View Samples
View Samples

Leveraging Ecopia AI's Building-Based Geocoding to Enhance Replacement Cost and Risk Estimation

Exploring how Tokio Marine North America Services (TMNAS) leverages Ecopia AI’s Building-Based Geocoding to transform risk estimation and underwriting.

For Property & Casualty (P&C) insurance carriers, location intelligence starts with the foundational ability to accurately understand the number and location of individual buildings at each address. 

This insight is essential for two key functions: (1) understanding replacement cost estimates for each building at a given property; and (2) assessing the respective risk of these buildings. This is because, following typical convention, an insurance policy should be priced based on the respective value and risk of each building at the insured’s address. However, historically, P&C carriers have lacked access to reliable data to determine these factors independently – instead, relying on aggregated address-based estimates often provided by an agent or the insured.

This case study will highlight the experience of Tokio Marine North America Services (TMNAS) surrounding the risks of relying on an aggregated address-based database, and the transformative benefits of a building-based database made possible through Ecopia’s Building-Based Geocoding – the most comprehensive, accurate, and up-to-date representation of buildings and related addresses across the United States.

The Risks of Address-Based Underwriting

Like many insurers, the historical foundation of the TMNAS underwriting and risk assessment workflows was an address database that tied each building on a policy to an address and ultimately to a single coordinate. The challenge with this system is that the data (provided by the agent or insured) often aggregated the property into a single “location” without providing the ability to individually calculate or validate the reconstruction cost or risk of each insured building. Given that across the United States, 45% of occupied land parcels have more than one building on the respective land parcel, this dynamic results in a very complex challenge when underwriting across the country. An extreme example below shows an address reported as a single “location” by the agent, despite the presence of 104 buildings on the property.

A single address marked as one “location”, despite having 104 buildings on the property
A single address marked as one “location”, despite having 104 buildings on the property

Traditionally, there would have been a single address point associated with this policy (either in the middle of the land parcel, or along the road), and the TMNAS team may not have been able to determine the number of buildings except that the limit would indicate the potential for a group of buildings (For this property, 104 buildings were only flagged once this address utilized Ecopia’s Building-Based Geocoding system).

The inability to consistently identify the number, location and estimated square footage of buildings insured results in major gaps in estimating value and related risk. Without this information, carriers can find themselves paying much higher replacement costs for insured properties than expected. For TMNAS, these challenges resulted in not only increased claims costs but also reduced premiums where values were understated.

Better Balance Portfolio Risk with Building-Based Underwriting

To fill the gaps of unreliable replacement cost values, TMNAS leveraged Ecopia AI’s Building-Based Geocoding solution to make a transformational shift from an address-based database to a building-based database. Ecopia's Building-Based Geocoding offers the first and only complete building footprint collection in the USA, paired with best-in-class address data. This unique offering was generated by leveraging artificial intelligence to mine the most up-to-date geospatial imagery available - creating a unique source of truth surrounding every building in the United States.

By gaining access to this solution, TMNAS was able to identify all insured buildings on each property. Specifically, integrating Ecopia AI’s data into TMNAS internal tools and workflows enabled the determination of replacement cost and related risk on a building-by-building basis (instead of an aggregated address-by-address basis). The example below highlights a single address with 14 different buildings on the property – each now having been assigned a unique replacement cost value and respective risk.

A single address now identified as containing 14 different insured buildings, Based on Ecopia’s Building-Based Geocoding
A single address now identified as containing 14 different insured buildings, Based on Ecopia’s Building-Based Geocoding

As a result of this integration, TMNAS will be able to better estimate and balance the insurance-to-value ratio across their portfolio. Further, once any differences versus historical data are identified, there is now an opportunity to increase premiums upon renewal, or balance-out portfolio risk. In addition, by leveraging Ecopia’s Building-Based Geocoding, TMNAS can now determine additional peril-based measures of exposure on a building-by-building basis, such as estimated insured roof area when considering hail risk.‍

“As a result of the integration of Ecopia’s Building-Based Geocoding, TMNAS was able to move to a true building-based database of policies, resulting in the ability to better estimate reconstruction cost and risk.” – John Ferraro, SVP & Pricing Actuary at Tokio Marine North America Services

Building upon this foundation, TMNAS has the opportunity to expand its work through this partnership, exploring the additional property intelligence information products that Ecopia is building across the entire US.‍

To learn more about Ecopia’s Building-Based Geocoding product, or how we are supporting enhanced risk assessment and underwriting across the P&C insurance industry, contact our insurance team.

Learn more about Ecopia's insurance solutions

Ready to get started?

If you're ready to leverage groundbreaking advancements in artificial intelligence, let's chat.

Let's talk