Optimizing UBI at Scale

Written by David Lukens

It’s been a wild ride for everyone, but particularly the insurance industry, for the past several years. As people start to see a potential of post-COVID times or at least an endemic situation which is more stable and manageable, massive changes are beginning to appear in the lifestyles of policyholders. These changes will profoundly impact driving habits and patterns as compared to the same behaviors in a pre-COVID world.

During the initial COVID shutdowns and quarantines, when policyholders stopped or greatly reduced driving, started telecommuting on a massive scale, and intentionally shrank individual geographical footprints, there was a corresponding reduction in claims frequency and losses. Now that things are getting back to “normal,” the pendulum is swinging wildly in the other direction, as people are driving more, faster, and worse, and as supply chain challenges continue to drive up parts and vehicle prices.

While everyone went into lockdown and experienced reduced driving in largely the same way, recovery or a return to normal looks different for everyone. Or, as Tolstoy wrote, “All happy families are alike; every unhappy family is unhappy in its own way.” Undoubtedly, the overall loss trends post-COVID look bad on a macro level, but there are pockets of positivity as individual policyholders have changed jobs or changed commuting patterns due to new flexibility and work-from-home (WFH) policies.

What does this mean? It means that understanding the post-COVID driving patterns of policyholders both individually and en masse is critical to insurers returning to profitability in this changed world.

The need to understand current driving patterns of policyholders is intensified by the fact that traditional rating attributes – age, gender, marital status, credit, prior accident histories, continuous coverage, etc. – are all in flux as compared to pre-pandemic baselines. The best way to understand how individual policyholders’ mobility patterns have changed post-COVID is by getting real-time or near real-time data about when, how, and how much policyholders are now driving.

Insurers positioned to understand these dynamics of insureds, and specifically at scale, are going to be far better positioned for challenges in a post-COVID world. These same insurers will also be more nimble and able to adjust prices tactically through the current and future hard market than those insurers without this capability (or without the ability to scale) and still operating under the assumption that legacy pricing attributes still mean the same thing post-COVID as before.

Why is Scale Important?

Except for a handful of usage-based insurance (UBI) programs, the majority of insurers are still focused on standing up voluntary, opt-in only programs that account for a small portion of any given book of business. While this is fine for understanding how driving patterns impact risk and getting the benefit of self-selection improvements in loss performance for those drivers who choose to participate, this approach does not help understand the true risk profile of an insurer’s total book of business.

If you have read the book “21” or seen the film about a group of MIT students who counted cards playing blackjack in Vegas, it provides a good analogy for this. This approach to risk selection (which is similar in some ways to insurance pricing) doesn’t work when only five or six hands are played. Simply put, there isn’t the volume of cards required to tip the odds. It has to be done at scale to see the benefits. In the same way, UBI programs that only address a small portion of an insurer’s book is like counting cards for only a couple of hands. It simply isn’t optimized unless you go (pardon the pun) “all in.”

Keeping UBI confined to the risks opting-in to the programs when it is beneficial (and opting out when it is not) really narrows an insurer’s an ability to understand which drivers are coming out of COVID in an adverse position. This is because the only visibility available to insurers is into the “good” risks, which arguably have less segmentation opportunity than those risks coming out of COVID and choosing not to participate. Programs that are extended to cover a much wider swath of an insurer’s book (or an entire book), not only allow for better understanding of changing risks, but also open the door to other valuable use cases, the most important of which is the use of driving data to adjudicate and settle claims more quickly.

It is not operationally efficient for claims organizations to retool or change processes for new data that is only available on and applicable to a very small subset of policyholders, especially if that subset of policyholders are the lowest-frequency customers with the fewest claims. Anecdotally, customers who have opted into UBI programs traditionally exhibit very, very low loss frequencies. This dynamic minimizes the impact UBI data can have on the claims process. In order to realize the value of UBI data, it must be available on a significant portion of claims, or, ideally, all claims, and able to be applied at scale. At this point, it becomes valuable enough to warrant investment in changes to the claims processes to incorporate the data into the claims workflow and realize the benefits to reduced cycle times and severities.

What Does the Future Look Like?

Many multi-line insurance companies in the UK and Europe are already incorporating UBI functionality into corporate apps which are used daily to make payments, check status, and facilitate communication between an insurer and an insured. In many instances, this naturally leads to cross-sell opportunities. And, by incorporating this functionality into a corporate app, the insurer is then well positioned to offer claims support and other services, like rewards programs, to policyholders for about the same annual cost as a motor vehicle report (MVR).

Coming out of COVID, there has never been an environment in which understanding individual policyholder’s driving patterns has been more important. But, this knowledge is not optimized until insurers have visibility into the majority of policyholders’ portfolios at very large scales, scales which also support the largely unrealized claims use case.


David Lukens can be reached for further comment or information via email at dlukens@ims.tech.