Case Study: Insurance tech provider delivers disruptive solutions, projects saving 40 percent with cloud

Few industries are more conservative than the insurance industry. So, when one of the industry’s leading global providers of technology and consulting solutions turns to cloud computing and encourages its customers to do the same, it’s time to take notice. Willis Towers Watson uses Microsoft Azure and Agile development to support innovative—even disruptive—solutions that wouldn’t be practical on other platforms and to make those applications available to a broad market segment. It expects to save up to 40 percent.

Situation

One of the world’s most conservative industries now faces one of its most disruptive challenges. The industry is insurance—in particular, auto insurance. The disruptive challenge is the usage-based insurance (UBI) model. And the most important company to watch as this transformation shakes out may well be Willis Towers Watson.

While not a household name outside of its industry, Willis Towers Watson is a leading force within it. Now in its second century, Willis Towers Watson is a trusted advisor to insurance companies that have come to depend upon its consulting and software solutions. They’ve used its software and guidance to implement a string of innovations over the decades, such as new ways to analyze risk, price auto insurance, and evaluate capital requirements.

Willis Towers Watson helped insurance companies worldwide to implement what is now the standard way to price auto policies. That method categorizes applicants based on predictive factors such as where they live, what they drive, how old they are, and what their prior claims experience looks like. It became the standard model because it worked well, and it was the best that technology could support.

Now, neither of those reasons remains true—or, at least, true enough. Telematics and mobile-phone technologies are now able to stream real-time data on every move that drivers make: how fast they drive, how hard they brake, even how they adapt driving behavior to changes in the weather. And having that much information about an individual driver gives an insurance company much better data for pricing a policy than merely knowing what demographic and other categories the driver fits into. Companies that offer usage-based insurance can price policies more competitively and more profitably than companies that only price them the traditional way. That’s what makes it disruptive. Finding a way to take advantage of UBI successfully, rather than succumbing to a more nimble UBI competitor, is what makes this development a challenge for insurers.

The opportunity for Willis Towers Watson is, ironically, to help its customers adopt UBI despite the inherent obstacles that implementation seems to pose. Processing and analyzing highly granular data feeds from automobiles, and integrating that information with third-party feeds that provide contextual information supporting the analytics, make for a massive computational workload. For many smaller insurers, the capital, operating expenses, and technical expertise needed to build and support a technology infrastructure for UBI are daunting if not prohibitive. And then there’s the risk of either overbuilding, and sitting on unused and unprofitable capacity, or underbuilding, and being unable to meet unexpected demand.

Willis Towers Watson wanted to provide the UBI technology infrastructure for its customers, so they could adopt the service without having to build and maintain it themselves. But that merely transferred the challenge of UBI onto Willis Towers Watson itself—and multiplied that challenge by the number of insurers the company would ultimately want to support.

Solution

It was a familiar challenge for the company. Willis Towers Watson had faced a similar need to provide its customers with highly scalable computing resources for another service, its Risk Agility FM service (formerly known as “MoSes”). Customers use Risk Agility FM to model all sorts of financial risk (see box, “The 189 Trillion Dollar Question,” below).

The company’s solution to support both Risk Agility FM and UBI is Microsoft Azure. “There has been a major evolution over the past few years in how we support big computing at Willis Towers Watson, and Microsoft plays a growing role in that support,” says Wayne Bullock, Global Leader of Life Insurance Software Development at Willis Towers Watson. “We’ve moved from on-premises implementations to hybrid deployments with the cloud, and we continually find enhancements in Azure that make it increasingly useful to us and our customers.”

Willis Towers Watson first prototyped a UBI solution three years ago, and has been working with customers on pilots and full production systems since then. Its initial solutions were hosted on parallel database appliances “that scale well beyond conventional relational database solutions,” according to Andy Lingard, Global Leader of General Insurance Software Development at Willis Towers Watson. “When we thought about preparing for millions of vehicles—we’re not there yet, but the number of vehicles is growing rapidly—the expense was extraordinarily high.”

That’s where Azure came in. Willis Towers Watson uses it as part of a hybrid UBI hosting solution. The company runs an on-premises data warehouse supporting an off-line R&D environment used to continuously evolve and improve the analytics running in the system. It maintains the bulk of the solution data in Azure Table Storage—data that includes trip and sensor information, and third-party data feeds on weather, road characteristics, and other factors that Willis Towers Watson includes in its analyses.

The analysis is conducted through an Azure-hosted application designed to scale to many hundreds of Azure Cloud Services worker roles. Scores and feedback are returned to an on-premises reporting application and a database running Microsoft SQL Server software, with results pushed out to insurers. Willis Towers Watson developed the solution using the Agile methodology and Microsoft development technologies that are increasingly popular at the company (see box, “Agile: The Method and Mindset Behind the Innovation,” below).

Willis Towers Watson had choices when it decided to extend its big-computing environment to the cloud. It chose Azure to gain several key benefits, according to Lingard. “Our customers have standardized on the Microsoft technology stack and so have we,” he says. “That makes Azure our go-to choice for the cloud. Also, our software pushes compute resources really hard, so working with a platform that supports us and our customers reliably is crucial.”

Benefits

Willis Towers Watson uses Microsoft Azure to gain competitive advantage, save costs, and broaden the market for its solutions.

Delivers Competitive Advantage
Lingard says that the power, reliability, and elastic scalability of Azure are crucial contributors to the competitive advantage that WillisTowers Watson gains from its implementation of UBI.

“Unlike some other UBI vendors, we’re not just counting driving events for covered vehicles,” he says. “We actually monitor position, speed, acceleration, and other factors at every second of every journey for every vehicle. We integrate granular environmental data. And we run analytics that are more sophisticated—and more predictive—than those of competitors, which make them more valuable to insurers. Azure is crucial to our ability to do this.”

Projects Saving up to 40 Percent
Azure is also crucial to the company’s ability to conduct those analytics cost-effectively. Willis Towers Watson estimates that using Azure saved it 20 percent in initial capital costs compared to building an entirely on-premises solution. Further, it forecasts saving between 30 and 40 percent in capital and operating costs as the number of vehicles scored by the solution develops into the millions.

Much of that savings will come from the company’s ability to provision its solution dynamically to meet fluctuations in demand without the need for additional hardware purchases and time-consuming deployment processes.

Broadens Market to New Customers
As disruptive as UBI is to the insurance industry, Lingard says that Willis Towers Watson’s UBI service, based on Azure, is even more disruptive—because it succeeds in bringing UBI within reach of many insurers who couldn’t afford, or choose not to build it on their own.

“Usage-based insurance can be an opportunity or a threat to insurers,” he says. “With a cost-effective UBI service hosted in Azure, we’re ensuring that they can see it as an opportunity.”

The 189 Trillion Dollar Question
The insurance industry is famously conservative—which, ironically, is a key reason that insurers are keen to embrace a technology that is anything but.

One of Willis Towers Watson’s key offerings is a platform for financial modeling, Risk Agility FM (formerly “MoSes”). Insurers use it, for example, to understand the financial reserves they need to maintain in order to cover their policies for decades to come. When insurers were smaller, risks lower, and regulations fewer, those models were relatively straightforward to devise. Back then, insurers typically ran the Willis Towers Watson software on a workstation. Today, that modeling is a far more complex proposition, and a single workstation can’t handle the load, especially not in a timely manner. So, it’s not unusual to find insurers running the company’s software in datacenters on scores or even hundreds of compute nodes. In this context, being conservative means an insurer needs sufficient technology resources to calculate the financial resources it needs to have on hand.

Back in 2008, Willis Towers Watson customers could run its software on grid networks running Windows HPC. In 2011 it was an early adopter of Azure. Now, the company is leading the industry again, this time in offering Risk Agility FM on the Azure Batch service. Azure Batch reduces the cost and complexity of running massive computational models in Azure by, for example, eliminating the need to configure and manage virtual machines or to include infrastructure elements such as head nodes. Willis Towers Watson offers insurers this version of Risk Agility FM as a simple subscription service suitable for virtually any insurer, regardless of size. It’s a service that can scale to tens of thousands of cores and support models with millions of policies.

To test the scalability of its software on Azure Batch, Willis Towers Watson ran combinations of large policy data models against numerous economic scenarios. One example model took six hours to run on a single workstation; on Azure Batch, it took two minutes. On a lark, the company decided to find out how much it would cost to insure each of the world’s approximately 7 billion people for $100,000. If calculated on a workstation, the model would take more than a decade to run; on Azure, using 50,000 cores, the calculation took just two hours. (The answer, by the way, is $189 trillion.)

Willis Towers Watson sees a variety of uses for its solution on Azure Batch. Insurers who have “bursty” compute needs—for example, those who see their compute needs peak to accommodate quarterly reserve runs—can avoid building extra, and expensive, datacenter capacity that will sit largely unused. Insurers can begin runs in their own datacenters and extend them to Azure Batch when needed. And small insurers can act like big ones without spending $1 million or more on a large datacenter.

But reduced cost is only one benefit of the new Willis Towers Watson option—and it may not even be the most important one. “I don’t expect insurers to take their gains as savings,” says Wayne Bullock, Global Leader of Life Insurance Software Development at Willis Towers Watson. “I expect most of them to reinvest the savings to run more scenarios and gain more-accurate projections. This is a conservative industry, after all.”

Agile: The Method and Mindset Behind the Innovation
Application innovation is at the core of what Willis Towers Watson does. And Agile development—which enables development teams to be more responsive to business needs—is at the core of its approach to application innovation.

“We’re building products that have never been built before,” says Sean Morley, Software Development Manager at Willis Towers Watson. “That has to be done through an iterative, evolutionary process, because when we envision these products, we can’t capture all the specs up front.”

Willis Towers Watson has been evolving its development process over several years. Now almost all of its insurance software products are developed according to an Agile methodology. The process of adopting Agile has involved a range of training initiatives, but perhaps the biggest effort has taken place where no one can see it: within the mind of each team member.

“The essence of adopting Agile is the individual’s understanding of what that means,” says Morley. “Team members have to think ‘How can I deliver quality today within this sprint.’ Some team members naturally have an Agile mindset. Others need mentoring. Either way, it’s a never-ending journey.”

Nor is it a journey for development teams alone. “The Agile process only works properly if the whole business embraces it,” says Morley. “The business leadership that has expectations for revenues and deliverables has to accept that if the business means what it says about Agile, then sometimes we have to change direction during development in response to changing conditions, and that has to be OK.”

The support of business leadership isn’t the only contributor to Willis Towers Watson’s success with Agile; its use of the Microsoft Visual Studio development system is another. “To implement Agile properly, you need visual indicators of who’s doing what, and what progress is being made,” says Morley. “You need to see Kanban charts and backlogs, and be able to predict when those backlogs will be resolved. The charts in the Visual Studio Team Foundation Server toolset enable us to do this.” Morley also cites the value of using Microsoft Build products for continuous integration of nightly builds; Microsoft Test Manager to test items in the Team Foundation Server repository; and the Microsoft Solution Framework (MSF) Agile Template to set up new projects quickly.