How bringing AI to the underwriting process will reduce combined ratio
Charles Murat Charles Murat
September 3 10 min

How bringing AI to the underwriting process will reduce combined ratio

It won’t be news to veterans of the commercial property and casualty (P&C) insurance industry that underwriting excellence plays a pivotal role in overall company performance. Neither will it be a surprise that reaching profitability is no easy feat: even top P&C insurers are currently operating near a 100% Combined Ratio.

As a significant customer touchpoint, underwriting is also subject to new and rising demands from customers, who today are increasingly accustomed to the easy, convenient, rapid-fire and customer-first service made possible by advancements in digital technologies and e-commerce. For insurance customers, the service they want isn’t a fancy online tool or mobile app, according to one recent survey. What they say they want more than anything else is “knowledgeable salespeople,” a confident voice on the other end of the line to quickly assess their situation and accurately answer their questions.

Providing customers with this resource is increasingly challenging. For one thing, experienced talent is scarce and getting scarcer: by 2020, it’s projected that insurers will be unable to fill almost half a million jobs just in the U.S. alone. Meanwhile, existing technology solutions are doing too little to help staff keep up. Some 30 to 40 percent of underwriters’ time is devoted to administrative work, such as data entry and executing analyses by hand, according to one report.

At every step of the underwriting process, accurate information helps reduce uncertainty around decision-making. Yet the cost of obtaining, validating and analyzing information adds up fast, and to maintain accuracy, speed is often sacrificed first. As a result, most insurers regularly lose out on policies that they could have written if they had been able to process submissions and renewals more quickly.

Artificial intelligence (AI) changes all of that. With the capability to collect, interpret and analyze data in real time, AI can give underwriters the ability to accurately and quickly answer complex customer questions. Beyond just information, AI is powerful enough to provide underwriting professionals with decision recommendations to help them decide faster and with greater confidence. In some cases, AI can even automate tasks, freeing up underwriters to spend more time focusing on higher-value activities, such as building relationships with brokers or evaluating higher-risk cases.

At Element AI, we’re developing AI solutions that can be applied across the underwriting process. Below we outline how AI can introduce new efficiencies at each stage of the process.

Underwriting submissions.

1: Submissions

Process applications faster and at lower cost

An application to purchase an insurance policy triggers the first step in the underwriting process: the submission. Today, many underwriting clerks still manually transcribe documents⁠—including application forms, analyst reports, notes and emails⁠—and manually cross-check for completeness. Not only is this time-consuming; data entry is also a prime opportunity to introduce typos and other human errors. Leveraging techniques such as optical character recognition (OCR) and natural language processing (NLP), AI can automatically read, digitally transcribe and structure this information, whether it is handwritten or typed, directly into a company’s system of record.

AI can further help improve and accelerate the submissions stage by flagging specific fields or challenging cases for underwriting professionals to review. If authorized to do so, AI can assess what key information is needed to process a specific submission and automatically request the correct forms or any missing or erroneous data, saving time and enabling the underwriter to receive the required information as early as possible.

2: Risk Appetite

Automatically flag submissions that don’t fit your risk appetite

Determining quickly and early on whether an application fits into an insurance carrier’s risk appetite, and assessing into which segment that risk falls, can lend underwriters an important new advantage over the competition.

Take renewals: when underwriters opt to review a renewal, a range of different high-level features may be considered, including losses on the account, major changes in business structure and standard application features such as age of the property. A lot of judgment is needed at this stage, and the manual process can be time-consuming.

The risk appetite stage of the underwriting process can be optimized with AI through automation. In the example of a renewal, if the model is confident, based on a confidence level determined by the insurer, the AI system is able to quickly determine whether the renewal should be cancelled, renewed or re-underwritten based on learning from characteristics present in the application and previous transactions, in addition to combining loss experience and various data sources.

Once an underwriter has decided to renew a policy, he or she must then determine the need for it to be re-underwritten or not, either because its needs changes in structure or price, or even because there is an opportunity for upselling. By analyzing historical data, AI can either automate this decision or, when uncertain, make a recommendation, which the underwriter can accept or modify.

With AI, underwriters can gain a better ability to detect submissions that do not fit their risk appetite. They also gain new efficiencies by avoiding a lengthy process for submissions that should be declined and offering well-aligned packages in a timely manner.

3: Assignment

Deliver quicker turnaround time with better assignment

It’s not always easy to assess which team member is best-suited to handle a given application. In many companies, it takes a dedicated team to assign and triage this work, and the assessment is often done by hand by an experienced staff member, or automatically put into team logs for underwriters to self-assign.

The assignment process can be continuously optimized with AI, which can assign cases to underwriters based on its analysis of a variety of different factors, including estimates of expected processing time, renewal rate, underwriter specialization, workload and signing authority. This is especially important because, for certain lines of business, underwriters can’t process all submissions.

AI can also fast-track the assignment stage by flagging which cases underwriters should prioritize, for example based on predicted closing ratio, lifetime value, premium and effort. This allows underwriters to focus their time on the most important cases, improve speed and consistency and, ultimately, increase gross premiums written.

4: Risk Evaluation

Make sense of complex information faster and with better consistency

Information overload can slow down even the most experienced underwriters. At most companies, underwriters leverage data from a wide range of different sources to assess the specific risks an applicant has and whether these risks can and should be covered. Since each submission is unique, each one also requires a unique review strategy, guided by the judgement that individual underwriters have developed and honed over years of experience. Although guidelines exist, experience counts for a lot; indeed, experience has been shown to correlate directly with precision and consistency in risk evaluation. Yet as veteran underwriters retire, they are increasingly being replaced by inexperienced new workers, making it difficult for underwriters to maintain consistency and efficiency.

AI optimizes the risk evaluation process. It can recommend information sources that were useful for similar applications in the past, or suggest a ranked order in which to review data sources. These recommendations help underwriters prioritize information from submissions and outside sources such as databases, allowing them to zero in on impactful information more quickly, increasing risk assessment speed and accuracy.

In addition to giving underwriters access to the most relevant and timely information, AI can provide a recommendation on whether the underwriter should quote or renew and at what degree of risk. It can also tell the underwriter how and why it came to that recommendation, explaining its rationale through visuals such as graphs and in the context of similar previous applications and decisions. The underwriter decides whether or not to accept the recommendation, and the AI can learn from this decision and use it to inform its future proposals. By automating the parts of the risk evaluation process that are predictable and repetitive, AI can help underwriters both make more consistent decisions and better control risk.

5: Coverage recommendations

Deliver the most relevant coverage options

Coverage recommendations have an important impact on underwriting profitability, and for many underwriters, this step remains more of an art than a science. To build out coverage options, underwriters assess the policy needs of the customer as well as the risk exposure of the existing book of business. They leverage their experience together with high-level guidelines that are encoded into existing systems of record, which can flag violations for underwriters or block them from issuing a quote altogether. This isn’t a quick and easy step, but rather a complex and lengthy process that involves analysis and multiple decisions.

Feeding off of past policies, AI gives underwriters access to a menu of recommended coverage options, including limits, deductibles, surcharges and rebates. In addition, the system can provide past examples of policies with similar application characteristics for reference and comparison. The coverage options proposed by the AI system may also take into account the company’s strategic goals. Loss ratio and closing ratio are just two examples of factors that could be calibrated in the system. This information helps underwriters prepare coverage options more quickly and with greater insight, as well as helping to build internal consistency.

Underwriting referal and authority.

6: Referral & Authority

Streamline reviews and approvals to accelerate the whole underwriting process

At the last step, right as the underwriter is ready to send the quote off to the broker, it sometimes happens that the case could contain an unusual and potentially problematic risk or other issue. AI can monitor for inconsistencies and flag any possible problems for the underwriter, who can decide whether to adjust elements entered into the system or continue forward without making changes.

AI can also help to ensure that decisions are approved by underwriters with the right level of authority. In a case where an underwriter is handling an account that requires managerial approval, the AI system is sometimes able to approve on a manager's behalf based on its analysis of the manager's history of past decisions. If the AI system isn't confident, then it can make a decision recommendation and flag it for the manager's review.

AI: the only tool to optimize all steps of the underwriting process

Across the underwriting lifecycle, underwriting professionals have the opportunity to capture powerful new efficiencies through AI that continuously learns and improves. By automating simple, repetitive tasks and making recommendations on more complex ones, AI helps underwriters process more applications in less time and at lower expense, leading to improved customer satisfaction and retention rate, and ultimately driving profitability up and combined ratio down. The technology is here—and so is the opportunity for those who are ready to grasp it.

To read on, download the full whitepaper.

Going to be at Insurtech Connect on September 23-25 at the MGM Grand in Las Vegas? Book a meeting with one of our experts to learn how AI can optimize your underwriting workflow.