Predictive Analytics in Insurance – Top 6 Use Cases for 2022
Predictive Analytics in Insurance – Top 6 Use Cases in 2022
- Predictive Analytics for New Customer Risk and Fraud
- Predictive Analytics in Insurance Pricing and Product Optimization
- Predictive Analytics in Insurance Claims
- Predictive Analytics for Insurance Agent Fraud and Policy Manipulation
- Optimizing User Experience through Dynamic Engagement
- Big Data Analysis
But first, some history on the impact of AI, Machine Learning, and Predictive Analytics in Insurance
Predictive analytics in the insurance industry is nothing new, but over the past decade, we witnessed a titanic shift in the way insurance companies operate.
The companies that embraced the “Digital Transformation” thrived, while the companies and business models that ignored it or were slow to adopt an Internet/mobile strategy have sunk.
Hopefully, as the surviving insurers view the floating remains of their fallen competitors, they understand that a new threat has emerged.
And it has a name – Artificial Intelligence.
The use cases and applications of artificial intelligence in insurance analytics are seemingly endless.
While you shouldn’t expect to see an iron-clad Schwarzenegger approaching in your rearview, the impact of AI, machine learning, behavioral intelligence and the threat it poses on those who ignore it is very real.
The rise of applicable AI has been described as the 4th industrial revolution. Data is the new oil – and AI is the key to unearthing it.
For some perspective, 90% of the world’s data has been created in the past 2 years.
By this time next year, it’s estimated that 1.7MB of data will be created every second for every person on earth.
Don’t bother trying to do the math, I promise you, your calculator is not big enough.
AI and machine learning are the only ways to harness the insights from such an immense amount of information.
By using AI to look at the past, we are able to glean a previously unimaginable look into the future.
Predictive Analytics in the Insurance Industry Today
Unlike their digitally native counterparts, traditionally brick-and-mortar industries like Insurance have been very slow to adopt newly available technology.
They are lucky – their moats have, for the most part, yet to be breached. But times are changing.
For years, these behemoths have survived based on minor product enhancements and customer loyalty.
But decades of stagnant physical infrastructure, legacy business partnerships, and technological neglect have made their seemingly impenetrable fortresses a little less daunting.
And the newcomers like Lemonade are attempting to flip the insurance business model on its head.
Customers, especially millennials, no longer care that their parents used a certain broker or that the retail branch is in their town, they largely don’t trust insurance companies, according to EY and Accenture.
Insurance Competition is Stiff
For traditional carriers, when factoring in the availability of pricing transparency, reviews, blogs, articles, social networks, and industry influencers – there is no shortage of ways for a customer to discover everything they need before buying a policy.
And while the industry as a whole isn’t fully commoditized, it’s getting pretty close.
Turn on a Football game and you will see 6 different insurance companies vying for the same customers…
This one saves me 15% or more, that one has a quacking duck, the other one has Jake in khaki’s, another shows the mayhem in life.
I didn’t even mention the woman running around in the all-white commercials or the ones with Peyton Manning singing a jingle, but surely you get the point
Ignoring the companies with clever commercials and talking animals, a majority of the Insurance industry is still acting as if it is 1997.
Fax this. Snail mail that. Print, sign, scan, return. Or, those dreadful four words, “We do that manually.”
According to a recent PYMNTS case study – just 5.5% of Financial Institutions have adopted AI and only 12.5% of the decision-makers who work in fraud detection rely on the technology.
They instead rely on “more limited – and increasingly outmoded – technologies like business rule management systems (BRMS) and data mining.”
I genuinely fear for companies choosing to keep their heads in the sand.
And a lot of the time, it isn’t their fault – their systems are built on severely outdated technology. This makes it either physically impossible to improve upon or so costly to reconstruct that they choose to stick with the old, “It’s worked for us so far!” mentality.
Ignorance is bliss, as they say.
The digital transformations these companies must undergo to survive likely feels an awful lot like trying to steer the Titanic away from the impending iceberg. You’ve twisted the steering wheel as far as you can, but the ship only turns so fast.
So, turning our attention to what the future holds, what should these companies do? (Hint: here are a few ideas)
While waving the white flag and milking their cash cows until someone inevitably displaces them is certainly an option, it isn’t the one I would recommend.
Embracing the future and implementing an AI strategy could very well mean the difference between life or death for insurers.
Example: Predictive Analytics in Life Insurance
Using advanced machine learning and new digital datasets, insurers are finally able to apply the same risk measures they have been utilizing manually for centuries in a much more efficient manner.
As life insurers continue heading in the direction of accelerated underwriting and straight-through processing of claims, predicting customer behavior and intent is more important than ever.
The balancing act that Risk and Customer Experience teams go through can be exhausting. How do you juggle creating a seamless experience for your customers without opening up the gates and letting in a trojan horse?
The answer lies in understanding user behavior to predict their intent.
For instance, most life insurance carriers are attempting to reduce the number of fluid tests required by applicants to complete policy applications. Not only are they expensive, but they are challenging logistically. COVID has exacerbated this problem quite a bit.
Rather than relying on spot-checking policies after they have been approved, or retroactive analysis after claims have been filed, life carriers are turning to predictive analytics such as Behavioral Intelligence to determine who may be misrepresenting themselves on their applications.
Medical and tobacco usage non-disclosure is the #1 issue facing life carriers today so proactive measures must be taken to protect against future losses.
Solutions such as ForMotiv’s for tobacco usage non-disclosure are helping carriers identify high-risk behavior in real-time so they can take action before it’s too late.
According to our customers, 11-13% of digital applications have some level of misrepresentation or fraud, and of those, 20-30% are underwritten. Given life insurance policies pay hundreds of thousands, sometimes millions of dollars in death benefits, it’s no wonder the industry loses nearly $4billion a year as a result of this issue.
Life insurers use ForMotiv’s predictive analytics to solve this problem in a number of ways including:
- Identifying high-risk customer behavior
- Predicting cases of application misrepresentation
- Increasing opportunities for accelerated underwriting
Learn more about our Tobacco Usage Non-disclosure Solution here.
Example: Predictive Analytics in Health Insurance
Health insurance companies are using predictive behavioral analytics and beginning to integrate Internet of Things devices as well.
With 5% of all patients accounting for nearly 50% of all healthcare spending, it’s more important than ever to utilize available predictive analytics solutions to identify risk factors in patients before they become problematic.
Wearables such as Fitbit and or Apple Watch can provide ongoing assessments of the individual’s health risk exposure.
Adding predictive behavioral analytics and predictive analytics, in general, helps limit losses for more advanced insurance carriers. The use cases for Behavioral Intelligence and artificial intelligence especially in applications and claims are seemingly endless. According to LexisNexis Risk Solutions, the top three areas where health insurance companies benefit from the use of predictive analytics are:
- Data-driven claims decisions
- Reduced operating expenses
- Improved profitability and expansion in new and existing markets.
Data-driven claims decisions are paramount in ensuring profitability and getting in front of costly patients and policies.
And according to GenRe, the top six ways predictive analytics are being used by health insurers to optimize claims processing operations are as follows…
- Allocation of resources/triage
- Reserving/settlement values
- Identification of potentially fraudulent claims
- Early warning of potentially high-value losses
- Expense management
- Trend analysis
The Future of Predictive Analytics and Machine Learning for Insurance
To its credit, a majority of the insurance industry has become keenly aware of the technological advances that threaten their incumbent businesses.
Mobile-first business models have stripped away the costs of having a heavy physical presence.
This opened up holes in the canopy for new entrants to grow.
In an effort to stay ahead and fight off companies looking to dis-intermediate traditional insurers, 66% of the legacy players are choosing to invest in and adopt their own AI and technological solutions.
Investments range from car sensors and telematics that monitor driving behavior and AI software that analyzes social media accounts to Drones, IoT device networks, behavioral intelligence, and predictive analytics for insurance underwriting.
The amount of data created on a daily basis is incomprehensible for most humans. And because of that, insurers are looking at new ways of analyzing that data for a competitive advantage.
We have already seen a significant amount of process automation and digital transformation in the last decade.
The next ten years, however, will be all about behavioral intelligence and predictive analytics insurance software.
How is predictive analytics used in insurance?
Simply put, by looking at our past, we are able to better predict our future.
Looking at the past decade, the insights are fairly obvious…
Streamlining online experiences benefitted customers, leading to an increase in conversions, which subsequently raised profits.
Add in operational automation for increased efficiency and you’re looking at millions if not billions of dollars a year in additional revenue and cost savings.
That strategy worked for a while. However, simply automating repetitive tasks and giving your website a makeover will not be enough to withstand the onslaught of competition.
In order to survive, insurers must integrate AI/machine learning, behavioral intelligence, and predictive analytics everywhere they can.
Integrating predictive analytics insurance software has quickly become the leading initiative on most of the top insurance carriers’ roadmaps.
What used to be a traditional, rule-based framework is now transforming into a data-driven, automated, highly intelligent and predictive system.
So, without further ado, here are the Top 6 ways Insurance Carriers are using predictive analytics today…
Insurance fraud has many faces…Stolen identities to obtain a new policy, false payee information, false declarations, computer bots and so on.
According to the FBI, the annual losses related to insurance fraud are as high as $40 billion, costing the average American family $400-$700 in increased premiums each year.
To combat this, companies have begun adopting predictive analytics insurance software to reduce risk and prevent fraud.
For example, by crunching data collected by behavioral biometrics and behavioral analytics software companies, companies can correlate user behavior against past customer records to detect fraudulent activity and suspicious behavior patterns.
This newly created “Behavioral Intelligence” is leading the charge into a more secure and smarter future. We’ll discuss the diverse use cases of Behavioral Intelligence more below.
1A. Predictive Risk Scoring with Behavior Analytics
An important use case of Behavioral Intelligence and predictive analytics in insurance is determining policy premiums.
Not too long ago a majority of business interactions were done face-to-face, making it exponentially more difficult to get away with risky behavior.
For instance, if a customer pulled out a sheet of paper and was copying over their home address, social security number, and the spelling of their middle name – that would likely raise some red flags.
Today, however, as businesses have shifted online, most business interactions are now ‘faceless’ and that type of behavior happens every day. Customers, fraudsters, even bots attempt to appear as good as they possibly can on paper.
Bots can automatically apply to thousands of financial service companies for thousands of different products. They only need one approval to cause serious harm.
Because companies and their agents have lost the ability to read and react to their customer’s body language, they are forced to grade that customer’s risk based on whatever the ‘final answer’ is that they submit. So utilizing artificial intelligence in insurance applications and other similar use cases is imperative.
Can you imagine sitting down face-to-face with an insurance agent today, but before you begin filling out the papers they put on a blindfold?
That would be like a teacher walking out of the room after handing out the test.
And on top of that, the teacher didn’t require that you ‘show your work.’ Instead, they simply graded you on your final answer.
But what if a life insurance applicant was correcting answers on their medical history, first putting they were a smoker, filling out the drop-down questions, but then changing the answer to say they’ve never smoked. “Smoker’s amnesia” as we’ve heard it called.
As an insurer, isn’t that something you would want to know?
Luckily, with Behavioral Intelligence, you can know that and much more. Companies like ForMotiv are using Behavioral Intelligence and predictive behavioral analytics to both alert companies of specific customer/agent behaviors, as well as predict the severity of these offenses to help grade risk appropriately.
By measuring customer (or agents) “Digital Body Language” – i.e. keystrokes, idle time, mouse movements, copy/paste, corrections, etc. – ForMotiv is able to use machine learning to correlate certain behaviors to outcomes like risk and fraud.
The way a user fills out an application can be highly indicative of their actual risk versus the risk assumed by their final answers. Digital body language like hesitating on certain questions, correcting important fields, viewing rates, and going back a screen to edit answers and re-view rates, even agents changing previously submitted customer answers can all show higher signs of risk that companies would be otherwise blind to.
ForMotiv customers are alerted of these behaviors and many more in real-time and can either take manual action or automatically price their risk accordingly.
This is a far superior solution to what most companies are doing today – which is waiting until there is a claim in the future and attempting to figure it out then.
It’s the difference between prescriptive medicine and reactive medicine.
Speaking of healthcare…
1B. Behavioral Biometrics to Prevent Account Takeover and Fraud
Behavioral Intelligence, not to be confused with behavioral biometrics, is great for assessing new customer risk and comparing it to every other user.
It uses predictive behavioral analytics to measure how unknown user John Smith compares to the millions of other applicants and their outcomes and predicts what John Smith’s likely outcome is.
This is often confused with Behavior Biometrics, and while they play in the same arena, they’re playing different sports.
Behavior biometrics is all about comparing John Smith to John Smith.
Behavioral Biometrics helps companies with identity proofing, continuous authentication, account takeover fraud, and vishing scams.
Your signature, voice, thumbprint, and face are unique to you – so is the way you interact with a device. Behavioral biometrics measures how John Smith uniquely interacts with a device.
Is he typing the same way? Was he right-hand dominant and now left-hand dominant? Does he swipe up or down the same?
Using behavioral biometrics, companies can determine if a logged-in John Smith is, in fact, John Smith. Another way this can be helpful is Voice Biometrics for account verification, which is often done over the phone.
Companies like V2verify are changing the game when it comes to voice verification, needing only 2 seconds of speech to accurately identify someone.
“The AI is the secret sauce of our Voice Biometric technology. Most Biometrics suffer from an inability to change and evolve after initially mapping a person’s vectors. With our patented processes, our AI enables the ability to evolve at the user level, real-time; something that is extremely important when using a person’s voice to ID or authenticates,” says Alan Smith VP of Sales with V2verify.
Because of this, behavior analytics software can help drastically reduce account takeover, prevent fraud, and enhance identification protocols.
Armed with more granular data and predictive analytics insurance modeling, actuaries can now build products better suited to dynamic business and market conditions, risk patterns, and risk concentrations.
In other words, historical costs, claims, expenses, risks, and profits are projected into the future.
Predictive analytics algorithms give insurers the opportunity to dynamically adjust quoted premiums.
For instance, in property insurance, continual monitoring of variables like claim history in the neighborhood, construction costs, and weather patterns helps to predict risk and price more accurately.
By analyzing customer preferences, behavioral signals, buying patterns, and pricing sensitivity, companies are able to use their predictive algorithms powered by machine learning to constantly optimize and showcase more relevant insurance products.
Up until now, it was difficult to customize policies at the individual level. However, companies can now use pay-as-you-go and dynamic pricing models based on customers’ predicted risk, behavioral signals, and buying preferences.
And many of the digital-first products are a result of millennial influence.
As Richard Hartley, CEO & Co-Founder of Cytora puts it in Gina Clarke’s “How Your Insurance Quote Is Powered By A.I.” article…
“Millennial consumer behavior is forcing irreversible changes across financial services leading to the emergence of digital-first and app-based services for banking, loans, mortgages, and investment. As the millennial cohort start their own companies and move into decision-making roles in business, commercial insurance is beginning to undergo the same revolution.”
Given millennials and Gen Z are quickly making up a majority of the buyers in the insurance market – what should traditional insurers do?
There are a few options, but simply put, they need to skate to where the puck is going.
By integrating currently available AI and predictive analytics tools, they can avoid a full reboot of their legacy systems.
Not only that, but they’ll be able to thrive in the new age of digital transformation.
Automating insurance claims processing was a huge step forward as insurers continue their digital transformations.
Given that claims are the part of the insurance lifecycle that has the highest percentage of attempted fraud, it is one of the first places companies are looking to integrate AI.
While legacy insurers are integrating AI software into their legacy claims process, companies like Lemonade are starting with an AI/behavioral-first approach.
They tout that they can process claims faster and by using a chatbot, they’re able to provide customers with faster payouts.
Lemonade isn’t the only company using chatbots during the claims process.
These chatbots are getting more sophisticated and can review the claim, verify policy details and pass it through a fraud detection algorithm before sending wire instructions to the bank to pay for the claim settlement.
This can help speed up processes and reduce human error.
Other companies like Tractable offer machine vision software to help insurance agencies automate claims.
Insurance agents can upload imagines associated with a claim, such as a damaged car, and an estimate of what they think the appropriate payout is.
The software then compares the image to a database of similar images and allows the agent to make smarter payout decisions.
This helps companies avoid overpaying for claims.
Some companies like Cape Analytics offer a service that they claim can help property insurers underwrite more accurately and more cost-effectively using satellite-based machine vision.
They can assess information about the roof, property, treeline, pool, trampolines, etc. saving the company from needing to send a human inspector to the property.
AI is also used to spot anomalies and unknown correlations that would be impossible for the human eye to detect. This lead to increased opportunities for straight-through-processing.
Companies are smart to look at reducing insurance fraud during new account opening and claims, but if their fraud prevention efforts stop there they are missing out on a hugely important area.
While fraud continues to evolve and affect all types of insurance, the most common in terms of volume and average cost are automobile insurance, workers’ compensation, and health insurance / medical fraud.
The tricky part for insurers, however, is that large percentages of fraud are actually coming from inside their own walls.
Believe it or not, customers are not as savvy when it comes to committing fraud as their agent counterparts.
One very common but hard-to-prove way insurance agents commit fraud is application manipulation.
ForMotiv recently worked with a Top 10 Life Insurance carrier to identify and solve this exact problem.
The original use case was to determine how many questions customers were manipulating on their life insurance applications.
For instance, were they changing their source or amount of income? Or were they trying to game their e-med questions to receive a better rate?
As it turns out, after a month of behavioral data collection we found some phenomenal insights regarding the agents.
The data showed the following… 72% of the applications had 2 or more questions corrected by an AGENT after being submitted by an applicant.
The Insurer had this to say…
“You helped us find the agents who represent themselves better than their employer and customer.”
Yes, we were able to identify a significant amount of customer manipulation as well. But what we did not expect to see was how often and aggressively agents were gaming the application.
Changing a few key answers to receive a better rate helps them convert more customers. More customers = more commissions. Simple formula.
We’ve heard this from a few customers and prospects… “Oh, no, our agents would never do that.”
Well, I hate to be the one to break it to you, yes they would.
To think there is absolutely zero suspect or blatantly fraudulent activity going on is like thinking your kid didn’t have their first beer until they were 21. This is why predictive analytics in life insurance is paramount in detecting and preventing fraud.
Using ForMotiv’s “Forensics” tool, customers are able to clearly determine not just WHAT answer is being provided, but HOW and by WHOM.
This level of insight was previously impossible to extract.
Today, it is being used by 4 of the Top 10 life insurance carriers. Gathering behavioral intelligence with behavior analytics software can protect carriers with claim contestability and special investigations.
Not to mention, it can save companies millions of dollars.
Companies need to be aware of the fact that internal or distributed agents often act in their own best interest. And their self-service ignores the ramifications for their clients and companies.
According to ITL and their prediction of InsurTech trends, the main focus is on a digital-first customer-centric approach.
A KPMG report also stresses how customer satisfaction and retention is becoming a more important KPI than operational efficiency.
As products are commoditized, loyalty becomes a thing of the past.
So what do you do now that maximizing customer satisfaction has become the name of the game?
I believe predictive analytics for insurance holds the key to achieving optimal customer experience and, ultimately, customer loyalty.
Using behavioral AI tools, companies are able to uncover behavioral insights at the form field level.
For instance, ForMotiv gives its customers behavioral intelligence on how their users and agents are actually interacting with the forms and applications, in ranked order, and provides explanation-based A/B testing recommendations.
This insight allows marketing and customer experience teams to remove bottlenecks, troublesome questions, and chokepoints and optimize their form fields for increased conversion and great customer & agent satisfaction.
In addition, companies can use innovative predictive behavioral models to measure user intent, in real-time, and can uncover insights into the actual intent of the users.
Does this look like a profitable customer? A fraudster? Is someone having trouble with the application?
By reading a customer’s digital body language, companies can use predictive behavioral analytics to create dynamic experiences for customers.
This helps to reduce friction for ‘good’ customers and add friction for seemingly ‘bad’ customers.
Do they seem confused or stuck on a question? Offer contextual help, a chatbot, live chat, and more.
Are they behaving in a risky manner or acting like a bot? Dynamically add friction, such as an “Upload a government-issued ID” process question.
Using these same tools, companies can predict application abandonment with almost pinpoint accuracy.
This allows them to dynamically engage a user who seems likely to abandon the application.
As we mentioned before, the amount of data created every second is virtually incomprehensible.
For a little context- the difference between a million seconds versus a billion seconds is 11.5 days versus 31.75 years.
So comparing a million IoT devices to a few billion?
It’s mind-numbing when you consider the data created by these devices.
In 2020, it is estimated that there will be 20.4 billion IoT devices. That is ~130 new devices connected to the Internet every second.
Cisco expects the total data generated to exceed 800 zettabytes, with a single zettabyte equal to about a trillion gigabytes.
Using the above time example, a trillion seconds equals about 31,710 years.
With about 90% of the data being unstructured, companies will be forced to embrace machine learning and predictive analytics more than ever to keep up with the demands of IoT.
Telematics (in-vehicle telecommunication devices), drones, wearables, smart speakers, refrigerators, washing machines, toasters…
By adding Internet access to every device imaginable, predictive analytics for insurers will be crucial for survival.
For instance, the behavioral data of applicants is computed when underwriting premium rates for vehicle insurance.
Does the driver slam on the brakes? Do they peel around corners? Do they park their car in deserted locations? Are the road conditions good where they drive?
By applying predictive analytics, insurers can assess the likelihood of the insured in being involved in an accident, as well as the odds of having their car stolen by matching behavioral data with external factors like safe neighborhoods.
Ultimately, this helps tailor policies and premiums that protect the insurer as well as the insured.
Artificial Intelligence, Predictive Behavioral Analytics, and Behavioral Intelligence Analytics have never been more important to implement for insurers.
Using cutting-edge insurance analytics solutions is the best way for insurers to fend off competition and thrive in a competitive market.
As the digital shift continues to impact the industry as a whole, transforming user data into actionable intelligence is imperative, and integrating artificial intelligence in the insurance application process is a perfect use case.
Companies that integrate predictive analytics into their insurance analytics solutions will undoubtedly increase their market share. They will also boost customer loyalty and can significantly grow their revenue while reducing their costs.
Other Articles You Might Like:
- Did Lemonade Kill the Combined Ratio?
- What is Digital Behavioral Intelligence?
- Top 5 InsurTech Companies Disrupting the Insurance Space