Digital underwriting is reshaping risk assessment by integrating diverse data sources. These tools offer unprecedented insight but also introduce complexity, regulatory scrutiny, and the need for critical thinking. Success depends on balancing innovation with compliance, efficiency, and human judgment.
In 1961, President Kennedy declared America would go to the moon – not because it was easy, but because it was hard. That mission was not achieved by rockets alone. Data provided the fuel at every step of that historic space race.
Today, underwriting faces a similar frontier. The tools have changed, but the principle remains: Data drives progress. Underwriters have access to a richer, more nuanced set of health data than ever before. Each source offers a different lens into an applicant’s health profile. For example:
This is not about replacing traditional underwriting expertise; it is about enhancing our view, improving efficiency, and making more confident decisions.
As with any meaningful change, the incorporation of this novel data presents both challenges and opportunities.
Digital evidence is information dense. While its reams of data provide a comprehensive view of an applicant’s health history, it also means there are more data points for the underwriter to consider in every case.
More is not always better. Repetition of data within and across sources, introduction of unfamiliar visual layouts, and the presence of immaterial health history contribute to the noise from which the underwriter must discern a signal.
To mitigate this challenge, underwriters can leverage the decision making, scoring, and triaging capabilities many vendors offer alongside their digital evidence products. These tools help direct underwriters’ attention, allowing them to find the needles in the digital haystack.
Another challenge underwriters face in this era of digital evidence is increased regulatory scrutiny. This runs in tandem with questions from producers and consumers who may still be unfamiliar with its use and relevance.
One example is the use of credit-based behavioral risk scores to assess mortality risk. States such as Colorado and New York have introduced regulatory guidance that addresses concerns with insurers’ use of External Consumer Data and Information Sources (ECDIS). Compliance with these guidelines is critical, and underwriters play a key role in ensuring fair and non-discriminatory use.
At the individual producer and consumer level, underwriters may be tasked with explaining the use of data products for risk assessment. Having a strong understanding of the product, its inputs, and its relationship to mortality risk is important for successful conversations.
Finally, digital underwriting evidence sometimes lacks the level of detail of traditional sources. Underwriters should apply critical thinking skills to make decisions with the evidence at hand when appropriate. This will improve decision turnaround times and appropriately manage underwriting evidence unit costs.
An expanded digital data profile can provide underwriters with a more complete picture of an applicant’s health, allowing for a more accurate risk assessment and more effective case placement. As a bonus, by making the easy decisions even easier, primary insurers can devote valuable underwriting resources to complex cases where they are most needed.
Increasingly sophisticated insurtech helps to efficiently process and interpret this more robust data – no matter how much is returned on any particular applicant. Rapidly improving rules engines and predictive models can make sense of volumes of information that might otherwise overwhelm an underwriting team, while large-language models (LLMs) can immediately summarize unstructured data.
In addition, the ability to leverage a wide range of digital health data assets allows insurers to streamline the application process for consumers. With more comprehensive information available upfront, primary insurers can make quicker, more definitive decisions and reduce reliance on invasive or costly follow-up requirements.
This added convenience increases the likelihood that applicants will complete the process and acquire insurance coverage.
Unique protective value (UPV) arises when a single piece of evidence leads to a more adverse underwriting decision than would have been made without it. For example, digital data may reveal current smoking habits or undisclosed congestive heart failure, prompting risk reassessment. Identifying UPV is crucial for determining if underwriting evidence is worth the cost.
Consider this example: An applicant admits to cigarette use four months prior to the application, but not since. However, medical claims data indicates nicotine use one month prior to the application. The underwriter approves the case at Standard Tobacco rates.
Did the medical claims data provide UPV? The answer is likely no. Most primary insurers would assess tobacco rates for cigarette use within one year. Because the applicant admitted use within this timeframe, tobacco rates would be applied regardless. While there was misrepresentation by the applicant, it was not material.
Once UPV is established, a cost-benefit analysis (CBA) evaluates the cost of obtaining evidence against its protective value. Studies show that even with limited data hits, primary insurers can achieve significant savings by using digital underwriting evidence.1 This underscores the financial viability of integrating digital data into underwriting workflows.
When performing a CBA as part of a back study or live pilot, a key consideration is determining how much evidence is needed for a successful study. This involves balancing the need for a large-enough sample size for statistical credibility with the resources required – labor and expense.
One approach for managing resource constraints is to limit manual underwriting review to cases that are most likely to deliver significant UPV (e.g., older ages, higher face amounts, larger risk class deltas). Importantly, this underestimates benefit/cost ratio, allowing the study team to under-promise and overdeliver in terms of evidence value once it is used in production. However, this approach may not be appropriate for every primary insurer. Successful studies usually involve a multi-disciplinary team (e.g., underwriting, actuarial, financial operations, data science, etc.).
From the vendor’s perspective, any new product should offer a significant return on investment (ROI). If it does not, it is a hard sell. When assessing the value of a new data asset for underwriting, one important question is: What unique and significant information does the data provide? An equally important follow-up is: What resources are required to structure and interpret the data effectively?
Prescription histories (Rx) or medical claims data (Dx) excel on both fronts. They clearly identify unique and significant findings, and because these data sets are already structured, they lend themselves to automated interpretation. By contrast, EHRs contain rich and distinctive information, the value of which is still emerging as it can require significant effort to structure and interpret.
More information at the time of underwriting is generally a good thing, but what is the value of that additional information in relation to its cost? There are different approaches to assessing the impact of additional information, and choosing the right method depends on the context and objective of the analysis.
One example is the increased detection rates resulting from the addition of medical claims data to commonly used prescription data.
One could analyze how frequently the addition of medical claims data would result in a more adverse underwriting decision and, from this, estimate the protective value. This approach can effectively approximate potential mortality savings and also quantify how frequently medical claims data uncovers findings that a carrier considers significant. The results for this type of analysis depend on the data that drove the original decision and the logic used to interpret the newly added medical data.
Protective value and mortality savings are, in isolation, positive outcomes. But it is not always appropriate for underwriting to become increasingly conservative with each new data asset. The ultimate goal is to use additional information to make more accurate — not just more restrictive — decisions.
One way to address this dilemma is by updating underwriting guidelines to reflect the increased information available. Rather than ignoring new data because it may trigger more adverse decisions under existing rules, carriers can adjust their criteria to better match the new information mix. This may involve reclassifying certain applicants: Some who would have been declined may now be approved, and others who would have been approved may now be declined.
Predictive models present an altogether different way of assessing the value of a new data asset. Unlike humans, models have a nearly unlimited appetite for data and the ability to digest new information in ways that lead to the identification of more issuable risk, rather than simply narrowing eligibility.
We note that a model can only use the type of data that it has been trained on. If, for example, a model has only been trained on prescription histories, inputting medical claims data will not improve its output. But when a model has been trained on the new data type, additional input should improve its output. That improvement can present a carrier with some useful options: the opportunity to issue more business without increasing mortality, maintain sales while improving mortality, or maintain sales and mortality while streamlining the underwriting process.
In today’s data-rich environment, underwriters face a critical challenge: distinguishing meaningful insights from background noise. A condition-centric view — one that aligns data sources with specific medical conditions — can dramatically improve the signal-to-noise ratio and enhance decision making.
At the research and development level, extensive RGA studies have provided an answer to a fundamental question: When is enough, enough? Starting with traditional accelerated underwriting (AU) program inputs — disclosures, MIB data, and prescription histories — these studies sought to identify the necessary inputs for a confident risk assessment and decision.
The answer varies by condition and hinges on understanding the strengths and limitations of available data sources, from EHRs and medical claims to clinical labs and dental records.
As a leading global reinsurer, RGA’s approach is grounded in four underwriting imperatives:
Research shows that risk loads – the additional premium to account for uncertainty — are often comparable across data sources. For example, EHRs and APS, or clinical labs and insurance labs, yield similar risk distributions across many conditions. This insight opens the door to automation. Lab panels and medical claims can be automated. EHRs, though rich and complete, still require human review to ensure accuracy and context but can be pulled quicker than traditional APS.
Regardless of the data source, critical thinking remains the most valuable tool. The central questions continue to be: What do I truly need to assess risk accurately? What is my end goal? These questions should guide the evaluation of each case.
For example, for an applicant with diabetes, a current A1C result from LabPiQture may offer a complete-enough picture depending on factors such as age, treatment, and severity. For mood disorders, medical claims and prescription data can provide essential context. The key is to match the condition with the most relevant and reliable data source – and to know when human judgment must intervene.
On July 20, 1969, Apollo 11 touched down on the moon. Neil Armstrong and Buzz Aldrin stepped out of the spacecraft, sunk their boots into the powdery lunar surface, and began exploring this new frontier opened to humankind through the successful use of data.
Today, digital underwriting evidence is opening a new frontier for how insurers assess risk – bringing speed, precision, and depth to these evaluations. By leveraging emerging data sources and understanding their UPV, underwriters can make smarter decisions, drive operational efficiency, and deliver better experiences for clients.
However, technology alone is not the answer. The same case viewed through different lenses – EHR, claims, labs – can yield similar insights, yet each perspective offers nuance. Much like it took the human intervention of the third Apollo 11 crew member, Michael Collins, to keep the command module in orbit while his colleagues explored the moon, critical thinking remains indispensable in underwriting. Whether evaluating diabetes through lab results or mood disorders via claims and prescriptions, an underwriter’s ability to interpret and integrate data thoughtfully is what sets great underwriting apart.
As the insurance industry moves forward into this new frontier, underwriters must remain agile, data-driven, and committed to continuous improvement. The future of underwriting is not just digital; it is intelligent.
Contact us today to learn how a partnership with RGA can help your business put these principles into practice.
