Digital intelligence in underwriting: The human perspective
Digital intelligence in underwriting: The human perspective
Data should be utilized to inform and enable, not overwhelm. Artificial intelligence (AI) and machine learning (ML) tools can analyze torrents of data for us mere mortals at the other end. Which means that while working to increase the amount of data at our disposal, it’s important to keep the users – underwriters and agents – and the customers at the forefront.
When considering the challenges of big data and what it can bring to the underwriting enterprise, one could argue that all data is important. However, there is a difference between what is sought, what is kept, and what can be used.
The right data has to be clearly connected to the business outcomes one desires. In underwriting, the right data is the data that gets you to the risk assessment as fast as possible and enables the best possible customer experience. This data focuses not only on the current decisions being made, but also helps determine what could be significant in the future.
What is the cost of data and how do we optimize it?
We must look at the holistic ecosystem, not just the data elements when we think about cost. To begin with, cost of data depends on the data source. Some data sources can be tapped into for cents per item, while others may cost $10, $20, or $30 per record retrieved.
Additionally, one must take into consideration the infrastructure that supports the data and how it’s processed, as well as the human costs associated with training and ongoing maintenance. The cost of data storage must be factored into the price equation, as well as real-time availability.
When it comes to optimizing the cost of data, the answer lies in finding good partners and thinking strategically. Early in the process of gathering data, there can be red flags. There could be early warning signs when using new products that might signal future difficulties, warning signs that it might be difficult to understand, to train to use, and to manage. Plus, the associated human costs might be high as well.
As a general rule, if a data source poses too many questions or warning signs, one should take a step back and ask the hard questions about the types of data one needs to bring in and how to utilize the data.
How will underwriting handle big data?
This exponential expansion in data increases the requirements on the data system, including the ability to triage data, to compress it, and present it to the underwriter in a usable way. Therefore, it is imperative that we search for pertinent technologies that can scale and increase the speed of the underwriting process.
At the same time, it is important to take into consideration the actual use of collected data. The traditional underwriting process consisted of getting an application form, some disclosures from clients, and possibly some lab tests. In many ways, it was just a snapshot of what we are now capable of receiving.
Now it is common to receive seven years’ worth of lab results, along with electronic health records containing years’ worth of history, including every lab test that a person has had over the last few years.
New sources of data are available and waiting to be tapped. These sources include digital health data, electronic health records (EHRs), and billing codes. Access to these types of data will continue to expand. With various devices such as personal wearables, we will begin to get more insight into people’s behaviors. The ability to lean into the behavioral data may become more important for the future of underwriting. It will be critical for collection, storage, and management of this data to be done in a way that protects the rights of both the customer and the insurer.
Will it be data or data science that differentiates insurers?
Traditionally, insurers are looking for evidence of their customers’ mortality. That was the universal risk assessment. The future, however, is really in looking at all the parameters as to why someone might live longer, and then collaborating and aligning with customers in ways that support that result. When mortality tables say you are going to die may become less important than how you are going to live as more data becomes available.
There is an ongoing debate as to whether data or data science is more important. On one hand, with data, you never know what precious metal lies within the ore. Ultimately, though, data scientists extract the gold. They are the ones with the actual domain knowledge who understand risk assessment and mortality.
Take the example of location information. When this information was first made available within various apps, most individuals had no idea or never thought about the level of privacy that was exposed. People’s lives were revealed in ways like never before. It was not the data that led to these revelations. It was the data scientists interpreting the data.
How does this apply to underwriting?
Ninety percent of the world’s data has been created in the last few years, and we are still barely touching the tip of the iceberg. However, while facing this increase in the amount of available data, the market has produced better rules engines, better data analytics, and better technology.
A great deal of technology is entering the underwriting process. It started with electronic manuals, which progressed to electronic rules based systems, mostly processing paper. Then came electronic applications running straight through the system.
However, realism has set in on some of this technology. Some things take much longer to come to production than you might imagine. While we are now at a point where more business is written electronically than on paper, we have yet to conquer an entirely paperless process.
While the adoption of some technologies is taking longer than we would like, we are now starting to see some technologies, such as rules engines, reach maturity. The majority of larger underwriters now use rules engines yet smaller companies are lagging behind due to a lack of resources to invest in the latest technology. Fortunately, there are ample more choices that are easy to deploy.
The one thing that is lacking, though, is confidence. From an underwriting perspective, these new models and applications need to be proven. People will trust a machine to land the plane they are sitting in, to drive the car they are sitting in, but would they trust a machine to underwrite their insurance policy?
What are the pain points?
Even with today’s modern technology, underwriting is still a painful process for agents. There are not many messages that are as hard to deliver as telling a prospective policyholder that they have to get invasive tests completed and the results may not be available for weeks. It is a difficult process to explain to a consumer, particularly in today’s society where people expect instantaneous feedback.
For insurance applicants, the underwriting process can be viewed as an inconvenience. However, it is a necessary step. This is the challenge. We cannot make it go away. Insurers need to examine how to make it less painful and how to communicate better with agents and customers.
Because of this difficulty, over the years, we have seen more of a focus on the simpler processes to sell products such as final expense products and guaranteed issue. Agents love the accelerated issue products because they can sit with a client and go through the application, with a clear line of sight to the enforced policy by the end of the session. They want to sell these types of products. Our challenge as an underwriting community is to make the more complex underwritten products, the higher face amounts, go through as quickly as those simpler products.
The advent of accelerated underwriting is very positive. We are now seeing people being accepted through the underwriting process without lab testing and without medical exams. This involves using data to pre underwrite people and select those who can go through the process more quickly. While this now occurs less than 50% of the time, it’s a big step in the right direction.
The other action we must take is to prioritize improving the agent experience, the customer experience, and the experience of all users involved in the process. We must find a better way to communicate and make the process as fast as possible, in order to sell products easier and more in line with positive consumer experiences.
How are roles and responsibilities evolving?
The underwriter
As we visualize the tools or services that the underwriter of the future will be using, the focus is on simplifying what the underwriter has to do. Traditionally, an underwriter is burdened with the task of collecting data and analyzing it. This leaves very little room for assessing the actual risk.
Underwriting platforms of the future will give underwriters the ability to provide more manual tasks to machines, making the underwriter the custodian of the process. Rather than collating and processing information, and then producing the outcome, underwriters will go straight to the outcome.
Although the value of the underwriter will no doubt increase, there may be cases in the future where expertise is not required and cases are completely automated. Carriers in the U.S. are investing time, resources and money to obtain greater automation within their underwriting processes. However, despite the prevalence of technological solutions, only a small percentage of carriers have achieved the level of automation desired. There is much work to be done to increase automation throughout the industry in the U.S.
With the increased adoption of technology within the underwriting process, there is a wider use of rules engines. Real time data is now available for rules engines to ingest. As these engines become more sophisticated, they enable more business to go straight through to completion. There is still, however, a decent chunk of business that is complex to underwrite, and has multiple risk factors. In these cases, there is such high face amount coverage, and so much data, that some issues are red flagged and an underwriter must manually review. In these types of cases, the technology will be used more to compress and consolidate the information, and then present it in an easily assessable format, helping to guide the underwriters in their decision-making.
Even on the straight through business, underwriters will still have a big role in training these systems, as well as auditing and monitoring them. Technologies must be carefully monitored and managed.
There will always be a human element to underwriting. Humans create technology and data scientists discover new methods of assessing risk. It is a combination of human beings and technology, more like augmented underwriting than automated underwriting. That is the future. The human component is going to change dramatically, but is never going away.
Underwriters are going to manage every application as well as exceptions. The future will be about managing exceptions in such a way that creates the data necessary for machine learning algorithms to decide whether something needs to be an automatic decision.
The new role of underwriting will require a different skillset and knowledge base. Moreover, it will consist of an absolutely new process driven completely by humans, assisted by machine learning and technology. Machines are going to bring different aspects to the underwriting process but the underwriter is going to be in charge of orchestrating all of the elements.
The customer
The customer experience will become akin to the Amazon experience, offering suggestions based on previous search and purchasing histories as well as the purchasing trends of other customers. There will be more value given to validating information. For example, frequently, 50% of the information underwriters receive is objective data and 50% is what the client has provided. In the future, there will be various techniques to validate all information.
Whether they realize it or not, the underwriter has always controlled the customer experience. So much of the customer journey is impacted by and takes place in underwriting. Yet, underwriting, for a long time, was behind the veil.
In the future, that is going to change. Underwriting will control the customer experience: where it happens, how it happens, and who is engaged in it.
The agent
The role of the agent is changing. As we think about the customer experience, we also need to think about the agent, who, in some ways is the custodian of both the customer and the information that’s given for risk assessment.
The agent, being an active participant in delivering the right customer experience during the risk assessment process, is paramount. In a perfect world, the agent would be behind the chatbot. When a person has questions, they connect to the chatbot, but the agent is there to have the conversation.
This is where the agent adds value.
Life insurance will always be with us, and like the underwriter, the role of the agent is not going away. Their effectiveness and productivity, however, is going to be dramatically higher as we integrate into a single, seamless, customer experience.
What will the new digital underwriting landscape look like?
The new digital underwriting landscape will be one that’s a lighter lift for everyone involved.
While it may sound like the underwriter’s workload will increase, in terms of orchestration, their engagement with new platforms and technology ecosystems will be one that’s more conducive to their work.
Traditionally, the underwriters and case managers have received the short end of the stick in terms of technology deployment. If there is a work around, it’s going to land on their desks in order to preserve the agent experience and the customer experience.
The new digital landscape will be one where all the key players are equally involved, where the tools and the different providers will work better together using new technological advances. New tools are heading in the direction of smoother processes specifically when taking into consideration maintenance and optimization.
In terms of the platforms that will enable the future of underwriting, simplicity will be the key -- being able to resolve all problems simultaneously. The underwriting experience needs to be as slick for the underwriter as it is for the agents and the customers. In order to achieve greater levels of simplicity, digital laborers or robotic process automation (RPA) must be applied to the more complex processes.
Exceptions need to get smaller so that underwriters spend more focused time on doing assessments. The machine should be used to screen risks, raise risks, and flag risks. Solutions such as cognitive computing should be used to predict models. These concepts need time to work together to develop a smooth underwriting experience. Allowing underwriters to use this information, and process it quickly and efficiently, is key.
Platforms of the future are going to be able to simplify behavioral statistics as well. They will allow future underwriters to consume data, and make sense of it immediately by providing risk assessments based on the machine’s algorithms.
There will also be elements where machines will be able to take data from social media platforms, and other areas, to say, for example, “You said you don’t do any dangerous sports, but your Facebook profile shows you were skydiving yesterday.” How far we will want to take this, or where our platforms will go with this, is going to be key.
Another benefit to the new platforms is that there’s no capital investment needed. You can use the cloud -- it’s secure. It’s when legacy systems come into the picture that the customer experience suffers. Unless legacy systems are transformed completely into the newer technologies, the customer-forward experience is just not possible. It’s not just about data and integration anymore. It’s about delivering data at the right point in the process with the level of processing to accomplish an exceptional experience. Legacy systems impede progress and therefore need to be retired.
Multi channel selling platforms
As we see a rise in the adoption of new platforms and channels for sales, it is becoming clear that multi-channel selling platforms will shape the future of underwriting. Traditionally, the sales process has been very much agent focused. Life insurance has always been sold through agents, originally door to door, then face to face or through phone calls.
We are now seeing exponential interest and growth in the direct to consumer channels, both from startups and B2C channels. Customers are being offered different ways of accessing products. There is also an increase in the omni channel model, which is a single platform with different entry points. The advantage of this option is that it only has to be built once, and can be deployed wherever one is doing business, making it easier to enter into new channels or offer different channels to customers. It also means that, regardless of what entry point a customer is using, there is a consistent customer experience as well as a consistent agent experience across the different channels. This great development will make it much easier for people to buy life insurance, and have a better experience during the process.
Multi-channel platforms will bring changes to underwriting. The underwriting process won’t necessarily change, but the pricing side will be different. Traditionally, actuaries have placed a lot of importance on the channel producing the business. If you’re selling through an agent sales force, it will be a different segment of risk versus a customer coming to you directly. And that is another interesting paradigm. With multi-channel platforms, customers can come from a variety of different channels. In the future, we cannot count so much on the channel to assess the segment of risk that is coming in.
When implementing new underwriting technology, carriers are essentially replicating their paper based processes, and using machines to adapt to what was being done. We are on the cusp of using machines to inform better, learn better, and make recommendations. There are plenty of platforms that will do the data capture, offering seamless, end to end journeys and launching products very quickly, but we are not yet using the full power available. Realizing this potential should be the goal of adopting a genuine underwriting platform.
Looking to the future
We need a philosophical shift in the way we launch technology projects. Rather than trying to recreate a process along traditional lines, we need to think about how we can work differently and move the process to a whole new level. We need to view technology as an enabler, so that we can harvest the opportunities new platforms provide. In order to change the process, insurers must first have a strategy and a road map of where they want to go and how they plan to get there.
EXL offers in depth underwriting automation for a wide variety of applications. We can do everything from the simpler applications to more in depth underwriting. We understand what the client needs and brings to the endeavor.
The hardest thing about adopting new platforms is the needed change in management itself. The platform itself is an enabler. What you are looking at then are changes in business processes as well as changes in the roles of the people who will be implementing and using them. That in itself is a big shift and a big ask.
Insurers need to embrace these changes and figure out how to draw people in. Many insurers want the system to adapt to their process rather than take advantage of new systems and develop a better process. That is the key for most insurers.
In addition, perhaps most important, we need to remember the importance of adopting a holistic approach. We must look at the customer experience and the agent experience, the underwriter’s experience and the actuarial experience. Ultimately, we must dispel the notion that data is gold. Data is the ore. Without data mining by data scientists, there is no gold.
It is digital intelligence, the combination of human and machine that is going to take underwriting forward into the future.
Written by:
Ajmal Malik
Vice President, Product Management
Sandeep Manchanda
Business Head, Life & Group Solutions