Tag Archives: Tractable

Some Respect for Insurance Innovators

The insurance industry is good at beating itself up for not innovating faster — and many analysts and customers are all too happy to join in — so it was a welcome surprise to see this article last week in the New York Times, enthusiastically describing a nearly frictionless future for auto claims.

The article quotes an executive as saying, “In the near future,… we’re going to take the [auto claims] process from days or weeks to minutes.”

When’s the last time you saw such a glowing statement about insurers in a national, non-insurance publication? Doesn’t it feel good?

The Times article said the most aggressive innovation is happening outside the U.S. because of the complications caused by state-by-state regulation in the U.S. The piece singled out Tractable, in London, as a leader in the technology that lets customers take pictures of their damaged cars and have artificial intelligence instantly generate an accurate estimate — cutting days out of the process by removing the need for an appointment with a human adjuster and for extensive back-and-forth between the adjuster and the repair shop. The piece cited Admiral Seguro, an insurer in Spain, as an early implementer of the technology; it may even offer payment of a claim within minutes.

The Times said the innovations will move into production in the U.S. soon — there are lots of pilots, but a human appraiser currently still double-checks the AI, acting as a sort of safety driver. The pandemic is speeding up the transition because it discourages face-to-face interactions.

I’m not saying that what the Times wrote is news to those in the industry. At ITL, for instance, we’ve been closely following the trend toward a fully digitized auto claims process, including this piece just last week on automating payments. Early this summer, this well-researched piece explained how innovations are “producing almost touchless processes for claims and enabling cycle times of seconds, minutes and hours instead of days and weeks.” A bit later, this article laid out how COVID-19 was accelerating the trend.

What I am saying is that we should all feel good that the industry’s efforts to innovate are being noticed by the general population.

Of course, we now have to redouble our efforts. After all, everyone’s expectations have just been raised.

Stay safe.

Paul

P.S. Here are the six articles I’d like to highlight from the past week:

Digital Future of Insurance Emerges

Patterns are emerging out of the fog of this pandemic and paint a clear view of the future of insurance, leaving only the timing uncertain.

AI in Commercial Underwriting

Machine learning and AI are incredibly well suited for helping to deal with the masses of data that underwriters now face. Here are five keys.

Selling Where Life Happens

What if life insurers reinvented themselves to be like retailers, obsessed with the point of purchase – digitally and in-person?

How to Minimize Flood Losses

Flood warnings have two weaknesses: lack of detailed information on the precise locations at risk of flood and too many false alarms.

5 Safety Keys for COVID-Era Building

Construction crews need continual training in the best of times, and the evolving challenges of COVID-19 heighten the need.

A Study on Expanded Use of ‘Presumption’

A shift on the presumption of coverage for COVID-19 under workers’ comp risks undermining the Grand Bargain.

5 Challenges When Innovating With AI

Artificial intelligence is booming in insurance. In a recent report, Celent identified AI use cases around the globe and across the insurance value chain.

Uses include customer engagement (USAA’s Nina); product optimization (Celina Insurance Group, Protektr); marketing and sales (Usecover, Insurify, Optimal Global Health, Ping An); underwriting (ZestFinance, SynerScope, Intellect SEEC, Swiss Re); claims (Tractable, Ant Financial, Gaffey Healthcare); fraud detection (Ant Financial, USAA); risk management (Achemea); and business operations (Ping An Direct, Union Life).

Insurers are wise to innovate with AI technologies. Early adopters will face challenges but will also have the potential to reap greater rewards by improving efficiency and customer engagement.

Here are five challenges for carriers to consider when innovating with AI:

1. What technology to use when. When embarking on a digital transformation, there may be a number of solutions available for a given problem, one of which could be AI. But while AI may resolve an issue, it is important to examine all the potential solutions and decide which one is the best fit. Perhaps robotic process automation (RPA), application programming interface (API) or another automated solution is best suited. Can an existing technology be leveraged?

Deciding what solution to apply when requires you to look at the whole organization and all the issues upfront. This allows CIOs and CEOs to examine each problem, decide on the right technology solution and make sure it complements the overall strategy and budget.

See also: Strategist’s Guide to Artificial Intelligence  

2. Big data + AI = big strategy. A second challenge surrounds the management of big data obtained from customers, core systems, brokers/agents and insurance exchanges. Add to that the varied types of data that AI is managing, analyzing, communicating and learning from and things get a little more complicated. Here’s a list of the different data types AI may be working with:

  • Structured, semi-structured and unstructured data
  • Text
  • Voice
  • Video
  • Images
  • Sensors (IoT)
  • Augmented/virtual reality

Data is also classified as real-time, historical or third-party — yet another dimension to consider. Make sure your strategy takes the necessary data variables into account: where data will come from, where it will flow to and how it will be handled at various points in the customer journey.

3. Managing customers across swim lanes. This leads us to challenge No. 3: the ability of AI to engage with customer data at key touchpoints during the customer lifecycle. For example, if Lucy has group benefits as well as voluntary products, car and house insurance, how will her data be managed and optimized across swim lanes?

What will be the touch points for AI? When will other insurtech solutions be present? When is human intervention required? And how will this data be used to inform future risk decisions?

4. Harnessing AI’s multidisciplinary capabilities. AI encompasses machine learning, deep learning, natural-language processing, robotics and cognitive computing, to name a few. You can read my blog post here to learn more. Deciding what technical abilities will be required to solve your problem could present challenges as the lines between disciplines blur.

Additionally, the next wave of AI could come from entirely different industries, such as aerospace, environmental science or health — but  it will still have applications for insurance. The best way to overcome this is to examine your AI needs across solutions and select vendors with the right capabilities to execute them.

See also: The Insurer of the Future – Part 3  

5. Communicating past tech speak. As AI becomes mainstream, the challenge of helping non-technical business professionals understand these complex applications is real. AI systems can require a level of technical expertise beyond the everyday scope of business.

True digital transformation, regardless of technical complexity, affects everyone in the organization. Ensuring the vision is shared will matter as day to-day operations, tasks and activities change. Find someone who can break down the benefits of these new solutions into bite-sized pieces that everyone understands to ensure buy-in and ultimate success.

The question of whether AI will indeed disrupt the industry or simply enable its full digitization is still not known. It will not be the solution to every problem. However, if implemented strategically, it may hold the capacity to create an entirely new way of insuring — and delighting — customers in a rapidly changing landscape.

Insurtech Is Ignoring 2/3 of Opportunity

Fifty-six cents of every premium dollar is indemnity (loss costs). A further 12 cents is needed to assess, value and pay those losses. Given that two-thirds of the insurance industry economics are tied up in losses, it would be logical that much of the innovation we are now witnessing should focus on driving down loss costs and loss adjustment expense — as opposed to the apparent insurtech focus on distribution (and, to a lesser extent, underwriting).

This is beginning to happen.

What do you have to believe for loss costs and adjustment expenses to be a prime area of innovation and disruption? You have to believe that the process (and, thus, the costs) to assess, value and pay losses is inefficient. You have to believe that you can eliminate the portion of loss costs associated with fraud (by some estimates, as much as 20%). You have to believe that there is a correct amount for a loss or injury that is lower than the outcomes achieved today, particularly once a legal process is started. You have to believe that economic improvements can happen even as customer experience improves. And you have to believe that loss costs and adjustment expenses can decline in a world in which sensor technology starts to dramatically reduce frequency of losses and manufacturers embed insurance and maintenance into their “smart” products.

See also: ‘Digital’ Needs a Personal Touch  

Having spent years as an operating executive in the industry, I happen to believe all of the above, and I am excited by the claims innovation that is just now becoming visible and pulling all of the potential levers.

We are seeing an impact on nearly all aspect of the claims resolution value chain. Take a low-complexity property loss. Technology such as webchat, video calls, online claims reporting and customer picture upload are all changing the customer experience. While the technologies aren’t having a huge impact on loss adjustment or loss costs, they are having profound impact on how claims are subsequently processed and handled.

One such example, as many have heard, is how Lemonade uses its claims bot for intake, triage and then claims handling for renters insurance. Lemonade’s average claim is a self-reported roughly $1,200 (low value), and only 27% are handled in the moment via a bot as opposed to being passed to a human for subsequent assessment. Still, Lemonade certainly provides a window to the future. Lemonade is clearly attacking the loss-adjustment expense for those claims where it believes an actual loss has occurred and for which it can quickly determine the replacement value.

More broadly, Lemonade is a window into how many are starting to use AI, machine learning and advanced analytics in claims in the First Notice of Loss (FNOL)/triage process — determining complexity, assessing fraud, determining potential for subrogation and guiding the customer to the most efficient and effective treatment.

While Lemonade is the example many talk about, AI companies such as infinilytics and Carpe Data are delivering solutions focused specifically on identifying valid claims that can be expedited and on identifying those claims that are more questionable and require a different type of treatment. These types of solutions are beginning to deliver improvement in both property and casualty. New data service providers — such as Understory, which provides single-location precision weather reports — can be used to identify a potential claim before even being notified, which can reduce loss costs through early intervention or provide reference data for potentially fraudulent claims.

Equally interesting is the amount of innovation and development appearing in the core loss-adjusting process. Historically, a property claim — regardless of complexity — would be assessed via a field adjuster who evaluates and estimates the loss. Deploying technical people in the field can be very effective, but it is obviously costly, and there is some variability in quality.

In a very short time, there are very interesting new models emerging that reimagine the way insurers handle claims.

Snapsheet is providing an outsourced solution that enables a claimant of its insurance company customers to use a service that is white-labeled for clients. The service enables the claimant to take pictures of physical damage, which is then “desk adjusted” to make a final determination of the value of the claim, followed by a rapid and efficient payment.

WeGoLook, majority-owned by claims services company Crawford & Co, is using a sophisticated crowd-sourced and mobile technology solution to rapidly respond to loss events with a “Looker” (agent) who can perform a guided process of field investigation and enable downstream desk adjusting process, as well.

Tractable provides artificial intelligence that takes images of damaged autos and estimates value (effectively a step toward automatic adjudicating). Tractable — like, Snapsheet and WeGoLook — has made great strides. Aegis, a European motor insurer, is rolling out Tractable following a successful pilot. In each of these instances, the process is much improved for customers — whether it be self-serving because they choose to do so (Snapsheet), rapidly responding to the event (WeGoLook) or dramatically reducing the cycle time (Tractable). All provide material improvements in customer experience.

See also: Waves of Change in Digital Expectations  

Obviously, each of these models is attacking the loss adjustment expense — whether through a more consistently controlled process of adjusting at a desk, using AI to better assess parts replacement vs. repair or improving subrogation, among other potential levers.

Today, all of these solutions are rather independent of each other and generally address a low-complexity property loss (mostly in the auto segment), but the possible combination of these and other solutions (and how they are used depending on type and complexity of claims) could begin to amplify the impact of technology innovation in claims.

digital innovation

The 7 Colors of Digital Innovation

InsurTech is now established in a class of its own, no longer a sub category of Fintech. In 2015, $2.65 billion of venture capital was invested in InsurTech. We now have InsurTech-focused accelerators, with the excellent Startupbootcamp in London, the Global Insurance Accelerator in Des Moines, Iowa, (about to start its second cohort) and Mundi Lab announcing its start-ups for its insurance program in Madrid.

In the past year, I have interviewed more than 50 InsurTech start-ups, and I have seen the full spectrum of characteristics and common themes that run through these innovative digital insurance businesses, which i call:

From Distribution to Data, the Spectrum of InsurTech

Red – Distribution

Distribution is all about making insurance easier to buy, consume and understand. Innovators put the customer first and build their insurance proposition from the customer out (unlike incumbents, which organize their business around internal capabilities).

These start-ups are all about the customer, and their propositions are characterized by convenience, on-demand, personalization and transparency (and, of course, digital).

Examples include;

  • Bought by Many
  • Knip
  • Cuvva
  • Insquik
  • PolicyGenius
  • Moneymeets

Orange – Enterprise

Here we see a new breed of enterprise-class software providers. These are software as a service platforms running on the cloud. They have consumption-based pricing models that replace the traditional, million-dollar, up-front license fee and multi-year implementation.

In the main, these InsurTechs have taken hold of the small and mediums-sized business (SMB) space, but it is a matter of time before they prove themselves as genuine enterprise solutions for Tier 1 insurers.

Examples include:

  • Vlocity
  • Zenefits
  • Insly
  • Surely
  • Riskmatch

Yellow – Mutual 

New peer-to-peer business models return insurance to its roots of mutualization and community. The model relies on the notion that social grouping and affinity will change behavior and address moral hazard (thereby reducing claims payouts and premiums).

The question of scalability still hangs over P2P insurance, but, if it succeeds as a business model, it could form the foundation of a new breed of insurer. Just as kids call to their parents in their hour of need, customers will call to the insurer in theirs.

Examples include:

  • Friendsurance
  • Guevara
  • TongJuBao
  • Lemonade
  • Uvamo
  • Gaggel

Green – Consensus

Blockchain technology will fundamentally change the way the insurance industry works (as well as banking and society as a whole, IMHO).

The promise is huge although as yet unproven. From smart contracts to identity authentication, from fraud prevention to claims management, blockchain technology will provide the underlying technology foundations for a trustless consensus that is transparent to all parties.

Examples include:

  • Everledger
  • Tradle
  • SmartContract
  • Dynamis
  • Blockverify

Blue – Engagement

For me, this is the most significant of the characteristics from InsurTech in personal lines. The product becomes integrated in the customer’s lifestyle. It becomes sticky and overrides the annual buying exercise, where price is the key buying criterion. Digital natives are responding well to lifestyle apps that sit on top of the underlying insurance product.

Examples include:

  • Vitality
  • Trov
  • Oscar

Indigo – Experience

The true value of insurance is only realized when the customer makes a claim. New tech solutions that improve the customer journey through the claims process will not only improve the customer experience, they will also reduce the cost of claims and claims payouts.

Examples include:

  • 360Globalnet
  • RightIndem
  • Tractable
  • Vis.io
  • Roundcube

Violet – Data

This is all about new sources of data to rate and underwrite risk. This is about using data science, machine learning, artificial intelligence and high-performance computing to process data in completely new ways.

While distribution is vital to change the way customers interact with insurers, it is the data players that hold the key to fundamental change in the way insurance is manufactured, especially in personalisztion of insurance premiums and policies.

Examples include:

  • Quantemplate
  • Analyze Re
  • Meteo Protect
  • The Floow
  • Fitsense
  • Influmetrics
  • RiskGenius

How to Use All the New Data

Most people who purchase an insurance policy are faced with the daunting task of filling out an extensive application. The insurance company – either directly or through an intermediary – asks a myriad of questions about the “risk” for which insurance is being sought. The data requested includes information about the entity seeking to purchase insurance, the nature of the risk, prior loss experience and the amount of coverage requested. Insurers may supplement that information with a limited amount of external data such as motor vehicle records and credit scores. The majority of information used to inform the valuation process, however, has been provided by the applicant. This approach is much like turning off your satellite and data-driven GPS navigation system to ask a local for directions.

According to the EMC Digital Universe with research and analysis by IDC in 2014, the digital universe is “doubling in size every two years, and by 2020 the digital universe – the data we create and copy annually – will reach 44 zettabytes.” That explosion in the information ecosystem expands the data potentially available to insurers and the value they can provide to their clients. But it requires new analytical tools and approaches to unlock the value. The resulting benefits can be grouped generally into two categories:

  • Providing Risk Insights: Mining a wider variety of data sources yields valuable risk insights more quickly
  • Improving Customer Experience: Improving the origination policy service and claims processes through technology enhances client satisfaction

For each of these areas, I’ll highlight a vision for a better client value proposition, identify some of the foundational work that is used to deliver that value and flesh out some of the tools needed to realize this potential.

Risk Insights
Insurance professionals have expertise that gives them insight into the core drivers of risk. From there, they have the opportunity to identify existing data that will help them understand the evolving risk landscape or identify data that could be captured with today’s technology. One can see the potential value of coupling an insurer’s own data with that from various currently available sources:

  • Research findings from universities are almost universally available digitally, and these can provide deep insights into risk.
  • Publicly available data on marine vessel position can be used to provide valuable insights to shippers regarding potentially hazardous routes and ports, from both a hull and cargo perspective.
  • Satellite imagery can be used to assess everything from damage after a storm to proximity of other structures to the ground water levels, providing a wealth of insights into risk.

The list of potential sources is impressive, limited in some sense only by our imagination.

When using the broad digital landscape to understand risk — say, exposure to a potentially harmful chemical — we know that two important aspects to consider are scientific evidence and the legal landscape. Historically, insurers would have relied on expert judgment to assess these risks, but in a world where court proceedings and academic literature are both digitized, we can do better, using analytical approaches that move beyond those generally employed.

Praedicat is a company doing pioneering work in this field that is deriving deep insights by systematically and electronically evaluating evidence from various sources. According to the CEO Dr. Robert Reville, “Our success did not come solely from our ability to mine data bases and create meta data, which many companies today can do. While that work was complex, given the myriad of text-based data sources, others could have done that work. What we do that is unique is overlay an underlying model of the evolution of science, the legal process and the dynamics of litigation that we created from the domain expertise of our experts to provide context that allows us to create useful information from that data built to convert the metadata into quantitative risk metrics ready to guide decisions.”

The key point is that if the insurance industry wants to generate insights of value to clients, identifying or creating valuable data sources is necessary, but making sense of it all requires a mental model to provide relevance to the data. The work of Praedicat, and others like it, should not stop on the underwriter’s desktop. One underexploited value of the insurance industry is to provide insights into risk that gives clients the ability to fundamentally change their own destiny. Accordingly, advances in analytics enable a deeper value proposition for those insurers willing to take the leap.

Customer Experience
Requiring clients to provide copious amounts of application data in this information age is unnecessary and burdensome. I contrast the experience of many insurance purchasers with my own experience as a credit card customer. I, like thousands of other consumers, routinely receive “preapproved” offers in the mail from credit card companies soliciting my business. However appealing it may be to interpret this phenomenon as a benevolent gesture of trust, I know I have found myself on the receiving end of a lending process whereby banks efficiently employ available data ecosystems to gather insights that allow the assessment of risk without ever needing to ask me a single question before extending an offer. I contrast this with my experience as an insurance purchaser, where I fill out lengthy applications, providing information that could be gained from readily available government data, satellite imagery or a litany of other sources.

Imagine a time when much of the insurance buying process is inverted, beginning with an offer for coverage, rather than a lengthy application and quote request. In that future, an insurer provides both an assessment of the risks faced, mitigations that could be undertaken (and the savings associated), along with the price it would charge.

While no doubt more client-friendly, is such a structure possible? As Louis Bode, former senior enterprise architect and solution architect manager at Great American Insurance group and current CSO of a new startup in stealth-mode observes, “The insurance industry will be challenged to assimilate and digest the fire hose of big data needed to achieve ease of use and more powerful data analytics.”

According to Bode, “Two elements that will be most important for us as an industry will be to 1) ensure our data is good through a process of dynamic data scoring; and 2) utilize algorithmic risk determination to break down the large amounts of data into meaningful granular risk indexes.” Bode predicts “a future where insurers will be able to underwrite policies more easily, more quickly and with less human touch than ever imagined.”

The potential to use a broader array of data sources to improve customer experience extends well beyond the origination process. Imagine crowdsourcing in real time the analysis of images to an area affected by a natural disaster, getting real time insights into where to send adjusters before a claim is submitted. Tomnod is already crowdsourcing the kinds of analysis that would make this possible. Or imagine being able to settle an automobile claim by simply snapping a picture and getting an estimate in real time. Tractable is already enabling that enhanced level of customer experience.

The future for insurance clients is bright. Data and analytics will enable insurers to deliver more value to clients, not for additional fees, but as a fundamental part of the value they provide. Clients can, and should, demand more from their insurance experience. Current players will deliver or be replaced by those who can.

I’d like to finish with a brief, three-question poll to see how well readers think the industry is performing in its delivery of value through data and analytics to clients. Here is my google forms survey.