Property intelligence using Geospatial Data in P&C Insurance

property intelligence image

Remote property intelligence is now becoming a core part of processes in property and casualty insurance. The advent of AI and geospatial technology has provided us with a fast, reliable and socially-distant way to gather information and details about properties. Insurance carriers need to access various details of property at every stage of the process from customer onboarding to claims disbursement. These details are dimensional, geographical, situational, structural, and even historical. Traditionally, carriers have relied on physical surveys and audits which not only leads to a high expense & huge turnaround time to collect this information but more than that, there is no sure way to verify the information’s accuracy. Gaining property intelligence through remote ways is a solution to all these problems.   

What is remote property intelligence?

Remote property intelligence is a relatively new term that gained currency since the advent of advanced GIS technology. It is an application of geospatial imagery and AI to identify numerous property attributes. Some of these attributes are roof surface area, roof material, roof type, number of solar panels, distance from fire-prone vegetation. When these data points are gathered remotely from satellite and aerial imagery using AI-based feature extraction to enable better decision making, with the help of the generated insight, it is known as remote property intelligence. 

Who uses remote property intelligence?

Insurance carriers

P&C insurance carriers utilize property intelligence data for quoting, policy underwriting, risk assessments, fraud detection, and claim settlements. Each step in the workflow requires updated and accurate intelligence about the property being insured or assessed for claims. This helps insurance companies define policies, premiums, the sum insured, and evaluate and validate claims based on a proven, scientific approach.

Real estate companies

The entire real estate sector deals with buying, selling and leasing of properties which require detailed knowledge of the dimensions, geography, neighborhood, valuation etc. Many of these data points can be extracted using geospatial imagery and AI.

Public sector

Urban planning, emergency management, land management, public works, and other government parties require remote property intelligence on a large scale to plan and execute effectively.

Apart from these, there are many other sectors that need property intelligence to serve their purposes. These include but are not limited to landscaping, mobile ad-tech, construction industry, social welfare organizations and map developers

How is property intelligence used in P&C Insurance?

One of the most widely used, ‘textbook’ applications of property intelligence is in the P&C Insurance industry. Here, the biggest challenge is that carriers insure/assess thousands of properties every day and only through remote property intelligence, a swift and accurate ground-truth can be obtained for all of them. With geospatial imagery, one can also verify the accuracy of the information received, unlike physical audits where verifying information was extremely difficult. 

What does a remote property intelligence survey look like?

Below is a brief example of a remote property intelligence report template. 

Attributes required to prepare a quote Square footage/ Building footprint
Roof Geometry
Roof covering
Number of solar panels, presence of swimming pool/trampoline
Attributes required to underwrite a property Roof Geometry
Roof covering
Vegetation cover
Distance to fire-prone vegetation
Trampoline
Swimming Pool / Hot Tub
Solar Panel
Attributes required during insurance renewal Change in the material features listed above
property intelligence

What is property intelligence replacing in P&C insurance?

Insurance is a 200-year-old industry that heavily relies on data collection and analysis to evaluate risks and damages. Carriers traditionally operated by relying on in-person inspections, audits, and long-drawn surveys to collect data & associated risk of the carrier object. Even to verify a claim request, another round of audits would take place to inspect the extent of damage or verify whether it is a fraud request or a genuine request. Over time, what carriers started realizing was that these audits were time-consuming and filled with manual errors. In fact, soon the redundancy of data points and the general accuracy of datasets thus gathered, was found to be so low, that it became a huge pain point for the industry. The problem still remained that there was still an absence of alternative data collection mechanisms. However, all this changed with the advent of remote property intelligence. Backed by high resolution and updated geospatial imagery, remote property intelligence is replacing the traditional way of assessing properties or quantifying damages. Field data collection can be done remotely and transparently now with high accuracy and lower turnaround time. 

What are the advantages of property intelligence?

Some of the most important advantages of remote property intelligence are as follows-

  1. Insights are generated remotely thus simplifying the entire chain of data collection and decision making
  2. The turnaround time of getting relevant insights about a property is highly reduced
  3. Insights are extracted through expertly trained algorithms and manual quality assurance to establish high accuracy and veracity of the data
  4. In the case of disaster-prone sites, it eliminates any risk and exposure to ground personnel
  5. It utilizes remote sensing data sources, which makes it possible to capture ground data that might not be possible manually. 
  6. In case of any discrepancy, a source of ground-truth is always available as backup thus making these analyses extremely transparent 

To summarize the advantages, take the case in point of assessing a claim request for roof damage. The risk associated with obtaining a remote roof damage report using an aerial survey is significantly lower than by a person inspecting the roof physically. Additionally, the amount of time taken is lower and the accuracy of the inspection is higher. Lastly, the imagery saved during the aerial survey not only helps in claim settlements but can also be used as proof in case fraud detection/litigation of a false claim. 

How to set up a source of remote property intelligence for your carrier?

Property intelligence has two indispensable components – a geospatial imagery source and an artificially intelligent algorithm to interpret the imagery and reap intelligible insights as per the requirement.

Geospatial imagery

Finding out the right imagery for your area of interest requires expertise and experience. Satellite, aerial, and drone imageries are very popular. But there’s a catch. Apart from making a decision of choosing one of these three, there are many more factors that come into play. Geospatial imagery has numerous properties like resolution, positional accuracy, spectral coverage, vintage, etc. which influence the quality of data that can be generated from them. Depending on the intelligence that needs to be extracted, the right imagery needs to be chosen. 

Computer vision/AI models

Once the imagery is ready, then comes the critical step of developing and training an AI to extract useful information. However, at present, the uniqueness of the pain points of each insurance company is a major deterrent in applying a one-size-fits-all approach to the problem. As a result of this, the AI development has to be customized to suit the needs of the carrier. 

One way to build AI models is to hire an in-house team to work on it. However, there are various challenges in building an in-house team. Some of these are as follows-

  • A high lead time of at least a couple of years is required to build a talented team 
  • Data sources are available but they are in-consistent and have numerous gaps that need significant effort before they can be fed into the models
  • Carriers do not have ready offshore resources which leads to high data labeling cost
  • Developing precise AI at scale requires consistent dedication and ground truth validation which is again a multi-year process. 

These points highlight why it might make more sense for a carrier to outsource the AI development and training of models to an insurtech company. However, there are a few pitfalls that carriers need to look out for while navigating through these insurtech firms –

  • Custom model development is expensive as insurtech firms prefer to push their off-the-shelf datasets and models to the carriers
  • Most of the solutions on offer are fragmented and might not lead to a coherent AI strategy for a P&C insurance carrier in the long run
  • Very few insurtech firms are open to share ownership rights for the AI models developed
  • Transparency is a major concern as the insurtech firms rely on 3rd party imagery sources whereas the carrier will prefer to build models on the imagery content they have already licensed

If you represent a carrier and are reading this, what is clear is that setting up an AI practice to extract remote property intelligence is the need of the hour. However, it can be a very expensive proposition to build accurate and scalable AI models from scratch and it might be an even more expensive proposition, in the long run, to buy custom AI models from an insurtech firm. This creates a catch 22 situation for you. So how should you go about it?

Well, there is a way out!

Contact us to know how you can build an AI practice within your carrier in a cost-effective way and integrate custom property intelligence models into your carrier’s workflows.

Tags: , ,

Leave a Reply

Your email address will not be published. Required fields are marked *