Insurance data analytics: geospatial intelligence for New Zealand actuaries
AI-derived roof age, condition and hazard attributes feed actuarial models, improving risk segmentation and reducing premium leakage for NZ insurers.
)
Insurance data analytics software is the set of tools a property and casualty (P&C) insurer uses to collect, structure, and model risk data for pricing, portfolio management, and capital decisions. For an actuarial team, the category includes the data feeds that supply property attributes to rating models as well as the platforms that process those attributes.
This article covers how geospatial data supports actuarial risk modelling and how the Nearmap data analytics platform supplies AI-derived property attributes and Portfolio Intelligence to pricing teams. It then sets out how New Zealand regulation and catastrophe exposure shape the data requirement, and documents results from two insurers.
What is insurance data analytics software?
Insurance data analytics software combines data ingestion, modelling, and reporting so that a P&C insurer can price risk, segment a portfolio, and monitor loss experience in one environment. Actuarial teams run it alongside pricing engines and policy administration systems, and the accuracy of its output depends on the property data supplied to it.
The shift from statistical modelling to property-attribute risk pricing
Rating built on postcode-level and self-declared factors describes a group of properties and leaves the individual building unmeasured. Property-attribute risk pricing adds measured characteristics of each structure, such as roof material, roof condition, and surrounding vegetation, so that two properties in the same postcode can carry different rates for evidence-based reasons.
The Insurance Council of Australia has described the same gap in flood modelling, where models often rely on flood depth at the centre of a parcel and omit both the position of the building on the land and its habitable floor height. The Council recommended that those data points be prioritised for capture.
How geospatial data improves actuarial models
Geospatial data describes where a property sits and what it physically contains in a form that a model can ingest at scale. Nearmap refreshes aerial capture annually in New Zealand, so the attributes derived from it follow changes to the property over time and give an actuarial model the roof as it stands today.
Higher-quality inputs improve model performance during development, where candidate rating variables are tested against loss experience, and in production, where the same attributes are retrieved at quote and renewal.
Nearmap property intelligence for actuarial teams
The Nearmap data analytics platform delivers AI-derived property attributes to actuarial teams through APIs and through Betterview, the Nearmap property intelligence platform for P&C insurers. Nearmap owns the aerial capture program that feeds the platform, which keeps the quality and timing of every attribute under one owner.
AI-derived property attributes: roof age, condition, material and slope
Nearmap delivers more than 130 AI property attributes, among them Roof Age, Roof Condition, roof material, and roof slope, which together describe how a roof is likely to perform in wind, hail, and storm. Each attribute is extracted from imagery captured over the building, so its value updates with each new capture.
Actuarial teams can test these attributes against claims frequency and severity as candidate rating variables. Attributes that show lift in the loss data then move into the rating structure.
Vegetation, hazard zone and proximity data
Vegetation and defensible space measures describe the fuel around a structure, and proximity measures describe the distance to adjacent buildings and vegetation, both of which bear on bushfire exposure. Betterview Partner Connect adds third-party hazard data to the same property record, so an actuary reads hazard zone and property attributes together.
Portfolio Intelligence for risk segmentation at scale
Portfolio Intelligence is a pre-built tool in the Betterview platform that aggregates AI-derived risk detections and scores across an insurer’s policies in force. Actuarial teams view concentrations of roof condition and high-peril properties on a map, filter the portfolio by roof condition, third-party data, and custom risk flags, and drill down to the individual property.
The result is risk segmentation across a large property portfolio without manual aggregation of data from several sources.
Actuarial tools and workflows
Integrating Nearmap data into pricing models
Nearmap delivers attributes through APIs, so an actuarial team can match them to policy and claims records to test candidate rating variables, then retrieve the same attributes at quote for production rating. Betterview connects to core policy systems, including Duck Creek Policy and, through the Nearmap Accelerator for the Betterview Platform, Guidewire PolicyCenter, which places the attributes in the system where underwriters and rating engines already operate.
The same attributes support AI underwriting, which applies machine learning to property data at the point of decision, so the variables an actuary validates in a model also reach the underwriter reviewing the risk.
Identifying and correcting premium leakage
Premium leakage is the revenue lost when a policy is priced below the risk it carries, and it commonly arises when the rating record holds missing or outdated property data. Geospatial property data reduces leakage by exposing the gap between recorded and observed property condition.
An actuarial team can segment the portfolio by Roof Condition and Roof Age, compare average premium and loss ratio within each segment, and identify cohorts where observed risk exceeds the rate charged. Corrections then follow through rating factor changes, underwriting referral rules, or renewal actions on the affected properties.
Reinsurance reporting and capital allocation
Reinsurance purchasing and capital allocation depend on how accurately an insurer can describe the accumulation of natural peril exposure in its portfolio. Under the Australian Prudential Regulation Authority (APRA) standard GPS 116, an insurer must hold capital against a 1-in-200 year loss measured across the whole portfolio, and reinsurance that meets APRA criteria can reduce that requirement.
Portfolio Intelligence aggregates property attributes and risk scores across policies in force, which gives reinsurance and capital teams a property-level basis for exposure summaries and accumulation views by hazard.
Case studies: improving loss ratios with property intelligence
Ohio Mutual Insurance Group, a US regional carrier, reduced inspection costs by 62% within two years of using Betterview to flag its book by roof condition. The two case studies below document loss ratio results from the same platform.
Arden Insurance Services: underwriting time and loss ratio improvement
Arden Insurance Services is a managing general agency that insures residential condominium associations in the Western United States and certain Midwest states, and it relied on post-bind physical inspections to confirm underwriting decisions while the business grew quickly. Within several months of adopting Betterview, Arden improved its loss ratio by 5.5% and cut average underwriting time per property by 9%, and the underwriting team handled three or more additional submissions per week.
Frederick Mutual: 27% direct loss ratio improvement
Frederick Mutual Insurance Company, founded in 1843 and writing commercial lines across five US states and Washington, D.C., adopted Betterview in 2019 after identifying that it lacked an accurate view of roof condition. The carrier reduced its loss ratio by 26.8% year on year and has held an average loss and loss adjustment expense ratio of 35.6% over five years despite a higher risk of severe weather events. Historical imagery also lets the carrier verify whether a loss occurred on a roof during the policy term.