking. When the driving pattern is smoother, the chances of accidents occurring reduces. Auto insurers encourage smooth driving, by monitoring the telemetry data stored by the on-board computer of an automobile. This computer stores a log of relevant parameters, which are used to deconstruct an individual’s driving style, and determine the risk associated with that driver / the automobile. Feedback is provided to the customer to help improve their driving style.

2. Old Data Handling:

Old data refers to legacy information in the form of documents and manuscripts where OCR (Optical Character Recognition) helps in digitization. With OCR scanning, a varied range of paper-based documents, in different languages and formats, can be processed into machine-readable text that not only eliminated paper piles but also makes previously inaccessible details available to everyone, with a few clicks and within seconds. NLP (Natural Language Processing) helps to extract the required information.

3. Synthetic Data Preparation and Computer Vision:

Synthetic Data usually helps in a prior analysis of the insurable event, especially in the case of virtual data simulation for driverless vehicles in Auto Insurance. Synthetic data is created from real-world data by stripping out the identifying aspects such as names, emails, SSN and addresses from the data set so that it is anonymized.

Computer Vision helps in gaining higher-level understanding from digital images and videos, through pattern recognition from the given images using sequences of the algorithm like feature-point extraction and object tracking (for videos).

Need for Computer Vision:

The FNOL (First Notice of Loss) and several claims deal with incident pictures, analysis of the environment and surroundings. For example, to claim insurance, the policyholder has to register First Incidence report, followed by sending information to the insurance company with evidence of insurance. These claims are validated by documents, which are verified by investigation teams in multi-dimension, leveraging technologies like Computer Vision.

Computer vision does more than helping identify and adjust claims, collision avoidance helps prevent incidences of claims. Collision avoidance requires extremely fast identification speeds, and involves some relatively simple applications (e.g., drones avoiding objects), as well as complex applications, as a part of Preventive Insurance.

The typical process in claim handling & underwriting:

Computer vision analysis kicks in for on-spot DAMAGE/LOSS analysis as well as recasting the event scenario through ENVIRONMENTAL analysis. The analysis includes data from wearable devices to telemetric, data from geographical and video capturing units, which captures the live moments of the Damage /loss incidents.

More recently, video analytics has a fresh dimension, that brings in notions and principles from the social, affective, and psychological literature, and that is called Social Signal Processing (SSP), which adds more weightage to Behaviour characterization through computer vision.

The FNOL analysis with computer vision helps to solve for the HOW & WHEN part of the investigation but answering the WHO and WHY — that’s where fraud analytics comes into the picture.

b. Risk Monitoring and Fraud Detection

Premium pricing for policies is based on the risk involved in underwriting for them. The risk is dynamic in nature and varies continuously based on behaviour attributes (in case of Individual policy) or financial performance of the company (in case of group policy). Hence monitoring of the risk plays a vital role in premium pricing to manage losses.

Fraud Detection is where several advanced analytical approaches have been employed since handling fraud manually has always been costly for insurance companies. Even if one or two incidences of high-value fraud go undetected, it’s costly for the insurer, especially as it opens up the possibilities of a new type of fraud that is undetectable.

To tackle this, several fraud detection tools are available in the market that generates “SUSPICIOUS SCORE” based on fraud history and social network analysis, through Predictive Modelling.

Here’s how BRIDGEi2i adopted its Claims Fraud ManagementTM Solution to identify fraudulent claims better.

c. Gamification

Gamification enables better sales in insurance, where all you need are a selfie-and a chatbot, to activate the insurance policy instantly; the triangulated analysis of data from various social platforms, historical data, wearables data and more can help design products for every individual. This also helps in taking a minimalistic approach in collecting information from consumers and making it easier for them to buy, interact and get serviced.

Gamification enables fraud minimization through identity verification and authentication, making the pricing/quoting process much faster. A conversational BOT that engaged the customer in their journey can really help to understand various behavioural aspects.

· If we look at gamification in the industry today, Farmers leveraged a game called Farmville that revolves around the concept of virtual crop insurance to promote its brand name. American Family Insurance created the i-AMFAM simulation on Facebook, where users can create avatars and manage life activities such as buying a house and planning a career, helping them decide which insurance to purchase at what point in time.

· Lawley Insurance provided a gamified experience, rewarding sales teams for active on-time logging of sales data. Sun-Life Financial launched an online platform to engage and educate young customers on retirement and investment planning with levels, challenges, and social interaction.

In summary, the industry has a strong demand for technological talent and needs partnerships to drive the process of innovation through analytics. Mobile and ubiquitous automation have already become the reality of insurance. Insurance apps and agent management software make claim handling and communication easier for digital-savvy clients. AI, IoT, Blockchain, wearables, and Telematics are emerging technology trends of the near future that should be taken into account, to stay ahead of the competition, and to drive improved customer experience and enhanced operational efficiency.

If you enjoyed this article, here’s another read on how analytics can address insurance challenges!