By Jim Weiss
Not long ago, before the social media revolution, a “yelp” was the sound that a puppy made when someone pulled its tail. Likewise, “twitter” usually involved a conversation among finches. And by anyone’s standard, consumer technology was only as “smart” as its operators.
In short order, the digital world launched a new language that is blurring lines between home and workplace, between bricks-and-mortar reality and the Internet’s virtual space. No longer just for puppies, Yelp is a website that rates restaurants and posts reviews by the public at large. In the same vein, an application known as Waze helps drivers navigate past road hazards and through traffic, using real-time reports posted by fellow drivers. New meanings and methodologies are being derived from old words.
Information flashed across social media has also helped transform ordinary mortals into aspiring scientists. When facing decisions as simple as where to go out for dinner, people now consider data points as diverse as strangers’ reviews of establishments on Yelp, the quickest route to a restaurant on Waze, and whether businesses have posted promotions or coupon codes on Twitter. In the blink of an eye, social media’s revolution has reached the dining table.
A similar range of data is often considered when purchasing an insurance policy. With property and casualty (P&C) policyholders employing big data analytics, insurers whose decision-making processes fail to include insights derived from social media may risk making a poorly researched decision. Fortunately, a variety of tools are emerging to help insurers incorporate insights obtained from social media sources into decisions across their organizations at every point of the value chain — from initial reporting to settlement of a claim. So why aren’t all insurers putting the pedal to the metal on these analytics when policyholders themselves are doing so for decisions as simple as where to eat dinner?
One possible reason is that data analytics drawn from social media don’t scale easily. Although the data is in many cases free and available to the public (often through open-application programming interfaces), it’s also often unstructured and full of gaps, false statements, and hyperbole. Further, the data pool is vast and in constant motion. To make timely and incisive decisions using the data, insurers need to establish pipelines that draw out and filter the data to obtain the most relevant content. This may require a two-fold commitment, not only to replace outdated IT architecture but to recognize the voice of policyholders as expressed through social media. Insurers that rise to those challenges can derive practical insights that provide a competitive advantage.
In the social media marketplace, data analytics are commonly used to determine how consumers understand a brand. Social platforms have helped level the playing field between insurers with small versus large advertising budgets by enabling tech-savvy insurers to reach greater numbers of prospective policyholders at relatively low cost. Success is frequently measured by how often brands are mentioned on social media platforms, the color and sentiment of those references, and whether content authors promote the brand.
However, such metrics are not relevant in all industries; insurance is different from movies or soft drinks. An analysis of how well an insurer connects with target policyholders should correlate the social metrics described above with premium growth, profitability, and risk concentrations — and compare them in a meaningful way with competitors. By understanding how brand perception relates to business performance, insurers can make more informed marketing decisions that may result in not only being “liked” but also successful.
Marketing efforts aside, insurers entrust their reputations every day to professionals who interact with policyholders during “moments of truth,” at the point of sale or the reporting of a claim. Social media and data analytics can help representatives hit the right notes at those key moments. Many agents and brokers analyze data from social media to mark relevant leads, as when a homeowner indicates he or she is in the market for a new vehicle. Those analyses help ensure prospective policyholders’ future needs will be met, while potentially reducing the volume of mutually frustrating cold calls.
In the same way, social data has helped insurers identify and assist claimants after catastrophes. By superimposing images and profiles sourced from social media over the map of a catastrophe-stricken area, insurers can make preliminary assessments of damages and send assistance to the neediest policyholders. Those worst hit may be unable to report claims because of disabled phone lines or utilities. Those are the types of interactions that help create “policyholders for life” and can generate positive perceptions online that are reflected in an insurer’s brand metrics.
Just as data from social media may help insurers understand policyholders’ coverage and claims-handling needs, deeper data may also lead to more accurate assessment of risk. In developing countries, commonly used predictors for P&C, such as a credit history, may not be readily available, and engagement with social media may be more prevalent than the presence of a credit history. Parallels exist for certain market segments in North America, where social review sites such as Yelp offer unparalleled insights into the risk characteristics of small businesses. These could present opportunities for loss control; for example, in a restaurant where reviews indicate a problem with undercooked food, owners may earn scheduled credits for improving the restaurant’s Yelp rating by addressing areas of concern.
Likewise, just as the sharing economy produces ratings for an individual’s performance as a contractor (such as someone who contracts to carry passengers for Uber) or patron (for example, someone who “hails” a ride using Uber), a similar approach may be taken for personal lines insurance. Uses of social data for risk assessment would likely operate on an opt-in, discount-only basis, similar to recent innovations such as usage-based insurance (UBI). After an initial exposure to the concept, regulators and the public may be receptive to these approaches because they can help create incentives for policyholders and serve to promote safer habits.
Profitability of a P&C account often depends not only on any claims that occur but also on how long a policyholder stays with the insurance company. That concept, known as customer lifetime value, is high on CEOs’ radar as increasing levels of advertising expense and pricing refinement yield diminishing returns and the focus shifts toward reaching longer-term customers who may promote the business or make future purchases with the company. Social media data can be valuable in assisting insurers to make better lifetime-value decisions because of the data’s timeliness and raw insight into policyholders’ behaviors as customers. Aside from identifying opportunities for cross-selling (such as a homeowners policy customer buying a car), studies from life insurance have shown consumer data can be highly predictive of policyholder longevity. This data includes information sourced from social media about leisure pursuits.
An ISO focus group of ten P&C professionals found that policyholders’ past experiences as customers (including content posted on social review sites) can be predictive for the longevity of P&C policies. An individual who issues online blackballs for every business transaction in the past year may be a less promising candidate for solicitation than one who presents a more evenhanded mixture of positive and negative reviews. Aside from its long-term revenue prospects, the latter category may also contribute to a more positive perception of the insurer brand.
As insurers strive to raise business standards and satisfy their customers, undoubtedly some fraudsters will be looking to scam. Social media data has proven to be a markedly beneficial tool in diagnosing both hard P&C fraud (for example, a staged claim event) and soft P&C fraud (for example, failing to mention that one was speeding at the time of an accident). At point of sale and from a glimpse at the applicant’s available social media data, an underwriter can potentially check whether there are additional drivers in a household or if the applicant is a ridesharing contractor. Similarly, a claims adjuster may give closer scrutiny to claimants whose social media data displays windsurfing pictures following an allegedly debilitating back injury. If deeper investigation is called for, insurers requiring a greater degree of analytical sophistication may even employ social network analysis, which would mine the data for an individual’s hidden associations with known fraudsters on social media platforms (for example, “friends of friends”).
Using social media data to uproot fraud often results in a higher value proposition for all policyholders because fewer premium dollars are wasted in paying out fraudulent claims or undeserved policy discounts. And wouldn’t that would be something to “tweet” about?
Jim Weiss, FCAS, MAAA, CPCU, is director of Analytic Solutions at ISO.