Member Spotlight: Arturo
What does Arturo offer to the insurance industry?
Arturo derives property attributes and predictive analytics to support insurers across the policy lifecycle, including underwriting, portfolio risk, catastrophe response and claims assessment. By combining multispectral imagery (a collection of several image layers of the same scene), geographic data and customer portfolio insights with machine learning models, Arturo provides a single platform to deliver property intelligence.
Why are insurers looking for property intelligence solutions?
Three main areas have been impacting the insurance industry in recent years: climate change, inflation and COVID-19. Climate change is affecting weather events, making their impact more difficult to predict. At the same time, inflation is increasing labour and material costs. A combination of these, along with COVID-19 changing the work dynamic, have led to insurers needing to change their processes.
Imagery for remote property assessment was not readily available five years ago. In cases where it was commercially available, it was expensive. It is now more costly to send out inspectors to investigate a property for underwriting or claims assessment. Advancements in satellite imagery and machine learning have led to remote assessment being an accessible solution.
What property attributes do insurers find most challenging to find accurate data sources for?
It depends on the geography and the application of the data. One of the most challenging areas remains interpreting and contextualising data from raw imagery. It is relatively easy to understand property type and roof type. It is more difficult to understand how this affects risk. For example, what is the risk associated with a tiled roof in a hail-exposed part of Texas? Similarly, roof condition has been a hot topic in recent years – but what does “poor” roof condition mean for pricing?
Today’s technology has enabled a level of granularity which we never thought possible, such as understanding the precise square footage of rusting. The challenge in adopting and attributing this level of detail is knowing how it affects eventual losses.
There is a lot of property data out there, but pulling it all together and contextualising it to provide actionable insights is the hard part. Arturo aims to create a full view of the past, present and future of a property.
How do insurance clients use Arturo’s offering for its underwriting process?
One of the main areas is pricing and rating. Arturo can provide property data for a whole property portfolio or at an individual property level. Over 100 attributes are available including roof damage, roof condition detections, exterior property structures and solar panels. Arturo also analyses properties’ proximity to hazards, including flood and fire, to assist insurers with the underwriting process.
The other main area is pre-filling online quote forms with information about a potential customer’s property. Consumers can input their address, with property information then being autofilled through Arturo’s API. This allows consumers to instead just validate the information. In some cases, our insurance customers are so confident in the information we provide that they remove questions altogether. In both scenarios, this reduces manual input for consumers and improves their quoting experience. Arturo works with Australian insurers such as Suncorp and IAG on these underwriting processes.
We also help to improve the inspections process. With the ability to take those property insights and determine the risk level of a home, companies are adopting some degree of straight-through processing, meaning that they do not need to conduct on-site inspections. For those that do, the time the inspector spends at a home is optimised and focused on determining risks that humans are just better at determining.
How does Arturo help with the claims process?
We enable insurers to make claims decisions more quickly. Through remotely assessing property damage, insurers can decide whether they need to send out an adjuster to further inspect damage or if they can pay a claim straight away. One of our clients has been able to reduce the average time spent on each claim by 30 minutes, resulting in several million dollars of savings.
In particular, one of our data partnerships allows insurance companies to understand risk of claims before, during and after a catastrophic event. With the forecasted footprint of an event overlaid on the “resting risk” of a home, insurers can begin to understand their likely losses. During an event, the use of synthetic aperture radar can help understand the scale of damage as it’s occurring.
Afterwards, we can identify the areas damaged and in need of repair, helping claims departments to triage and prioritise their resources and adjusters. In this way, policyholders who are worse off can get help faster. Contractors know which tools and materials to acquire, so they arrive at an impacted property more prepared.
Are there ESG implications to Arturo’s offering?
Due to conversations we’ve had with our insurance clients, we now offer a sustainability module which can help insurers understand their scope 1 (direct) emissions. When customers use our solution, they reduce the need to send inspectors out to properties. This results in a material reduction in time spent, fuel consumed and carbon emissions. We show our clients how many trips they save, and therefore emissions they avoid or reduce, over a given period of time. They can create reports using this information and provide them to the relevant regulatory bodies or governments.
What is Arturo’s geographic coverage?
We cover all urban and suburban areas across the US and Australia. We use imagery acquired from aeroplanes, which offers higher resolution than satellite imagery. Whilst a lot of insights can be extracted from satellite imagery, it does not offer the granularity needed to accurately identify and understand whether shingles are missing from a roof, for example. We therefore provide data and insights on all areas that imagery-catching aircraft fly, which is over 70% of the US and the four major cities in Australia.
We also use satellite-derived synthetic aperture radar which allows us to “see through clouds” and determine impacts of flooding in real time. This, in conjunction with our aerial imagery, AI-derived insights and insurer information, helps to paint a complete picture of impact and risk.
How does Arturo use machine learning?
It starts with the raw images of a property. As we acquire these from aircraft flying back and forth, we combine these image slices to create a view of the entire property. We then label the different attributes, such as chimney and roof type, to help train our machine learning models. With the large volumes of imagery and data we receive, our models continue to get smarter. For example, we can identify roof type with 94% accuracy across the US.
However, we continue to improve and learn why the machine learning model has made mistakes in certain instances. For example, last year it identified a roof type as cement and it recently changed it to a tiled roof. When investigating why this was, we could see that last year the building was under construction and was using concrete slabs as part of the process. Now the building is complete, we can see that it is actually a tiled roof. Machine learning can make mistakes just like humans, so we are always looking to improve and learn.
How does Arturo approach data and technology partnerships?
Partnerships should be focused around trust, collaboration and flexibility. If these aspects are not present, it is more of a transaction than a partnership.
If an insurance client wants a property attribute that we do not currently offer, we partner with other companies to provide this. These companies need to be transparent with us. This is because governance committees within organisations may want us to prove the accuracy of our data. If this is the case, we also need to do this for the data we are obtaining from our partner companies.
Arturo already partners with companies including ICEYE, Vexcel Imaging and Lightbox. We are always looking for more sources of imagery and weather data, so we’d love to hear from any potential partners.
What is Arturo working on over the rest of 2023?
We are looking at geographic expansion: in particular, helping our clients with properties in rural locations. We are also exploring commercial property insights. We already offer some data and insights to our clients in this area, but we are looking to make this a more defined product offering over the second half of 2023.
What should readers do if they want to learn more?
Our website is a great place to learn more about what we do and access white papers that we’ve written. At Arturo we are looking to understand the problems that insurers are facing and create the technological solution to solve it. We’d love to hear from potential clients and the challenges they are facing – you can reach out to us by filling out the contact form on our website.