As part of our commitment to serving as a key business resource for the Insurance profession, Insurance Business is offering this guide on integrating intelligent automation into your organisation.
Insurance is rapidly becoming a real-time, data-centric service stream that demands technology built for its environment.
Intelligent automation enables an evolution from traditional approaches to insurance, and allows companies to serve their customers more effectively, and with greater efficiency.
This article features exclusive interviews with technology leaders working in the insurance space, and reveals what they’re doing with intelligent automation to create a superior customer experience.
Read now and gain insight into:
- The value of automation during times of change
- Why flexibility and scalability are a must for insurance organisations
- The business case for automation
The real-time opportunity
Enhancing the process
Fine-tuning the engine
Turning the industry on its head
ML and AI
Understanding the data
The business case
This article draws on exclusive interviews with technology leaders working in insurance, revealing what they’re doing with intelligent automation and how this is benefitting the customer.
It also explores the business case for using automation to understand and serve the customer more effectively – enabling greater efficiency, better response times, and, ultimately, a superior customer experience.
Finally, we examine the value of automation during times of change, when flexibility and scalability are a must.
The real-time opportunity
Insurance is an ancient industry that is being transformed by data and new technology.
In decades past, insurance was based on what could be called ‘monolithic data’ – static blocks of knowledge about a customer, including factors such as age and health, or potential risk to an insured item. But today, data can be gathered in real time from reams of different sources and devices. This changes the game and puts the customer front and centre of insurance decisions. Critically, it also speeds up and improves insurers’ internal processes.
Insurance is fast becoming a real-time, data-centric service stream, and this demands technology built for that environment. Intelligent automation is a natural fit for the sector, enabling both an evolution of traditional approaches to insurance, and a real transformation of it.
For insurers and their customers, intelligent automation’s promise is to enhance the way the industry serves customers and technology serves the business. But what is it?
Enhancing the process
The innovation lies in the fact that repetitive, day-to-day workflows can be enhanced by artificial intelligence (AI) and machine learning (ML). These can then be integrated into existing technology and data stacks via digital employees – software robots that can acquire specialist skills.
By automating essential but low-value-adding tasks in this way, human workers are free to focus on the people-focused tasks that they are better suited to, as they work creatively and collaboratively to drive the business forward.
By automating what happens behind the scenes in a smart, data-centric way, insurers can also create a frictionless experience for the customer across different services and claims, one that is more finely tuned to their individual needs.
So which companies are using these technologies?
By Miles is one of a new breed of data-centric insurance start-ups, founded five years ago to provide pay-by-mile insurance for drivers, based on the number of miles they drive month by month. In this way, drivers who don’t use their vehicles often are not penalised by steep premiums.
In theory, everyone benefits, including the environment by discouraging the unnecessary use of private cars in cities.
Such evolutions are inevitable as transport increasingly becomes an on-demand service itself, with similar usage-based insurance ventures emerging for autonomous vehicles and drones (such as UK start-up Flock). Over time, other traditional models such as fuel duty and road tax may also need to change to a more usage-based system.
By Bits is a B2B insurtech start-up that has emerged alongside By Miles, run by the same management team. It allows rival insurers to build similar usage-based offerings of their own, using By Bits’ software. Callum Rimmer is the co-founder and CTO of both companies. He explains that, while driving remains the preserve of humans for the time being, data-fuelled insurance providers are already encouraging vehicle owners to behave more responsibly.
“If you’re on the motorway in autonomous driving mode, you’re inherently safer than if you’re not,” he explains. “So, the idea of understanding when you turn that feature on means that you should get a premium saving. And all these little changes are making the usage of your vehicle become more respectful of the risk you should be paying for.”
Fine-tuning the engine
But to get to the point of insurance being a real-time, data-informed service demands automation and seamless integration behind the scenes. This is where AI, ML, digital employees, and other aspects of intelligent automation come into play.
In the case of providers such as By Miles, this is initially via in-car cellular telematics devices, which enable insurers to reconstruct a driver’s journey and calculate a risk-based premium. This is then charged to the customer, automatically and transparently, via a mobile app.
“Automation is essential for this, because you’re not going to have the real-time aspect of turnaround quick enough if there’s anything that isn’t automated in that process. We’ve got 100 million insured miles, and we’re doing millions more insured miles every week. So, there’s a huge volume of data that we need to collect, automate, analyse, price against, and report back out.”
Turning the industry on its head
Rimmer explains that his company is using this data to flip the traditional approach to driver insurance: instead of using telematics to penalise young or high-risk drivers, By Miles is using it to reward the normal- or average-risk drivers who use their cars less frequently with lower premiums.
Other elements of the service are backed by intelligent automation too, he explains, including risk analysis and underwriting.
“If you don’t have a real-time understanding of that volume of collected premium data, then your understanding of the risk that you have on the books isn’t correct.
“So, there’s a lot of very important accounting that needs to happen that gets automated as well. That means external systems stay in state with what the customer is saying, or our system is saying. There’s a lot going on behind the scenes to make it happen.
“We’re not full stack insurers – we’re not an Aviva or an Admiral, we’re a managing general agent (MGA). Our insurance partners give us a premium, we decide on the price and everything to do with how it’s facilitated from a product point of view, but the underlying risk is taken on by an insurance partner.”
James Gardiner is Head of Technology, Data and MI, for Admiral, which puts him in charge of the strategic direction of the full-service insurance giant when it comes to analytics and intelligence. The key difference is that Admiral itself is the core customer of his team’s business insights and data, and automation is core to that relationship.
“We work in what we call ‘tribes’ and ‘squads’ – the data warehouse squad, and so on. Essentially, we must implement some insight on how those various tiers are working for motor insurance, for example – how successful the uptake has been, that sort of thing. When you get something like that, the business wants reporting and insight at the drop of a hat; they really need it very quickly. This is where automation comes into the mix for us.”
For Admiral, there are two sides to automation: development and testing.
“Someone comes up with a bright idea for a new product and, to deliver that insight, we’ve got to be as quick and as quality-driven as possible. There’s no point in delivering insight if the figures are wrong.
“One of the keys to doing that is to look at a lot of the tasks we do and we question ourselves all the time, ‘Can we automate this? Because it is taking too long’.
“Part of what we do is behaviour-driven development [BDL]. What these tools do for us is put requirements into a comprehensive written statement. So, if I make a sale, I want it to appear on a report with the product that was sold and the customer it was sold to.
“And because we write it in a structured format like that, what we’ve found is we can push the testing elements of our delivery to the left. We get an SME, someone from the business who knows exactly what they want, and we get a tester and a developer.
“The developer will start developing what the SME’s requirements are, and they’ll start coding that up. And the tester will understand what those SME requirements are and know how to test them and know what’s important.
“And when the developer has finished, it goes through an automated release cycle. It schedules all our releases and our tests, then sends an email saying, ‘Sorry it’s failed’, or ‘Yes, it’s succeeded’.”
Like every aspect of financial services in the 21st century, insurance is heavily regulated, which is another area where intelligent automation is now critical, by ensuring that every box is ticked and every industry requirement is satisfied. “Understanding the risk that you have at any point in time is critical”, says By Miles’ Rimmer.
But for By Miles and for majors such as Admiral, intelligent automation has one overriding purpose: to save time within the enterprise. “What we try to do is free up time, to make something that’s either more innovative or better for the customer – self-service policy changes, for example,” says Rimmer.
“We’re more interested in retention and NPS scores and keeping our customers happy. And capitalising on the adventure of what will happen in a semi-autonomous and autonomous driving world and trying to keep ourselves relevant.
“It also enables us to understand where and when they’re driving, and proactively communicate to them that maybe they should try a different route or try to drive at a different time of day.
“Not so much because it’s good for us as an insurer, but because we can tell them they’re pumping out too much air pollution here, or are more likely to get caught in congestion there, or their premium will be lower if they go into fewer hotspots.
“Using the real-time information that we have is an advantage to us as a business compared to most other insurers, giving things back to the customer that improve their driving experience. Customers are more and more willing to give away data if they can see the benefits back, in the form of a premium saving or a better experience.”
RSA is another long-established giant on the insurance block. Automation is a core element of its strategy and keystone programmes, particularly around the automation of repeatable processes, explains CIO David Germain.
“Things such as streamlining and simplifying the customer journey, enabling greater selfservice. This is something that is being increasingly demanded in the back-office to enhance our testing, maintenance, and security capabilities, or to reduce processing time to minimise risk and disruption to our customers.”
Within Claims, RSA uses automated scripts across commercial and personal lines, which has driven down the amount of manual processing by more than 45%. It has also improved the quality of the firm’s output by proactively identifying and resolving issues within automation scripts and automatically notifying user groups.
“One of our automations compares the many documents generated as part of the claims process and highlights any variations,” Germain explains. “What was once a manual process and took days to complete now takes a matter of minutes, which means a much quicker response time for our customers.
“Another automation triages calls to ensure our customers reach the correct handling team as quickly as possible, which has reduced the cycle time by more than 20%.”
Through automation, RSA was able to reduce processing time for some claims from five days for 1,000 claims to just 45 minutes. As was the case for By Miles, the customer has benefited from this, says Germain.
“Taking one of our more recent examples where we have improved the customer journey across some of our core personal lines products. We have seen an increase in traffic to various platforms. As a result, our customer-facing support teams are more available for those customers who need specialist assistance, since most our customers’ needs can be met by our system processes.”
Overall, RSA’s customers benefit from automation as it improves their experience through the consistency of their interactions with the company, from the increased availability of services across multiple devices, and by speeding up all processes.
“Customers are not having to wait as long for their claim to be processed, as we have cut the processing time down dramatically via automation,” he says.
ML and AI
But what about machine learning and AI?
For By Miles, automating processes via AI and ML is similar to what an actuary does: “give it data and it trains itself,” he explains. “It uses experience and past data in order to improve its predictability. Then it publishes what it does, looks at the performance, and then iterates over it. It’s exactly what an actuary does, but you do it programmatically.
“It’s an obvious thing for insurers to do, to introduce machine learning in order for them to price better, to lower their operational costs because they don’t need as many actuaries, for example.”
But simply using automation to price better than other companies on a comparison website is an attractive, but short-term strategy, he explains.
“You’re getting data in real time about how people are driving, where they’re driving, the features they’ve got turned on in the car. And all this extra data is perfectly made for machine learning. It allows you to start building models and automation, which mean you can price better and in a more sophisticated way.”
Understanding the data
“So, I think the biggest changes in automation are going to be not so much the processes you use, or simply adopting things like machine learning, but understanding the data you need to take in. There are things that you need to be getting information out of, and you need to be automating that so as not to stymie innovation.
“We’re already at level 3 out of 5 for automation in cars and progress in car technology on the roads. And it’s at the stickiest point because we’re moving from the driver to the car. And in that switch, insurance really isn’t in a very good place to be able to understand how to price for that risk, or where the risk sits.
“So, the biggest changes in automation and the processes that you’re going to need to build are going to be in getting comfortable with that step. And the easiest way to do that is to look at a usage-based proposition.”
The business case
RSA’s Germain agrees that the real benefits of intelligent automation go far beyond cost savings and process efficiency to informing better decisions, improving risk selection and productivity, and facilitating product and service innovation.
But how do insurers go about establishing and measuring the business case for automation? There are four key elements, he says.
- Risk: Automation can eliminate the risk of human error, which can lead to customer harm.
- Productivity: Automation can improve the speed at which you can implement change, bringing new capabilities to your customers faster.
- Cost: Automation removes manual processes that are resource heavy and drive down the value proposition for your customers.
- Innovation: With your teams spending less time on making changes, there is more time to focus on customer interactions and product improvements which drives greater benefits and value to your customers.
For Admiral, much of its future direction will be into the cloud. But business culture, integration and traditional DevOps challenges remain.
“I think one of the problems with automation is you’ve got DevOps, but you’ve also got this DataOps initiative, which is like the data kitchen. But data isn’t as easy to automate as code.
“It involves the same principles, but there is a different set of challenges. You get bad data, people put the wrong credentials in, they put the wrong postcodes in, and so on, so there are data quality issues.
“We can’t push our live data into our test and dev environment, because of data protection. So, what we do now is we’ve got tools that take a copy of live data and mask all the personal details in it, so we can no longer recognise anyone or identify any individuals.
“And then we will subset that, make it smaller. We start coding around the intricacies of this bad data, malformed data, and we’ve got actual real-life examples, and that’s something that we also automate.
“One thing I’ve learnt is you need to build a team of good people who can understand this technology, and you need to ringfence them, and you need them to focus. We have a separate squad that works on enabling automation functionality.
“You have to focus on it, you have to take it seriously. It can’t be something you just do at the side of the desk.”