Tag Archives: artificial intelligence

Is Insurtech a Game Changer? It Sure Is

Several years ago, property and casualty insurance executives were looking over their shoulders anxiously at a growing number of internet startups. Who were these scruffy people wearing black turtlenecks? Could they really “disintermediate” legacy providers that had been around for a century or more?

Since then, we’ve all evolved. By now, most brands know they have inherent strengths that are hard to dislodge. The startups have matured, too, and they clearly have something to offer the market. We’re now working with companies in both camps, helping them navigate this new normal, where collaboration, acquisition and competition are all plausible options.

Some insurers may think they’ve dodged a bullet. But insurtech’s threat is more stealthy, and no less powerful.

Insurtech: the new, new thing?

At this fall’s InsureTech Connect trade show, literally thousands of people descended on Las Vegas to show and examine the latest offerings, from core systems, predictive analytics tools and anything-as-a-service to pitches addressing distribution, pursuing unserved niche markets, offering comparative pricing and broker services and more.

In our recent report on the state of insurtech, we cautioned insurers to look beyond the many truly interesting offerings now coming to market. As impressive as these tools are, we urged decision makers to stay focused on the capabilities that make their companies unique.

See also: Has a New Insurtech Theme Emerged?  

What do insurers really do?

So, what are those capabilities? At holiday dinner tables, you may find yourself talking to a relative about what insurance is, and why it’s important. You may say something like, “We create products that help manage risk by sharing the possibility of individual loss with a larger pool of users.” This explanation held true for a long time, but,
with the rise of insurtech, it may not be the best way to look at your business.

That’s because many insurtech companies have emerged to manage the firehose of data that now shapes our world: the Internet of Things (IoT), wearable health devices, connected cars, artificial intelligence and more. Of course, there’s still a role for insurers when someone else captures and gets the insight from that data. But it’s a commodity role, driven by who is willing to write a policy to offset the risk at the lowest rate. There won’t be many winners, and the margins won’t be attractive.

Some insurers see their business as settling claims and handing out checks. But when someone else is using telematics to assess driving habits, or social media to understand lifestyle risks, who will be able to monetize this data? Increasingly, underwriting depends on getting deep into the data-driven weeds. If you’re not there, recognize that someone else will be.

The rise of outside money

There’s another factor shaping insurance today: the amount of private equity (PE) and venture capital (VC) money flooding into the industry. An industry as highly capitalized as insurance was bound to have external investors come knocking eventually. Now, they have.

To be blunt, many insurance systems are too costly and too slow. PE and VC firms have seen this, and they’ve said to themselves, “I don’t have to be perfect, and I know I can be more efficient than this. Even if I’m only a little bit better than the legacy players, I can make a very healthy profit.” It’s a form of arbitrage, and competition could soon get a lot tougher.

With the acceleration of insurtech and related technologies such as cloud and artificial intelligence, PE and VC firms have found a way in that doesn’t require them to show a century of stability. They can do very well developing an insurtech play for very specific aspects of the P&C value chain. Many traditional companies are finding themselves in a commoditized business, without the structure of a commodity manufacturer.

Finding your way to play

Some of the most exciting developments in technology are now reshaping the insurance industry. That spells new opportunities and new risks. With the rise of PE and VC funding, we now see competition emerging from companies with significant resources—and they’re privately held so they can be more patient investors.

See also: Advice for Aspiring Leaders in Insurtech  

Legacy insurance companies still have enormous advantages, and many opportunities to win. But most won’t be able to do it alone, and there are many examples of insurers that wasted time (and money) on the wrong insurtech acquisition or partnership. As the cycles of innovation and capital movement accelerate, you’ll need to be more focused than ever on the capabilities that make your company great. Insurtech is a game-changer.  Make sure you’re playing the right game.

3 Big Challenges on the Way to Nirvana

We hear almost daily how insurtech is disrupting the once-staid insurance industry. The main ingredients are big data, artificial intelligence, social media, chatbots, the Internet of Things and wearables. The industry is responding to changing markets, technology, legislation and new insurance regulation.

I believe insurtech is more collaborative than disruptive. There are many ways insurance technology can streamline and improve current processes with digital transformation. Cognitive computing, a technology that is designed to mimic human intelligence, will have an immense impact. The 2016 IBM Institute for Business Value survey revealed that 90% of outperforming insurers say they believe cognitive technologies will have a big effect on their revenue models.

The ability of cognitive technologies, including artificial intelligence, to handle structured and unstructured data in meaningful ways will create entirely new business processes and operations. Already, chatbots like Alegeus’s “Emma,” a virtual assistant that can answer questions about FSAs, HSAs and HRAs, and USAA’s “Nina” are at work helping policyholders. These technologies aim to promote not hamper progress, but strategies for assimilating these new “employees” into operations will be essential to their success.
Managing the flood of data is another major challenge. Using all sorts of data in new, creative ways underlies insurtech. Big data is enormous and growing in bulk every day. Wearables, for instance, are providing health insurers with valuable data. Insurers will need to adopt best practices to use data for quoting individual and group policies, setting premiums, reducing fraud and targeting key markets.

See also: Has a New Insurtech Theme Emerged?  

Innovative ways to use data are already transforming the way carriers are doing business. One example is how blocks of group insurance business are rated. Normally, census data for each employee group must be imported by the insurer to rate and quote, but that’s changing. Now, groups of clients can be blocked together based on shared business factors and then rated and quoted by the experience of the group for more accurate and flexible rating.

Cognitive computing can also make big data manageable. Ensuring IT goals link back to business strategy will help keep projects focused. But simply getting started is probably the most important thing.

With cognitive computing, systems require time to build their capacity to handle scenarios and situations. In essence, systems will have to evolve through learning to a level of intelligence that will support more complex business functions.

Establishing effective data exchange standards also remains a big challenge. Data exchange standards should encompass data aggregation, format and translation and frequency of delivery.
Without standards, chaos can develop, and costs can ratchet up. Although there has been traction in the property and casualty industry with ACORD standards, data-exchange standards for group insurance have not become universal.

See also: Insurtech’s Approach to the Gig Economy  

The future is bright for insurers that place value on innovating with digital technologies and define best practices around their use. It’s no longer a matter of when insurance carriers will begin to use cognitive computing, big data and data standards, but how.

Finally Realizing the Promise of AI

It’s almost inevitable. Spend your working life identifying, analyzing, quantifying and ascribing monetary value to risk, and you’re likely to have a fairly strong aversion to it. More accurately, an aversion to undertaking new endeavors with inadequately understood consequences. The insurance industry is, on any number of levels, the very definition of risk-averse.

Yet, for all the commentary suggesting otherwise, insurance still has an appetite for innovation. If the insurtech sector is any indication, then an interest in and requirement for new solutions is being recognized and slowly addressed.

Insurance may not employ the language of disruption that runs through the wider fintech market, may be short a few unicorns and may be unable to boast some of the record-breaking funding rounds, but a quiet tech evolution has been building in insurance, nonetheless. Hence the advent of automated underwriting facilitated by more advanced algorithms and data analysis.

Where insurtech does overlap with its more vocal fintech counterparts is in the greater use of artificial intelligence (AI) and machine learning to solve age-old problems around data analysis and interpretation.

It’s about five years or so since AI first became a topic of conversation in insurance. Since then, despite the intensity of the debate, it has often felt like a reality that is always just over the horizon – a destination that kept moving even as more and more efforts were directed toward it.

But recent research suggests that the journeys made so far have not been in vain. We are at a point where embracing AI is about to step up a gear. The global value of insurance premiums underwritten by AI has reached an estimated $1.3 billion this year, as stated by Juniper Research; and they are expected to top $20 billion in the next five years. As a destination, AI is closer and more attainable than ever before.

See also: Untapped Potential of Artificial Intelligence  

However, AI is not an island. Its promise of $2.3 billion in global cost savings to be achieved through greater efficiencies and automation of resource-intensive tasks will not be achieved in isolation.

AI remains part of a more complex ecosystem of data gathering and analysis. It can apply new technologies to get the best out of the already established and still-emerging data sources that feature in underwriting offices around the world. It emphatically does not require these existing investments to be ripped out, replaced or downgraded.

It is more helpful, therefore, to see AI as the differentiating factor in the latest generation of insurance IT: augmented automated underwriting, or AAU for short.

AAU lets underwriters spot patterns and connections that are, frankly, either invisible to the human eye or that take normal, human-assisted processes unfeasible amounts of time and resource to identify.

Whereas earlier generations of automation were able to pick up the low-hanging fruit of insurance markets – the individuals whose driving history fit into clearly delineated boxes, for example – AAU can take into account all of the rich complexity of the human experience. It can spot the nuances and individualities that populate the life market, for example, and translate those into accurate policies.

That’s good news for both underwriters and their customers. AAU can significantly reduce the need for separate medicals, repeated questions, and lengthy decision-making processes and drastically increase the speed at which a potential insurer can get a quote and cover – while continually improving the way risk is calculated and managed.

AAU can make sure the decision-making process remains in the hands of underwriters rather than IT departments, enabling them to set and update the rules and parameters as befits their preferred business model. It consequently makes advanced, complex and precise decision-making available to a broader range of underwriting businesses – which is good for those businesses, good for customers and ultimately good for the entire industry.

See also: Strategist’s Guide to Artificial Intelligence

AAU – augmented automated underwriting – is an example of the realization of AI’s promise. As such, it’s set to become one of the key talking points and disruptive technologies of the insurance industry. And this time, AAU is both a journey and destination that all progressive insurance organizations need to be considering.

How AI Changes Everything

Insurers already collect heaps of data; with artificial intelligence, they can use it to its full potential and improve at every level, from automating call center request processing to helping make accurate assessments and executive-level decisions. Through its power to recognize patterns and anticipate actions, AI can provide a predictive environment where risks are anticipated and hedged.

So far, it seems that the main areas of AI application in insurance include customer experience (58%), process optimization (43%) and product innovation (19%), according to a 2018 study by Everest Global.

AI could also be applied to fraud, which a report by the FBI shows costs more than $40 billion per year. The insurance industry is prone to multiple fraud schemes. However, most fraudsters follow well-known patterns that AI can identify in minutes. The need to address fraud will become even more critical in coming years when more policies will be issued and handled exclusively online, and AI will be able to highlight any abnormal patterns during claims.

See also: Untapped Potential of Artificial Intelligence  

Claim Support

Right now, the insurance sector requires a lot of staff for processing and inspecting claims. This makes policies more expensive and case-solving more cumbersome. These manual tasks can be at least partially replaced with a chatbot to record the claim, verify the details, make sure there is not a fraud attempt and pass the claim along. Through computer vision, the chatbot can also analyze the evidence and assess the damage.

Better Underwriting

The complaint about insurance policies is typically their price. AI can create personalize rates based on the client’s actual choices and lifestyle. Factors such as the distance traveled, diseases, financial stability and more can create dynamically priced policies. Metromile, for instance, uses an AI-enhanced sensor system to monitor the driver’s behavior and any incidents.

A similar solution could be created for life insurance based on data provided by fitness trackers and medical records. This is called behavioral premium pricing, and it’s about paying for what risks you take. You are no longer just a data point in a statistic; you are paying for your actions. It’s not about approximating but about taking responsibility.

Better Marketing and Customer Experience

An enhanced online profiling tool can help insurance companies tailor products for a wide array of client segments.

Using natural language processing and scanning comments from online platforms and forums can lead to the creation of innovative insurance products that are more adapted to the modern client’s needs. It’s listening to the client’s voice but in a new, AI-powered way.

Customer experience is all about speed and reliability. The time to settle a claim is a key performance metric, and it can in many cases be reduced from days to minutes.

The next step of using AI is not so much about innovating but about integrating. We have different services providing us with insurance for health, car and home, but we can hope to see universal insurance models that are customized to the client’s needs and priced dynamically according to the perceived risk.

A McKinsey report, Insurance 2030, describes a fictional client who uses his digitally powered AI assistant to do some daily chores, while insurance premium is computed on the go, based on the client’s decisions and lifestyle.

People will become more careful and try to prevent claims instead of repairing or treating. Until now, joining an insurance pool was about sharing risks, but AI is making us more responsible for our own actions.

Is the Public Ready?

Any new technology is just as good as the adoption rate. The good news is that Accenture has found that as many as 74% of customers are willing to use computer-generated insurance advice through easy-to-use channels, such as messengers and voice.

See also: Strategist’s Guide to Artificial Intelligence  

In the era when a selfie is enough to buy insurance, like in the case of Lapetus Solutions, the customer gets much more than an insurance policy. The customer gets healthcare advice, possible savings, and dedicated products.

The downside of this approach has to do with privacy. In a world where your insurance company can determine what you did last night and if you took your medication, do you feel safe or do you feel part of a Big Brother system? What is the perfect balance between customization and intrusion?

What’s Hiding in Your Medical Records?

Every year, organizations pay millions of dollars in settlements for workers’ compensation claims for services they often shouldn’t. I’m not talking about insurance fraud but rather an unseen problem that routinely balloons out of control.

It is vital to understand how medical cost projections are calculated in Medicare Set-Asides (MSAs) to pinpoint the root of this significant and often inflated area. Let’s explore.

Today’s Process

Projecting medical costs for MSAs is done by estimating the future treatment likely to be required over the claimant’s remaining lifetime for workers’ comp injuries as well as any illnesses or conditions accepted under or exacerbated by that injury. The projection entails figuring out what treatment is likely to be required — the number of office visits, diagnostic testing, surgeries, medications, braces, basically anything that could pertain to a specific injury — and is typically calculated by summarizing the past two years of medical records.

Once the person charged with conducting the analysis — usually a nurse, an attorney, a claims person or someone with a managed care background — receives the records, she reviews them and generates a summary of two to three pages outlining the nature of the injury; history of medical treatment; and any recommendations for specific types of treatment, such as surgeries, hospitalizations or spinal cord stimulators. From there, a treatment table is developed, outlining everything the claimant will need for the rest of his or her life as per the expectations of the Centers for Medicare and Medicaid Services (CMS).

In general, CMS expects the following for any symptomatic body part paid for under the workers’ comp claim: X-rays every three to five years, MRIs every five to seven years and 12 physical therapy sessions. This is just the basic care; any recommendations or provision for future surgeries for these body parts will increase the costs considerably. Based on this information, all future predicted care is priced out according to the current workers’ comp fee schedule within the jurisdiction of the claim. Although a time-consuming process, it seems simple enough. But there is always a catch.

See also: Why Medical Records Are Easy to Hack  

Data Hiding in Plain Sight

Suppose an adjuster is reviewing a summary of a low back injury. The expectation would be to see treatment services related to the lumbar spine in the claimant’s medical records. Then, all of a sudden, the adjuster comes upon treatment being rendered for the knee. It may be a legitimate part of the treatment plan, and the knee may be accepted under the workers’ comp claim, but what if it’s not?

It’s become more obvious over the past 10 years — particularly with the advent of increasing recovery efforts from the Commercial Repayment Center (CRC) and Benefits Coordination & Recovery Center (BCRC) — that treatment for body parts or conditions is being paid for under workers’ comp claims to which it doesn’t pertain. The treatment has not been accepted under the injury claim, and, justifiably, insurance carriers don’t want to pay for it. Yet additional, out-of-bounds care often slips through the cracks and is paid for and now documented in the claim. This is a problem because, from the MSA standpoint, once a single payment is made for a condition, the payer has effectively bought it and all future medical care that comes with it. That knee is now part of the low back injury claim for the duration of the claimant’s life expectancy and will have to have future medical services included in the MSA.

How does this happen? It is actually very easy. Right now, adjusters often manage a desk of 150 claims at any given time. They spend most of their time talking to injured workers, to medical doctors who are working on their respective cases and to any managed care people involved in the claim. In addition, adjusters have a high volume of medical bills flooding in that need to be approved for payment. Adjusters can’t scrutinize all the information coming in on medical bills and think, “Is this injection actually connected to this claim?”

In the best of all worlds, the medical bill review teams, whether internal or an external vendor, would catch what has been accepted under the claim versus what’s being billed by the provider, but, in reality, charges still routinely slip through the cracks.

Fixing the Problem

New technologies that leverage artificial intelligence (AI) to “read” medical records are on the horizon. These systems can analyze all of the body parts and conditions being treated and compare that against the medical bill payment data. As such, smart systems are becoming the new front line for establishing exactly what gets paid under the claim and alerting claims adjusters to anything that doesn’t seem quite right. The adjuster can then take a look and discuss with the physician’s office the reasons for inclusion of the treatment before the bill gets paid and the abnormality becomes part of a future MSA, generating costs associated with lifetime care for the body part or condition.

For example, it won’t be long until applications can generate reports showing an additional alleged body part is now being treated under a claim. Alerts can be automatically generated to adjusters showing the scope “creep,” as what started as a low back claim has now expanded into the neck and shoulders. The injured worker is now also having problems with hips and knees, and three new medications have been added that the adjuster may not have been aware of. Applications can identify all of these vitally important nuggets hiding in the data and place them into context, allowing a wealth of information to be delivered to the adjuster’s fingertips in real time.

See also: How to Manage Risk of Medical Malpractice  

AI-based applications show tremendous potential for flagging issues that get missed. Machine learning fills in the blanks by understanding how things fit and how they don’t, even when it’s a little murky. The cost savings as well as the time saved in managing claims will be tremendous. Hidden data will finally be brought into the light so that people can make more informed decisions about what to pay and why. This is an exciting new frontier for MSAs, as medical payments are limited to only those body parts and conditions accepted under the claim, allowing the MSAs to be based on the most accurate, up-to-date information available, while holding down potential costs.

As first published in Claims Journal.