No one likes to think of their company as being behind the technology curve, but those insurers operating without benefit of an advanced analytics system are doing exactly that.
Scott Diekmann has a simple explanation for why his company, Premier Prizm, is currently in the process of installing an analytics system it licensed from Innovation Group.
“We are changing our analytics approach from putting numbers together and passing them to someone to figure out what the problems are and how to fix them—to this system that will tell [business users] where a problem is that needs to be fixed. This aligns with the industry moving from a reactive to a proactive mindset,” says Diekmann, business analyst for Premier Prizm.
The number of insurers behind the curve appears to be shrinking, though—at least in the mind of Jose Trasancos, who recently joined Utica National Insurance as senior vice president and senior personal lines officer.
“It typically starts out in the pricing or product development areas, with modeling applied to product design,” he says. “Organizations also have started to apply analytics to process improvement in areas like claims and underwriting. The pace of adoption has increased in the past five or six years.”
Trasancos, who recently left a position with Narragansett Bay Insurance Company, has been working in analytics in the insurance industry for about 25 years, cutting his teeth during a stint at Progressive, which he describes as “a wonderful environment. It was a learning organization, principally interested in making every effort to find out why things behave the way they do.”
Karen Pauli, research director for TowerGroup, believes the adoption of analytics software has been creeping along over the past decade, but she maintains that once an organization sees the value of the software it begins to create a force unto itself.
Pauli adds that carriers need to understand the differences between analytics and predictive analytics.
“Carriers are using analytics to look at their operations and once they get comfortable with that toolset, the providers of the analytics software also have excellent competencies around predictive analytics,” she says. “Now you get business people who are hungry for more business value with more accurate pricing and decisioning. An analytics-driven business organization is where you have to be. Predictive analytics is a high art form of that.”
Frank Petersmark, CIO advocate for X by 2, believes the software providers that offer predictive modeling and predictive analytics have products that are maturing.
“By maturing, I mean the vendors are coming up with solution packages—configurable-type platforms—that are going to allow carriers to mine some of that gold,” he says.
Trasancos maintains that to the extent that analytics are applied today by more insurance companies, much of the industry’s success has to do with software packages that are easier for insurers to use and relate to.
“Things have changed dramatically,” he says. “Today there are much more intuitive packages available for users who perhaps aren’t as technical, but analytically very capable. They find [the software] much easier to use and apply. Packages like SAS Enterprise Guide and others have brought analytics to the desktop of people that aren’t necessarily programmers.”
Breadth of capability made the SAS products stand out for Trasancos. “Not only are they quite strong in terms of statistical procedures, but they are end-to-end ETL tools that allow you to do the extraction, translate, and execute the outcome from one platform,” he says.
MATURITY LEVEL: INSURERS
On the carrier side, the maturity level is another story, points out Petersmark.
“If you game me a scale, I’d say the maturity level for vendors is medium to high, but if you ask for the maturity level of insurers I’d say low edging toward medium,” he says. “There are some issues there, not the least of which is the quality of data. It was true 20 years ago; it’s still true today. The quality of your data and the way it is organized, its accessibility, and the ability of carriers to make it actionable remain big issues.”
Pauli agrees that insurers have yet to reach a maturity level in their use of analytics, but she adds that often depends on a carrier’s product line.
“When we look at personal auto and homeowners, there definitely is maturity there. They have straight-through processing using analytics that don’t involve underwriters looking at anything—with just a few exceptions,” she says.
From TowerGroup’s perspective, Pauli sees adoption of analytics following the same road as the adoption of core systems.
“We’ve had some strong policy administration system adoption for personal lines and behind that comes the analytics,” she says. “It will follow the same pattern in the claims area. We have the adoption of some new claims systems and once this happens companies will begin to see the power of their information.”
Trasancos believes every time he recalibrates a pricing model he learns something new.
“Natural systems do change their behavior, but they do so gradually,” he says. “This has to do with being disciplined about recalibrating models you’ve already deployed. There are always things to learn.”
Trasancos feels there are points that are intuitively obvious, but nonetheless he finds it gratifying to see those points in the data once an analysis has been executed. For example, in the context of homeowners’ coverage, something seemingly as innocuous as the number of bathrooms in a home turns out to be predictive of future loss.
“Most household disasters have to do with a plumbing failure of some sort and those plumbing failures tend to happen in bathrooms,” he says. “If you use the number of bathrooms as a proxy for the number of potential failure points and you analyze the data and introduce that variable into an existing model, you suddenly start to see an improvement in pricing accuracy.”
When discussing results such as these objectively, Trasancos finds there is a great deal of acceptance among business users, but he adds that within the context of an organization, change control becomes a component.
“The development of tiering mechanisms and more capable pricing models has forced a level of change onto the insurance industry that’s required some management—particularly in areas of underwriting,” he says. “There can be some organizational skepticism, but it’s typically related to change control.”
“It’s fair to say a lot of companies are still going through the modernization cycle,” says Petersmark. “It could be the case that companies either have already bought an analytics system so they don’t need more investment, or they don’t quite understand the potential competitive differentiation that analytics holds for them.”
Petersmark agrees that using analytics can be a cultural change within an insurance company because insurers are asking employees—who are used to doing things in a certain way—to do things differently.
“They probably do [their jobs] pretty well, but you upset the apple cart when you say they need to look at this world differently, particularly when there is a potential to automate tasks that haven’t been done that way before,” he says. “A lot of work is still done manually even in big insurance companies. Throw all those points in a bucket and shake things up and maybe that’s what’s holding some people back.”
Premier Prizm is a medical management company that provides medical administrative services for insurance carriers. The company houses data from industry-leading companies and Diekmann reports Premier Prizm’s wants to pull that data to analyze and identify emerging trends.
What initially led Premier Prizm to look at analytics tools was the detailed work involved in a report Premier Prizm presents annually to its clients that involves what Diekmann calls “a significant amount of our IT resources internally to program different, specific queries to get the data.”
Diekmann reports Premier Prizm sought a solution that reduced the time and involvement of the company’s IT department and put the power with the business users.
“We had a choice: program it internally or look for an outside solution,” says Diekmann. “We pitted ourselves as a competitor against analytics products and when we did the numbers we found we couldn’t come close to the number of features and the capital expenditure to get the basics, compared to what [Innovation Group’s] Insurer Analytics offered at half the cost. It was an easy decision to go with them for the needs we have.”
Premier Prizm believes it is important to get the data questions out of IT’s hands and into those of the business users.
“Let [business users] ask the questions and get the data they want instantly,” he says.
Premier Prizm is in the process of building an internal data cube so that at any point in time the company can analyze and identify trends within the industry.
Premier Prizm first became aware of Insurer Analytics in a demo and the company was able to use the application within 15 minutes time. Business users were creating their own KPIs and seeing the possibilities of what those numbers were and what that meant to Premier Prizm’s bottom line.
“It will give you answers then and there,” he says. “You don’t have to wait for IT to put a project together, write, test, and release the code. You are talking three months before that flushes through [under the manual system]. That’s a totally different mindset.”
Premier Prizm expects to have the system in place by the middle of this year. The project plan includes requirements documents, data gathering, and building out the cube.
“It will be specific toward what our users need to know,” he says. “It will be the same application with our data within it. I would be comfortable with [business users] using the system] with a basic manual [for instruction].”
The company plans to migrate all of its reporting needs to the software over the course of the next year.
“We will be getting out of Microsoft Reporting Services into Insurer Analytics for all the reporting needs we need to do internally,” says Diekmann. “On a weekly basis every manager does about eight hours of data gathering. Once we go live we will replace those eight hours with driven reports and the managers won’t have to dig for the numbers. [The change] will remind them what is outside their normal and what they should take a look at.”
Petersmark believes analytics in the insurance industry predominantly focuses on underwriting and claims.
“If a carrier’s business strategy is more aggressive growth—new markets, new products—the reason to use advanced analytics is to help that cause,” he says.
“Think of the world of product or customization; analytics done well could go a long way in identifying new market opportunities or potentials.”
On the other hand, if a carrier is more focused on an expense ratio or expense model, Petersmark maintains that is where you see the focus on claims because insurers are able to get an early identification of what potentially could amount to fraudulent claims.
“Analytics can even stretch the claims side into reserving,” says Petersmark. “If you can identify claims better, you can do a better job of reserving so you have capital available for other investments. The mistake some carriers make is they acquire these platforms without having thought through some of the things they want to accomplish.”
Adoption does track with certain insurance products, according to Pauli.
“From a personal lines perspective, you are almost in the position of adverse selection if you don’t have analytics to support your decision processes and there is an older, manual decision-making process in place. You can get quick decisions when you are using analytics.”
Pauli believes claims is coming along in second place behind underwriting, particularly with discovering fraud. She cites the number of fraud solutions on the market as a reason for this.
“That’s kind of an intuitive use of predictive analytics,” says Pauli of fraud detection. “Most claims executives would say they need it for fraud because it is hard to find fraud hidden among all the claims data bytes. Now the carriers are looking for other purposes for its use in claims with an understanding that it can help in areas such as subrogation, litigation management, and vendor management. It definitely is a work in progress.”
For effective claims management, analytics can make that area a profit center if insurers are able to improve the life cycle of the claims, according to Ray Dowling, vice president, sales, for Innovation Group.
“If [an insurance carrier] can shave off a mere one percent, from a direct written premium aspect that goes right to the bottom line” he says.
Carriers also can track new products, their performance, and whether the products are driving profitable business. The systems also allow carriers to track profitable business lines—whether it is on renewals or other areas.
“The challenge is how to garner actionable information from the data,” Dowling says. “Often, the pain points are because the data is fragmented or siloed. How do you get that one enterprise look? You have to bring in data from disparate sources into one data repository and place the analytics on top of that.”
OLD VS. NEW
As with most technology issues, legacy systems continue to be a stumbling block, points out Pauli.
“There’s been a perception over the last 10 years that unless you have your data in perfect order—and we know there’s no such thing as that—you can’t use analytics,” she says. “That absolutely is not the case.”
Carriers find themselves in an evolutionary process with their core systems and data initiatives, explains Pauli. The good providers of analytics have the capability to bring industry standard data to insurers while the carriers are seeking to create some sense out of their own data so it still provides value.
“It gets you down the road, provides some support, and moves you further along with decisioning and customer issues,” she says. “We know carriers that are wrapping their old legacy systems with some integration technology so they can use analytics sooner than they can accomplish a core system replacement.”
Pauli points out some analytics vendors have created partnerships with policy administration vendors to provide carriers with an analytics capability.
“Most of the vendors that are really good at core systems know that analytics isn’t their bread and butter,” she says. “[Policy administration vendors] have created an open architecture with the capability to integrate their systems with an analytics product.”
That’s another example of how technology is pacing at this point in time, according to Pauli.
“A lot of the pure data players are now getting predictive analytics capabilities,” she says. “Most technology providers understand that analytics—particularly predictive analytics—is where the industry is going and where it needs to be so there definitely is maturity. TowerGroup sees data-driven decisioning as a strategic initiative. We are seeing the leading carriers treating it that way.”
Petersmark feels another issue facing carriers is the expertise of the analysts who use the software.
“The software vendors have done a nice job of hiring smart actuaries and developers who have this stuff down cold,” says Petersmark. “Insurers don’t have a lot of that yet. You may find a person or two in the actuarial department. We’re certainly aware of carriers that have bought [analytics] solutions, but some of them are still struggling to get some value out of the software.”
Petersmark brings up the old axiom among insurance companies: They all do the same things, just in slightly different ways.
Insurers can hire actuaries or a data management administrator, but the real issue is the data is a little bit different.
“Based on the edits in the systems they’ve used for years, data becomes specific to that carrier and even more importantly it is usually in specific repositories,” he says.
“Who knows the most about your data? Even with big companies there isn’t always a good answer to that,” says Petersmark. “People may know certain things, but that isn’t always valuable enough to pull information.”
Petersmark believes there is a “sweet spot” with institutional data and those who can translate that data for analytics platforms allow business users to do what they need to do with it.
“That’s not a readily available skill set,” he says.
The pain points within any insurance organization are different among competitors; therefore the analysis of data is different. Priorities also can change quickly, according to Denise Garth, senior vice president strategic marketing and industry relations for Innovation Group.
“Last year, low interest rates and challenges of the soft market performance were big issues because combined ratios were high,” she says. “In the spring, we started getting hit by catastrophes and all of a sudden claims became a top priority. On a dime, priorities can change because of events that are going on and the dynamics in the marketplace. Having a solution that has an insurance specific data model and the capability of handling policy and claims gives you the ability to quickly get new reports out and respond to the marketplace.”
If carriers are not careful about what they want to accomplish, Petersmark feels the whole purpose of an analytics project can be defeated.
“You might bulk up the staff along with spending plenty on the platforms and then find yourself a year later realizing you didn’t get much out of it,” he says. “That’s when the tough questions start coming. A lot of companies end up creating a small sandbox. Give [the tools] to the actuaries to start and then see if it gains traction and build it out from there.”