Using AI eye-in-the-sky to improve insurance coverage

Last year was the second-costliest ever for the insurance industry, with roughly $120 billion in losses caused by natural disasters alone. With the tally from Hurricane Ian already estimated to be over $70 billion, this year may shape up to be another one for the record books. Against this backdrop of catastrophic losses, rising costs and heightened competition from all corners of the marketplace, property and casualty (P&C) insurers need to find ways to streamline their operations, improve the accuracy of their risk forecasting and drive stronger customer engagement.

To get there, they are going to need more powerful data and better predictive analytics.

Each year, roughly $719 billion in P&C insurance premiums are written to help U.S. homeowners protect their property from the effects of destructive weather, fire, flooding, theft and countless other risks. However, a large portion of that total value is at risk due to inaccurate and incomplete property assessment data.

Based on our analysis of property claims data, we’ve consistently found traditional means of collecting and analyzing property risk through self-reported questionnaires and sporadic site visits provide wildly inaccurate readings of actual risk exposures. In fact, by applying new forms of analytics-driven aerial imaging technology — or computer vision (CV) — alongside predictive risk scoring analytics, we’ve found numerous unaccounted-for hazards, including: overgrowth and brush in fire-prone areas, damaged roofing and siding, and overhanging branches that could put homeowners — and their P&C insurers — at greater risk of a claim.

The rise of computer vision in insurance

CV is essentially what it sounds like: an AI-driven process through which computers can derive meaningful insights from digital images. Many insurance carriers are now using these advanced aerial images to survey large, remote areas of land and assess properties. These automated assessments include pertinent property information, risk data, and estimated damages from recent events.

It’s a game-changer compared to traditional man-powered assessments. CV-driven analytics capabilities can identify changes to the property since the last evaluation. For example, a policyholder could have installed new solar panels this year and not reported it to their insurance carrier. This type of addition, as well as the property’s deterioration over time, such as roof quality degradation or vegetation overgrowth,  increases risk beyond what was originally contemplated for both the property owner and insured in the event of a covered peril.

What’s more, a CV-enabled platform can push out updates related to property changes, alterations and modifications at regular intervals, especially during the time of policy renewal. By tracking how a property’s condition changes over time, both potential threats and risk optimization opportunities can be identified with less effort and resources, and earlier in the process.

Customized premium pricing

Getting real-time insight into how property conditions are changing over time is also vital to being able to deliver more accurate, personalized premium pricing. Take roofs for example. Condition is a far more accurate indicator than age. So the premium for a property with a roof installed in 1992 that was consistently maintained can be lower than a property with a roof installed in 2012 that was subject to more severe weather conditions such as ice dams in the winter, provided imaging captures those condition factors.

Using this type of image-based analytics, several factors can be evaluated to indicate the condition of a property and update premiums in real time. For instance, overhanging trees can increase the frequency and severity of a claim. Armed with this information, insurers can choose to either provide rating for this new risk or use the information to engage with the customer proactively and recommend that they trim overhanging branches as a loss avoidance measure.

Aerial imaging can also be used to access historical quarter-by-quarter property features, which, when paired with conventional pricing variables, can help estimate upcoming claims and align more risk-appropriate insurance premiums. This type of customization not only rewards customers for mitigating their property’s risk, but also drives a level of customer engagement all carriers encourage.

Speed and efficiency

CV can also help expedite the claims process at a time when customers are most in need. In the wake of extreme weather or a natural disaster, imagery can be used to estimate affected properties, and CAT response teams can be strategically dispatched where they are needed.

CV also enables more tactical efficiencies in insurance operations, such as eliminating the need for adjusters to climb onto a steep roof or severely damaged structure to get an accurate reading of damage and identify necessary materials. As a result, more claims could be handled by adjusters while still providing customers with quicker turnarounds and increased customer satisfaction. It is also possible to conduct a pre- and post-analysis, which can provide a second opinion on the extent of the damage caused by a natural disaster.

Finding your customer

Beyond more accurate underwriting and better claims processes, CV also lets carriers expand into new markets or increase their presence in existing ones profitably. By having more complete and accurate property information, an insurance carrier is better suited to assess each risk properly, thus gaining a competitive advantage over those who are not.

Compared to traditional methods, this can result in a sustained increase in marketing ROI by targeting properties that exhibit favorable characteristics using property attributes and risk scores, which lowers loss ratios, and using the preferred marketing channels to contact targeted prospects, resulting in higher bind rates.

The analytics that drive the future

As CV is adopted by more of the industry, the technology will undoubtedly advance, allowing future aerial data technology to become even more accurate. And as insurers look for alternatives to physical audits and surveys in an effort to make underwriting a precise, low-touch and automated process, image analytics will continue to have a profound impact on property insurance process efficiency, cost structure and overall profitability.

Forward-thinking P&C carriers can get a jump on the industry by integrating this technology now. Early adopters will have an easier time finding a comfort level and be able to tailor these solutions to their specific needs and challenges. With the likelihood of more mass-loss events on the horizon, P&C carriers need to find a way to be more agile, and CV could be one option.

Rahul Nawab is senior Vice President, Analytics, at EXL, a multinational data analytics and digital operations and solutions company. Joseph Nemet, FCAS, is Director, Modeling & Data Analytics at Erie Insurance.

Related:

Leveraging data to better anticipate & plan for NAT CAT events

Safety mitigation measures for wildfire risks could herald a new era in disaster insurance