Current geocoding techniques create a hidden risk of $43B across the US.

Having an accurate understanding of location is a foundational prerequisite to understanding property risk and underwriting a policy. However, current parcel-based and street-based geocoding practices only offer a best guess regarding the location of each building you insure - guesses which we've found often contain significant inaccuracies. If you use these techniques in your current underwriting process, you may be estimating risk based on the wrong inputs - adding significant hidden risk to your portfolio, and uncertainty to your bottom line. Learn more about the severity of this risk, and best practices to minimize its impact on your portfolio with our free whitepaper.

Common errors with traditional geocoding systems

Even a few meters can make a world of difference in property insurance. It can be the difference between being classified in or out of a hazard zone - a material difference when considering policy pricing and willingness to underwrite. Parcel and street-based geocoding are designed to be ‘close enough’ - however, a simple analysis revealed that the use of this data for flood zone determination alone resulted in the miscategorization of over 1.6 million properties across the US.


Parcel and street-based geocoding datasets were created based on assumptions and interpolations. The resulting data assumes the primary building for each property is in the center of the parcel, or somewhere along the road. Without a complete and up-to-date database of every building across the US, it is not possible to attain rooftop-based accuracy consistently at scale.


Parcel-based geocoding providers are often missing addresses in their database since not all addresses are captured by the tax assessor. Street-based geocoding providers often fabricate “interpolated” addresses and place them between two known points to mimic completeness. However, both approaches often result in datasets which are incomplete and inaccurate.

Case Study: Flood Risk

Underpricing risks

Flood-determination analysis was conducted by comparing parcel-based geocoding against Ecopia’s building-based geocoding, which is built on the foundation of Ecopia's Building Footprints. The analysis revealed that over 1 million homes across the US are likely underpriced for flood risk. These properties, incorrectly identified as low-risk properties by parcel-based geocoding, have primary buildings that reside within a flood zone and are exposed to flood risk. These policy holders are underpaying for their insurance and could expose the portfolio to substantial claims.

Overpricing risks

Further, we found that over 600,000 properties were likely overpriced for flood insurance policies. These properties, incorrectly identified as high-risk properties by parcel-based geocoding, have primary buildings that are well outside of the flood zone, and not exposed to flood risk. These policy holders are overpaying and more likely to search for more fairly priced offerings from other insurers, leading to potential customer churn and lost revenue.

This zone-based analysis can be applied to other perils such as fire, hail, hurricanes, earthquakes, and landslides, to reveal underpricing and overpricing scenarios that arise as a result of parcel and street-based geocoding inaccuracies.

Where $43 billion of hidden property risk resides
across the US

By taking the average flood claim for each of the underpriced homes, the risk of underpriced policies alone can be estimated at $43B. This figure does not take into account the risk arising from other perils such as wildfire, hurricane, earthquake, etc. When considering these other perils, the risk is likely multiples higher.

This heat map illustrates where the highest concentration of underpriced and overpriced homeowner policies are located.

Ready to protect your portfolio?

Fill out this form to download a free copy of our white paper. Learn how the next generation of geocoding technology can be leveraged to significantly improve your risk assessment.

By signing up, you agree to our terms of use, privacy policy, and cookie policy.