Where next for underwriting automation?

When we think about automating the underwriting process, we should be looking beyond just speeding things up. Automation should be done with an eye toward improving the entire process, not merely making it faster.

How we automate the risk selection process has been one of the big success stories of underwriting innovation. There's good reason for this: even though some people contended at first that the human underwriter role could never be replaced, much of the selection process was actually very repetitive and boring. Automating those elements of the process was an obvious step, freeing up people to focus their skills and insights where they could truly add value.

The past two decades have seen significant improvement of the technology and rules behind the point-of-sale decision making. Unfortunately, the focus of automation has largely remained on the part of the process that relies primarily on what the customer can tell us. 

By not looking beyond this, we're missing out on further opportunities for automation. We can exploit other areas where automation can become truly game changing in the insurance onboarding process and beyond.   

For instance, we have already begun using the same automation capabilities to assess not just what the customer can tell us, but also some of the additional evidence that may be available to us from other sources, such as laboratory reports.

There are still some hurdles, as this is limited by a lack of standardisation of how this data is stored and presented. The addition of consistent electronic health records, a process that is still evolving, will be a massive step forward for insurers and their customers, while also saving the valuable time of the physicians who would no longer need to complete manual reports.

As we transition to an era where there is more and more data available to help make decisions, data automation is an obvious area where we can make strides. For me, this can take two separate routes: 

1. tapping additional sources of information, especially common in the US, to supplement what is asked of the customer
2. point-of-sale extraction of data that is already available on the customer's devices 

The first is a logical step already being undertaken by many, with existing digital databases containing huge amounts of risk-relevant information. Being able to leverage these at the point of sale, in addition to disclosures, has an obviously beneficial impact on the accuracy of the assessment made. Having a more accurate price quoted at point of sale, facilitated by such information, also helps to remove the price shock common in the traditional process as new risk information emerges during the underwriting process.

The second avenue is not yet common but is also a logical step: our phones and health watches contain a wealth of valuable information that can provide insights for insurers. If we can utilise this information in a more immediate way – naturally, with the customer's consent - we can make the journey much easier for everybody involved in the transaction, but most importantly for the customer themselves.

In the best of all possible worlds, we could, for example, be able to request fewer medical examinations or other reports if much of that information can be reliably extracted from the phone, again saving time and hassle for medical professionals. Time savings are just one key element of the societal benefits of digitalisation, but a more efficient process also means more people can get insurance cover, thereby helping society further by helping close the protection gap.

We have long known that health questionnaire wordings and even supplementary medical evidence should be driven by more than simply age and the amount assured. We have been prevented from using this knowledge by the need for simplicity for the agent, who needs to know immediately what additional requirements might be necessary and how to manage customer expectations accordingly.

As more and more sales are online or at least digitised, we can start to match the underwriting hurdles more intelligently to each risk. We should always be asking ourselves important questions, like whether we can more effectively define what we ask of applicants by incorporating factors such as pre-test probability of disease.

By doing this, it will save operational costs - and mean what we do spend is more likely to get a result, and at the same time deliver a faster journey for many.

It will also help us to cut down on asking pointless questions of those at very low risk of what is being asked about, such as dementia risk at young ages, or hazardous sports at older ages. Having an automated triage and steering capability will help to deliver this in the optimal manner, meaning more people seeking insurance benefit from a speedier journey, without the usual cost consequences of such simplification. 

Many of the tools and information we use to assess risk at the point of sale could also be used to help people improve their health, either through chronic disease management or via incentivisation to adopt a healthier lifestyle.

Because such a step would benefit both insurer and customer, it holds huge potential. Regular manual assessments to monitor a person's health will never prove cost effective; consequently, this post-issue review of medical factors is an area where automation is almost a requirement to deliver the true benefits.

The instant rewards for the customer made possible through automation would also better incentivise further engagement.

It should be clear now that the original rules engines, designed to serve the primary purpose of quantifying customer disclosures, will need to evolve to cater to additional needs. What is common to most of them is the need for insurers to distance themselves from the notion that automation is merely a stand-alone solution focused primarily on efficiency.

Above everything, however, maybe it is also time to think about the entire set of principles behind automation. Instead of just replicating manual outcomes via systems, shouldn't we be using the power of automation to do an even better job than the manual processes that are being replaced? That surely must be our goal.

But the digital world never stands still, and ever improving techniques, particularly those facilitated by artificial intelligence and machine learning, will expand the capabilities even further. In particular, the ability to take underwriting engines beyond the still mostly rules based approach holds massive potential to further revolutionise the automated underwriting journey.

Once you accept that automation has even greater potential than merely doing what we have always done more quickly, maybe there is also on opportunity to go back to the roots and revisit how and what we ask clients at the point of sale.

There are situations where we'll be able to ease up on asking for details that the customer will almost never have, and in return increase the likelihood that they will receive a decision more quickly. Watch this space.