Underwriting has come a long way since 1668 when bankers at the Lloyd’s of London insurance market, were paid a fee, for literally writing their name under the risk they were willing to cover, as specified in the Lloyd’s document.

Profound technological advancements — especially the advent of big data and analytics— have substantially changed the rules of the game and precipitated radical shifts in traditional underwriting in recent years.

A steadily growing number of organizations are applying big data and analytics for risk assessment– using it to gather and verify data and drive improvements in underwriting precision and focusing on outcomes such as profitability and customer lifetime value.

The Business Case for Big Data in Underwriting

In Banking

New data sets, new opportunities

A vast amount of customer data is available and accessible to banks today.

It is Big Data that provides banks the ability to connect diverse and oblique data, residing across multiple sources that were not necessarily designed to converse with one another in the first place. And in doing so, it helps to unravel a single, comprehensive view of the borrower- deriving many new meaningful insights and unlocking a new kind of intelligence in the underwriting process.

In an era when tracking data was difficult, lenders typically relied on credit-bureau data, including credit-line utilization, delinquency status, and credit inquiries. It was all based on a thin stream of data. But, new sources of “alternative data”, such as mobile records, DTH recharges, social media presence, click behavior, granular customer-payment and spending behavior etc., might arrive by the gigabyte, even for a limited number of prospective borrowers.

Such analysis consumes enormous computing horsepower and cannot be provided in real-time. Moving over this processing to Big Data solutions and leveraging analytics helps banks proactively incorporate these various data elements into their workflows.

A new crop of online lenders like Kabbage are scouring a wide range of these untraditional sources of data for clues to creditworthiness. Rather than solely looking at credit histories, these companies crunch current, real-time data to develop credit risk models and provide short-term loans to small businesses. Through a partnership with UPS, Kabbage has found that shipping patterns can be a reliable indicator of credit risk.

Similarly, there are other online lenders that are using Big Data to lend to people with “thin files,” meaning borrowers with no built up credit record. For example, Jumo, a South African startup, taps into cell phone data to lend to unbanked “thin file” customers.

These are exciting areas that are ripe for further innovation in the future.

The Digital Underwriter:  Making Automated Decisions

To offer rapid real-time answers to customer requests (e.g., applications for loans, opening accounts), banks will likely need to find ways to assess risks and make intelligent decisions without human intervention. This calls for the use of more “alternative data” along with zero-based thinking.

A good example of this would again be Kabbage. The small business lender provides a quick, convenient online loan-application experience, eliminating the need to submit lengthy documents. It assesses oblique data sources such as PayPal transactions, UPS shipment volume and, e-commerce trade transactions of applicants to make instant, automated lending decisions.

To enable seamless and quick onboarding experience, many banks have now designed account-opening processes where most of the requested data are prepopulated from various public sources.

Accurate risk models, enhanced lending Performance

The rapid adoption of machine learning models, which identify complex, nonlinear patterns within large data sets, is making more accurate risk models possible. These models are able to learn every new bit of information they acquire over time to vastly improve their predictive power.

The Gini factor- which is just a measure of how effective a model is in terms of its ability to discriminate between good risks and bad risks– often improves substantially with such intelligent decision making.

risk model


On-demand access to granular data on active loans, accrued equity, and the ability to generate quicker and more accurate appraisals are driving important changes for the banking industry and helping players radically outperform traditional scorecards in both consumer and business lending.

Big data analytics also offers substantial upside in portfolio monitoring. By serving as an early warning system in situations where the risk of default is imminent, it contributes to significant reduction in defaults and enhances overall lending performance.

In Insurance

Precise Pricing, Enhanced Value

Insurance is also ripe for the incorporation of real-time external data within its underwriting operations.

The use of analytical models to underwrite and price policies is not new to the insurance industry. However, the use of near real-time big data would help insurers access much more granular, up-to-date information than is typically included.

In the new underwriting paradigm, the actuary can supplement traditional questions with scientific questions that drive better risk decisions. A real life use case here would be that of commercial auto carriers.

Traditionally, commercial auto carriers have relied on the business’s profile coupled with the MVRs (Motor Vehicle Record) of the drivers within the fleet to assess risk. However, some carriers are embracing emerging sources of driver data – to scrutinize individual drivers within a commercial fleet.

The rationale here is quite simple: MVRs are just one piece in the whole puzzle. Given that most severity-heavy claims are due to errors on the part of the driver, it behooves insurers to better manage risk at underwriting by looking more closely at the risk associated with individual drivers.

A big data driven approach here can drastically increase the likelihood of identifying new patterns or revealing previously hidden realities: For long haul trucks, for instance, fatigue is the leading cause of fatal accidents. Vibrations transmitted from the engine beneath the seat are the main factor behind driver fatigue. Preventing fatigue can significantly enhance driver safety, leading to lower loss ratio.

Armed with these insights, the underwriter can ask data the right question- are all trucks fitted with seat dampeners?

The end result is a significant leap forward in insight discovery and a more informed assessment of the policy submission, leading to more accurate pricing of the premium cover. 

Real-time Analytics with big data

Underwriters are enabling carriers to figure out optimal ways to use sensor-based technologies for constant monitoring of customer behavior and real-time pricing and policy term modifications. Such technologies can help optimize insurance protection and personalize premiums for the insured. In motor insurance, for example, telematics is being used to keep track of a customer’s driving habits and send data back to the carrier in real-time.

A major example in this area is that of Aviva in UK which encourages safe driving with its driver behavioral app called Aviva RateMyDrive and calculates tailored premiums based on how safely customers drive.

Though the use of flexible algorithms to build customer risk profiles coupled with the increasingly better understood role of real-time analytics, insurers can take current usage-based insurance (UBI) models to an altogether different level- offering   fully dynamic UBI in the future.

In Sum

With big data now being a part of the information supply chain, tomorrow’s winners will have underwriters who play considerably different and higher-value roles than they play today. To achieve this state, organizations must create operational data frameworks that ensure open and connected data throughout the enterprise. For those that do, a wealth of potential value awaits.