Today’s property and casualty (P&C) insurers are constantly pursuing the elusive gold standard of property intelligence. Geospatial imagery holds immense potential for risk assessment, providing real world context for address data that can form the basis of this gold standard. However, many insurers still struggle to scale their imagery analysis workflows, ensure data accuracy, maximize ROI, and eliminate data redundancies.

For example, there are some underwriters who admit to using Google Maps to supplement official data sources in their imagery-based risk assessment of a property. Many insurers have even blocked this URL on their employees’ devices because it can cause several downstream issues. For one, Google does not allow commercial use of this data. Additionally, Google Maps data is not controlled or validated, so cannot be considered the gold standard for firmwide property intelligence. What’s more, when different departments use disparate data sources, it introduces miscommunication and data discrepancies that can have costly consequences.

Underwriters resorting to using Google Maps is just one way P&C workflows can be affected by the increasing complexities of geospatial data and risk assessment, and there are countless other examples we could cite of sub-optimal analytics resulting from inefficient or inaccurate data creation. Luckily, advancements in artificial intelligence (AI) are changing the way insurers source and analyze property intelligence data, solving critical challenges that ultimately impact the bottom line. 

Top 4 challenges of imagery analysis in insurance

Imagery analysis is an extremely powerful capability for the insurance industry, but must be optimized for efficiency, cost, and accuracy in order to be truly effective. AI is making it easier than ever for P&C carriers to take full advantage of all that imagery has to offer by addressing these challenges. Here are the top 4 challenges insurance carriers face with traditional imagery analysis methods, and how AI is helping the most innovative firms rise above the competition.

1. Manual imagery analysis is resource-intensive

Interpreting imagery for one property can be simple, but when looking at thousands or millions of addresses across a book of business, manually reviewing imagery quickly becomes unscalable. Many P&C teams devote hours upon hours of valuable resources to manually analyzing geospatial imagery of properties for risk assessment - resources that could instead be allocated to developing sophisticated risk scoring models or other data-driven strategies that can really move the needle.

AI is increasingly helping carriers scale their risk assessment workflows by eliminating the need to manually review imagery of each property. Instead, AI-based algorithms can efficiently mine millions of geospatial images to produce the insights insurers need to assess risk, freeing up time and resources to leverage these insights in an impactful way. Of course, human verification is sometimes still needed, but AI can also flag properties that may need additional manual review so carriers can be confident in the results of their analytics. 

2. Many insurers still use inaccurate geospatial data

Geospatial tools like geocoding can help P&C carriers locate properties in their books of business, but many are still using geocoders that are only accurate 58% of the time. If these foundational geocodes are inaccurate, so are any risk assessments and insights derived from them. For example, an inaccurate geocoding engine could plot an address on a neighboring property, in the middle of a nearby field, or even miles away. One way carriers make up for these discrepancies is by supplementing their geocoding with manual imagery review, which we just established is not scalable across an entire book of business.

Fortunately, AI can not only analyze an image to determine if a geocode is correct, but also extract high-precision building footprints to provide carriers with a more detailed view of the property being analyzed. This is critical for insurers to realize the highest ROI from their tech stack, as an incorrect geocode can instantly render imagery analysis useless. Additionally, AI can place bounding boxes around each building on a property, facilitating stronger master data management (MDM) and better replacement cost estimation. This eliminates the need for employees to manually assess each property to understand the relationships between different structures or correct inaccurate geocodes, instead allowing them to develop more strategic reinsurance plans and data-driven underwriting practices.

3. Investing in imagery can be really expensive

Property insights gleaned from high-quality imagery are extremely valuable; however, the upfront expenses of capturing, purchasing, and manually analyzing imagery can quickly become cost-prohibitive for some carriers. To make sure imagery captured is fresh and high-resolution, insurers must invest in (or choose an imagery provider who invests in) quality cameras, frequent flights, experienced technicians, and similar expenses. While there is ROI to be expected, carriers incur a massive cost to obtain the best imagery, and are routinely challenged with squeezing the most they can out of this investment. That’s hard to do when teams are spending hours manually analyzing imagery and correcting inaccurate data.

AI-derived data that extracts characteristics of the imagery helps carriers get the most out of their investment by streamlining data creation and analysis. For instance, AI-based geocoding and building footprints enable insurers to properly geocode an address to its actual structure. If there are multiple buildings, AI can identify how many there are, their locations, and the unique identifier of those buildings - all related back to a master property record. This highlights the structures in question to fastrack analysis, and also standardizes the building notation across departments as each structure has its own unique ID that can be referenced.

4. Data redundancies are everywhere in P&C insurance

Standardizing data sources is top of mind for many P&C insurers, as many are experiencing high levels of tech bloat from the data redundancies existing across disparate departments. For example, the underwriting and claims group at the same carrier could be using different geocoding engines, address databases, or even geospatial imagery to perform their specific functions. This is not only unsustainable from a cost perspective, but also leads to discrepancies in property intelligence that can undermine the integrity of a book of business. 

The miscommunication across the insurance lifecycle can significantly increase operational costs. If the data is standardized across departments and unified through one MDM system, carriers can effectively reduce underwriting speed to claim settlement speed - both of which directly impacts an insurer’s combined ratio. However, the scale and efficiency provided by AI-driven imagery analysis empower carriers to unify data sources across their entire firm, eliminating data redundancies and establishing a single source of truth for operations.

AI-powered imagery analysis for scalable insurance workflows

Ecopia AI (Ecopia) has been working in the AI space for over a decade, constantly refining our imagery analysis algorithms to extract the most precise property data so carriers can be confident in their data-driven decision-making. In fact, top US P&C insurers trust our AI-based imagery analytics to bring together the information they need to optimize their workflows and improve their bottom line.

By using Ecopia’s comprehensive collection of AI-based building footprints, carriers can enter an address and instantly understand:

• the unique location of every building they insure;
• a standardized identifier for every structure;
• the condition of the property;
• and additional attributes that can fuel their risk analysis.

To learn more about how Ecopia is empowering P&C carriers with AI-based imagery analysis and insights, get in touch with our insurance team.