<img height="1" width="1" style="display:none;" alt="" src="https://dc.ads.linkedin.com/collect/?pid=306561&amp;fmt=gif">

How Flock are using geospatial data to mitigate risk

With Ed Leon Klinger, CEO, Flock

“Being within the proximity to certain high risk areas; such as schools, prisons, or airports (being the obvious one) hospitals and public parks. Anywhere where the risk of the drone flying is influenced by the Geospatial data we collect"

Catherine King, Production & Content Director, Corinium Global Intelligence

One of the biggest selling points of Flock is the real-time data element. Could you tell me a bit more about what's involved with this feature?

Ed Leon Klinger, CEO, Flock

Yes, I certainly can.  So, in its essence, the Flock platform is driven by a risk intelligence engine which is our own technology that we've built over the last two years. The way that works is that we inject various source of real-time data and we combine that with various sources of static and historical data.  That data for example could include real-time weather conditions, so that's wind-speed, weather humidity, probability of rain-fall and so on.

drone1Being within the proximity to certain high risk areas; such as schools, prisons, or airports (being the obvious one) hospitals and public parks. Anywhere where the risk of the drone flying is influenced by the Geospatial data we collect. The risk of environment to the flight, and also the risk of the flight on the environment.  Imagine flying a drone on a particularly busy public area, you will probably be flying over people and densely populated area. You might be flying over busy roads. We collect and build our own models for various data sets that communicate the risk of the area.   We model population density with time, traffic conditions and so on. We collect all the real-time data and combine it with spatial data and geospatial data, and we combine it with historical data as well (historical data, such as the number of claims occurred). It is not just real-time data within the model but a whole array of data. Simultaneous we collect information about the drone that you're flying. We know how much they weigh, how fast and high they're flying, how many propellers they have and so on. We also combine this data with specifications of the drone from various sources. We are in close communications with every drone manufacturer worth knowing so we can have the best and most up-to-date information about the drone itself.

Finally, we collect information about the pilot. The experience, how many hours of flying they've done, any previous claims. If they have already been using our platform, we have a record of every flight they've ever taken - we feed all the different data we collect, some of them real-time, some of them historical, and some provided by the pilot into a large unstructured data set, into a set of algorithms that we've been building the last two years.

Now what we're going to do is calculate the real-time risk of the flight based off of that flight's take-off position. The way that we do that is by calculating the probability of a failure taking place (a failure being a drone falling out of the sky). We calculate the associated severity of that failure; so the probability, for example, this may take into account the probability of a failure due to a weather conditions leading to a drone falling out of the sky. The associated severity might take into account, the cost of colliding with something onsite or an individual on ground level. What we do is multiply severity by probability integrated into the area of the site, such that we have an expected cost of any given flight. So for insurance that's the expected claim from any given drone flight. That's the rawest calculation for calculating the risk of drone flight in real-time. We have partnered with Allianz, they then convert that calculation into a price.

Catherine King, Production & Content Director, Corinium Global Intelligence

Okay! Just to pick up on that point, with this calculation, probability and severity, what are the key challenges in doing that? There must be so many, but what are the key ones?

Ed Leon Klinger, CEO, Flock

One of the key challenges in that is we were one of the first companies in the world to build a calculation like this, specifically for the drone industry. As a result, there was very little to go on. We are pioneering this. We didn't have millions and millions of past datasets and previous research that we could use to build our own model. What we do is work with leading experts in the drone space and the insurance space and we built a large and well scale advisory board on both those spaces. We work with the insurer Allianz and we inject numerous amount of published research on this topic so we can build our risk analysis models ourselves. All of models are actually built in house. That is the first challenge, with building the models.

The second challenge is working with multiple data partners to actually ascertain the best quality data, so we can produce the best risk quality metric. We work with a wide variety of data partners in order to couple together all the data that we need so we'll be confident that our risk metric is something that world leading insurance company can put under their underwriting capacity behind. Obviously working with Allianz who expect the highest quality of analysis, to go into their products we are held accountable and we are proud to be held accountable for our risk metric because it's the core of what we do as a company. Building a model and obtaining the data are two of the key challenges.

Catherine King, Production & Content Director, Corinium Global Intelligence

With these different forms of data that you're using, especially the real-time element and calculation based on things like the weather, why do you think the other insurers and reinsurers aren't perhaps embracing this as much as you have done?

Ed Leon Klinger, CEO, Flock

Good question! I think the world of insurance is starting to wake up to step-change and accuracy in algorithmic pricing data1that could be achieved through this real-time data. I do think there is a shift taking place in insurance companies and starting to shape. However, as with FinTech what we're seeing is start-up and deep tech companies are showing the way here. They are committed to innovation. They thrive and live as a result of their innovation. They only succeed when they're at the fore-front. They don't have the safety net of being a large multi-national organization. They live and die by how good the products they launch are. I think start-ups also take larger risks, which the larger insurance companies don't like doing. Start-ups are leading the way on not just real-time data but cutting edge technologies more generally. I admit, there are a plethora of reasons as to why start-ups move faster and disrupt more than larger organizations. What we're seeing is an increasing number of partnerships in the InsurTech space between big corporate's and start-ups.  Especially with some of the insurance companies in London, New York, and Silicon Valley. We're starting to see a much larger multi-national organizations waking up the importance embracing comprehensive technologies and taking risks. They just understand that they're not particularly good at it themselves, so they find start-ups who can provide the technologies that are important.

Catherine King, Production & Content Director, Corinium Global Intelligence

At the moment you specialize in drone flight cover, do you think there is potential for what you're doing in this specific industry and moving into other products and other solutions?

Ed Leon Klinger, CEO, Flock

Yes, is the answer in a word. What we built at Flock is not one single technology, it's a stack of technologies. Which together creates a platform. What that platform does is inject a wide variety of real-time data, they feed that data into our risk intelligence analysis engine at Flock. They produce a  quantifiable metric that represents the risk of a single activity taking place at a given time in a given place. They then convert that risk metric into a price and which is used to sell a bespoke hypo-local insurance policy. They then build the platform and distribute that insurance policy in a meaningful way, indicating the Flock cover by producing documentation and emailing that documentation directly to the customer and producing that documentation within the app. The platform then allow users to interact with their insurance meaningfully as well. They can go back and see what the flight and all the activities they've undertaken. They can forecast future risk and plan future policies as well.

What we do at the back-end is actually extensive work on producing really detailed automated reporting and analytics. So we can retrospectively analyse the policy that's been sold. Map them out temporally and spatially, and garner insight from the data that we collected. We then present those findings on a daily, weekly, monthly basis to all of our partners. That's the whole technology stack that we built at Flock the last two years. In order to serve the distribution of what we believe to be the world's best insurance cover.

Now what we've noticed in asking these questions, it very much is that the stack can be repurposed for multiple industries- not just the drone industry. We have been bombarded with interest by the insurance industry more generally who have seen how quickly we have grown, having not just products in the drone space and are keen to unlock a similar level of growth in their own industry. It's not just growth in revenue, it's a reduction in claims and the quality of risk complication litigation service as an insurance. If claims go down, expenses go down as well. That is very attractive to insurers in whole load of industry not just drones.

Catherine King, Production & Content Director, Corinium Global Intelligence

So you mention that you worked with several sets of data, you have data partners. Geodata is quite well-known for being expensive, and the quality of Geodata appears to range quite massively. How do you decide what data you use for Flock?

Ed Leon Klinger, CEO, Flock

We take that very seriously. We basically undertake a data ROI. We want to understand if we invest in, not just money, but time and resources and we bring in new data sets, the material impact our data may have on the accuracy of our pricing. If deemed it's worthwhile bringing in new datasets based on not just the cost of today but the projected cost of the data in future when we have 1,000, 10,000, 100,000 customers on the platform we will understand the significant impact on the pricing and, therefore, the quality of the experience for insurer and the insured. Our general methodology is not to bring on data because of the quality, but because of the quality, the price that it's being sold at and the resources that it requires. There are many data that we plan to use in future but we don't right now. There are some that we use to work with that we replaced. We can replace data quickly because of the module-like-platform of the system, we feel we can plug out one data set and plug in data set in a matter of days rather than months.

Catherine King, Production & Content Director, Corinium Global Intelligence

Just one final question...Obviously you're going to be a Keynote Speaker at the GeoInsurance Europe Conference. Why do think that engaging in these types of conferences are really important?

Ed Leon Klinger, CEO, Flock

It's just generally enjoyable to talk about what we do. We built a phenomenal team of data scientist, engineers and insurance experts. We enjoy boasting about the work that we've done and showing off the work that we've done. It great to get out there and have recognition.

The second reason is it actually helps us to find brilliant people, with whom we can either work with or employ. We've used conferences to draw in people who are genuinely passionate about building tomorrow's technology. If there is a way we can get out there and showcase what we've done and hopefully entertain people and impress people with the work we've done, it's actually more often or not, how we actually ended up hiring the people who show the most interest.

The third reason, partnerships. We like to be very open about what it is we build; we've fostered an open culture at Flock. Open internally so that we share our mistakes, our shortcomings and our findings. But also open externally, we deliberately work in a way that allow people to internalize and showcase the work we've done. A: it allows us to get brilliant feedback from our exceptionally talented community especially at GeoInsurance Europe, we expect there to be very talented people present in the audience who will be able to provide us with any type of feedback on the work that we've done. B: because the technology that we're building captures many, many different industries. Right now, just in this industry we're already straddling the aviation sector and the insurance sector. By showcasing what we've built and explaining openly how it works, we're opening ourselves up to opportunities that we didn't even necessarily know existed.  The reason we became an InsurTech company in the first place was because we built a risk analysis platform that simply quantified risk and speaking around London at various conferences, we discovered that there was an insurance application buried within that. By speaking at events like those and drowning ourselves in a very engaged intelligent community we are actually exposing ourselves to new ideas, thoughts, and content that we might not have necessarily thought of.

geoint

You might also like...

© 2018 Corinium LLC