Insurers have made significant investments in improving their data assets over time. Large-scale IT projects are replacing legacy systems, data lakes and warehouses are being built, and new policy and claims administration systems have been installed. All with the promise of better data quality and greater data availability.
The industry now has thousands of pages of reports and/or KPIs produced daily. Advanced analytics has unlocked significant value for businesses, but the sheer volume of information gives rise to a new challenge: how to assimilate all the internal and external information we have on a daily or even instantaneous basis and use that information to improve our business. strategic? the decisions? How do we make our models smart enough so that they can inform us, and not just give us the results? Automation can help. It can sift through all the data and glean the nuggets of information that can help allocate the human resources we have.
Building a smarter model and leveraging AI all help at the microscale. With a specific line of business or this information, insurers can create models that are smart enough to tell what is most important and what is not. How do companies aggregate all lines of business and reporting in place? How do they bring out what is most important and pressing that requires their attention?
From the head of underwriting perspective, they want to know where the company is getting rates and where it is losing out to competitors. If the company earns too much, they are undervalued in all lines of business. This is the kind of information that is critical to a COU’s decision-making. It’s different from the CFO, who sees rates as the most important piece of information. Does the CFO want to know if reserves are behaving as expected or if the company is seeing an unfavorable change in reserves?
From the Chief Actuary’s perspective, the most important issue is where processes have been automated so that users are notified when things start to exceed established expectations. The automation can be run on a weekly or daily basis to be able to process this information.
For example, some companies, from a provisioning perspective, prefer not to review all of their classes quarterly. Automation can be built to monitor data as it comes in, where information is extracted, compare it to expectations that have been set, and where it deviates from tolerances. The system can be configured to notify the business when the system falls outside of these tolerances. This means companies can know that they need to invest time and effort to better allocate their resources when preparing their financial reports. They can also do things outside of the quarterly process to address things that are wrong before they get to the end of the quarter. They can take steps to mitigate what they see while making better use of resources.
New booking trends
Automation can also help a business provision by looking at risk at a more granular level – a very common trend today. Granularity has significant advantages. This means more homogeneous groupings of claims, making them more predictable. It is also useful outside of reservation, subscription and other broader areas.
Instead of striving to be more granular, organizations should seek to find the optimal level of granularity; reduce the granularity in some areas where it has no impact and intensify the granularity in areas where it makes all the difference with the least effort. These are new trends that companies haven’t thought of. How do companies best use their resources? It’s a new way of thinking. There is an element where companies want to be efficient with what they have, but menial labor does not add value. Companies need to focus on strategic work. A company is more likely to retain an employee who does strategic work, instead of manipulating data – this is an important consideration. Through automation, companies can level the playing field and replace people learning about the process.
If the process is manual and not well codified, it is difficult to train someone to understand the process. If it’s automated, that person just has to use their judgment; they don’t need to be an expert, with years of experience, they just need to understand the booking process. In other words, companies can use people to their full potential.
Companies with long-time underwriters know the unwritten rules – where to go, who to talk to; this level of experience cannot be replicated by someone new to the business. Instead of having so much knowledge and analytics embedded in people, there can be an automated way to surface that information and the experience gap can be bridged. As a result, companies have the best information available and no one is disadvantaged by newer or less familiar work processes.
The course for companies
The question is how automation can enable the efficient digestion of all available data and, through the automated use of analytics, bring actionable insights to the end user, focus the limited bandwidth of human resources and ensure that human capital is focused on work of the highest value?
Case studies show that automation and analytics can be used to significantly reduce manual effort while maintaining high quality or where automation can uncover new opportunities to improve the way insurers do their job.
The “automation journey” means insurers don’t have to be overwhelmed; they can start simply. For example, they don’t need the fanciest model, but can start by ranking and looking at the top decile/quartile or the top and bottom decile. It may also include next steps for using statistical tests and analytical models, how to monitor models over time, and use the results to allow machine learning algorithms to tune the models. Ultimately, insurers can use the results to train AI that can enable even higher levels of automation and insights in the future. On the underwriting side, companies should look to triage submissions.
Commercially, you may not be able to evaluate them all. The best way to tackle it is to be the most profitable and set the most important price first. The question becomes, how do companies measure to win, and how much does it take to be profitable?
Insurers must understand a given quote, to win this account, based on similar accounts, you must have the best price in the market. Businesses can create analytical models that help them understand the probability of winning or losing on the account. To some extent, if a company consistently wins, there are two reasons: it undervalues but is not likely to make money or it understands this niche better and can target good profitable risk and better pricing than his peers. Companies should enter the top quartile and not quote the bottom quartile; they have to set priorities. If a company only makes the top quartile, it needs a sufficient bonus base. Where do they break even? Companies need to support scale, to balance their expenses.
We are entering a new era of insurance where we have bigger and better data; we are going to find ways to exploit new data external to the company and this world of data will continue to grow.
The companies that will win will know how to exploit this data. Just being bigger or having more history won’t be enough. Companies will succeed if they use this widely available information, where they take this wealth of data and distill what is most important and as close to real time as possible. It’s not just the most external data, but who best uses that information to gain a competitive advantage.