How unstructured alt data is helping insuretech disruptors change the game

Learn how: Algorithmic actuary models are decreasing insurance fraud, dynamic usage-based pricing models are driving sales among millennial demographics, and how structured, historic, publicly available data is enabling more efficient underwriting  

In this article we will discuss:

• Algorithmic actuary models are decreasing insurance fraud
• Dynamic, usage-based pricing models are driving sales among millennial demographics 
• Structured, historic, publicly available data is enabling more efficient underwriting 

Algorithmic actuary models are decreasing insurance fraud

Insurance companies are constantly looking for ways to analyze statistics, and calculate risk. In order to achieve this new actuary algorithms can, and are benefitting by being fed:

Alternative, structured, historic data 

By using automated data collection networks, insurance companies are able to collect historical data sets that are unstructured (e.g. old/public PDF medical records), and get them delivered directly to their AI, and ML in structured, ready-to-use formats. 

Depending on the policy in question, here is a sampling of some of the data sets being leveraged in this context:

• Crime rates in a specific area where a company or business is located
• Weather data from past years to calculate the possibility of natural disaster damages
• Social media sentiment on a business to see how customers, and suppliers ‘rate’ their trustworthiness  to be used as part of their ‘social credit score analysis’
• Public records in order to gauge criminally activity, employment of illegal immigrants, and any other misconduct which can raise said business’s risk profile 
• Search engine results that can bring to light news coverage, and user-generated content that can shed positive/negative light on the business in question

Algorithms can also collect records of past fraudulent, and non fraudulent claims, and use these to create an automated model which can flag dangerous anomalies. This saves a lot of time, and money throughout the vetting process, and helps streamline, and spotlight claims that need the most attention. 

Immaculate data leads to impactful output 

Data integrity is also extremely important ensuring that the data you feed algorithms are:

• Reliable
• Traceable
• Clean 

As was accurately pointed out in a recent data collection piece entitled: ‘Why ‘clean data sets’ are key to driving meaningful ROI for businesses using AI and ML’:

“You must always remember that your AI, and ML are only as valuable as the data sets you feed them. All roads lead to the quality and accuracy of the data sets that you collect, and compromised data means derived trends, insights, and conclusions need to be thrown out with the bathwater.”

Dynamic, usage-based pricing models are driving sales among millennial demographics 

The one-size-fits-all approach to insurance is no longer cutting it. Whether they are in the market for personal health/car/home insurance, or looking for policies for their businesses, younger consumers are not willing to settle. There are currently tools, and apps which allow them to compare pricing, and ever since companies like ‘Root’ have made their debut in the insurance market, the race for data-driven, usage-based insurance is on. 

Here is a user testimonial describing Root’s unique approach to pricing insurance policies based on user’s driving habits: 

This type of pricing model is very attractive to Gen Z’ers, and millennial business owners who crave event-specific insurance models such as:

• Only paying for employee liability policies when workers are on-premises, and clock-in, and out. 
• Paying for healthcode, and consumer food poisoning protection proportionately to the quantity of potentially harmful dishes actually being sold (e.g. if 50% of dishes are fish and the other 50% vegetable-based, this information could, and should reduce restaurant premiums).

But this type of data may only be accounted for once an insurance company establishes an intimate working relationship with would-be consumers. In order to gain a competitive advantage they need to be able to use anonymized alternative data in order to build a ‘plausible business profile’, enabling them to give real-time ‘tailored’ quotes. 

For example, Alternative information sources, such as geospatial data, can help companies better map areas with upcoming hurricanes, and snow storms, for instance, enabling them to make tailored property damage offers to consumers in relevant GEOs.

Other examples of data-elevated customer experiences include data collection regarding crime rates in specific business districts – offering customers in low-crime areas more attractive premiums, while offering additional, albeit customized protections to those in more dangerous parts.

Structured, historic, publicly available data is enabling more efficient underwriting 

Getting approved for a policy used to be a lengthy 60-90 day, hand-written, and mailed-in process. Business customers were hungry for a quicker, more streamlined process which brought about the advent of online forms. But the truth is that many of these were being checked by humans, and just giving the impression of digitalized efficiencies. 

More recently, companies in the industry utilizing insuretech have embraced automated, real-time application submissions, and approval. Customers want to be approved immediately, having very little patience in today’s ‘instant economy’. 

But as with the algorithmic actuary approach described above, this hyper-efficient, quick underwriting method requires models be fed structured, historic, publicly available data sets, such as:

• Motor vehicle reports 
• Geolocation data
• Transactional, and financial data [such as ‘Fair, Isaac and Company’ (FICO) credit ratings]
• Health code compliance records
• Social reputational harm data
• GPS, and geospatial smartphone data
• Open-source web data (including store opening times, product pricing, inventory levels, customer reviews, etc)
• Business/social network data (such as LinkedIn) 

Inaccurate misrepresentations of, for example, the number of stores, employees, and corporate activity locations will raise a red flag and be sent over for ‘human underwriters’ to further review. All the rest will be approved on the spot. Making ‘instant gratification’ part of your business model using data will ultimately increase your competitive appeal, and market share.  

The bottom line 

Since the actuarial sciences were established in the 17th century, the insurance industry has been heavily reliant on data for its risk assessment models. But as companies look to maneuver through the age of on-demand personalization, insurers will need to shed antiquated, time-consuming methods of data collection, analysis, and implementation. Policy pricing, underwriting, actuary analyses, and marketing departments will need to hook themselves up to a live pulse of consumer-generated data so that they can appeal to prospects with hard-to-refuse, out-of-the-box insurance experiences.