The metadata standards that streamline company-to-company communication can do double duty as prototypes for internal data management within a single insurance carrier."If it has already been invented in the industry, why should we think that we're going to be able to invent our own that's better?" reasons Rich Maynard, an enterprise architect for The Hartford Financial Services Group, Hartford, Conn. He's one of many insurance executives who view industry metadata standards that way.
Metadata, or "data about data," works something like a dictionary that defines data or like the index in a book that tells users where to find data, says Gary Knoble, who worked for The Hartford for 30 years before leaving a year ago in January to work as a consultant for Bao Rong in China. He's been a founder and active member of data-oriented trade groups, and he agrees that metadata standards designed for external use can serve as examples for handling data within a company.
"If I were the CIO, I'd say start with ACORD," Knoble says, referring to a widely used set of industry standards. "If they don't satisfy a need, then modify them." Observers agree that data and business processes vary so much among insurers that standards nearly always need some tweaking to work internally.
In a pilot project, ACORD standards fit 90% of the data at Erie Indemnity Co., Erie, Pa., says Jim Viveralli, a database administrator there. "The rest of them, we had to make up because they're Erie-specific," he says. His service on ACORD's Business Dictionary Validation Working Group helps him keep him in touch with others who are working with metadata, he notes.
That sort of contact can help carriers navigate the metadata scene but can't replace the need for familiarity with company metadata, says Mike Freel, a bureau statistics manager for EMC Insurance Group Inc., Des Moines, Iowa.
"We remain cognizant of all of these industry standards, yet we make sure that if the need arises for us to have our own internal standard because of specific business practices then we would do that [internally]," says Freel.
Juggling external standards and internal needs can prove challenging because most carriers face a "mishmash" of half-forgotten homegrown legacy metadata standards, says Knoble. Departments within a single company often use different names for the same pieces of data or use differing rules for processing the same data. Data from outside the company hasn't been subject to company rules.
"You may be using data that's in a six-digit field in one place and in a 10-digit field in another," Knoble says by way of example. "In one instance it has a decimal point-in another it does not. Dates may be in different formats." Drivers might be listed by age in one place and by date of birth in another.
One part of the company may say a claim file is closed when the final payment is made, while another may consider the claim closed when the last bit of activity ceases, notes Freel of EMC. Departments can have trouble agreeing on something as simple as a policy number, according to Maynard of The Hartford.
With separate entities within a company generating their own business and technical terms, one carrier found itself with 3,500 terms to standardize and define, a source at the company says.
Meanwhile, the volume of data that insurance companies process is growing rapidly. "The typical large insurance company is getting data from an unbelievable number of sources," says Knoble.
"There's a great, great dependency now on using third-party or external data sources for business use within a company or in the industry across companies," says Pete Marotta, enterprise data administrator and principle in charge of data management consulting for ISO, the Insurance Services Office, Jersey City, N.J.
Maynard notes that some third-party data sources have been supplying information for a long time, like motor vehicle bureaus, while many others have arisen in recent years, like catastrophe modelers.
"Not too many years ago the notion of longitude and latitude and geocoding were not thought about a lot," says Maynard. "The distance to the shore when underwriting a house for wind damage used to be a fine art." Now, insurers rely on a dramatically larger well of third-party information on geography and weather, he says.
How can insurers cope with wildly mismatched and rapidly growing volumes of data? That's where the metadata comes in again. Besides defining and locating data, metadata can automatically convert seemingly disparate data into the same format, allowing for meaningful combinations of data, says Knoble.
Yet most insurance companies lack the prerequisite for using metadata to manage data, adds Knoble. They have yet to develop a planned, coherent enterprise data strategy, he says.
"It's shocking how few companies actually think about what data they need," says Knoble. "They just collect it and say, 'We'll think about that later.'"
THE DATA STRATEGY
For Knoble, the data strategy begins with a definition of where the company wants to go-a destination that coincides with the company's business strategy. Once a company defines that business strategy, it's time to take the second step in the data strategy, which is to ask what kind of data is needed to make the business strategy work. Next comes figuring out what technology is needed to support that data.
Thinking up a data strategy and putting it in place can take three to five years, says Knoble.
To overhaul internal metadata, as part of the data strategy, carriers need to understand as much as possible of what has been handed down through the years. Sometimes that means trying to ferret out the meaning of programming created by developers who have long since left the company, says Erie's Viveralli.
"That's where the costs come in," Viveralli continues. "Do you want to dig into the program code and try to unravel nuances about each field-or just draw a line in the sand and say, 'We're going to go from this point forward.'"
The money to replace whole systems probably won't be available, warns Freel. "Realistically, it's very difficult to go to a senior vice president and say these two 25-year-old legacy systems have different standards so we need to drop both of them. That's not going to fly because replacement would be millions of dollars."
Dollars aside, the success of a data strategy can ride on choosing the right person to take ownership of the project. The consensus among industry players interviewed by INN was that the person responsible for data should come from the business side of the company, probably from the ranks of the actuaries.
Once someone assumes that mantle of leadership, he or she can begin to convene committees and working groups to tackle the work.
People from the business side should determine what data to create, store and manipulate, industry observers agree, while the IT takes on the task of making those business decisions come true.
People on the business side need to bear in mind, however, that IT cannot work miracles, says Pat Saporito, director, insurance solutions, in the Cliffside Park, N.J., office of Business Objects SA.
"One of the mistakes is to think technology is going to solve a business issue," Saporito says. "Technology is an enabler."
At any rate, everyone involved in making the data strategy work can benefit from exposure to peers at meetings, conferences, trade shows and membership in industry groups that include Pearl River, N.Y.-based ACORD; IDMA, the New York-based Insurance Data Management Association; IVANS Inc., the Stamford, Conn., provider of networking solutions and software; NCCI, the National Council on Compensation Insurance Inc. in Boca Raton, Fla.; ANSI, the American National Standard Institute, Washington, D.C.; DAMA, the Data Management Association, Belleview, Wash.; and IASA, the Insurance Accounting & Systems Association, Durham, N.C.
Don't worry if the information presented at an event seems superfluous at the time, advises Freel. "It's amazing how much of it you find you need in the next year and a half."
Events and organizations also provide opportunities for networking that can benefit everyone involved, says Freel. "You just have to be willing to step forward, hold out your business card, shake hands and meet people," he says.
Third-party software tools help some carriers get control of their metadata, while others, like EMC prefer to create their own systems. "We have a larger IT staff than a number of companies our size would and we do virtually all of our development in-house," says Freel. "All of our systems directly meet our needs, and we have direct control over all the standards that are used in them."
Knoble calls software tools a tactical response to a strategic question, but software providers maintain that they work with carriers over the long term.
Still, no matter what tools are involved, nobody says developing metadata standards for internal or external use will be easy. IBM worked more than a decade to define data before giving up because of the differences among companies, says Knoble.
IBM recently shared some intellectual property with ACORD, a move that has benefited the industry in Maynard's view.
ACORD standards will continue to evolve as business needs change and the effort to define business terms continues, observers say. Meanwhile, ACORD reaches out to find common ground with other standards bodies, because insurers exchange information with everyone from auto body shops to state insurance commissioners.
"It's definitely a new era of cooperation among standards bodies," says Don Allen, ACORD senior enterprise architect. The movement has gained traction in the last 12 to 15 months, adds Cindy Maike, ACORD director of standards planning.
At the same time, bodies other than ACORD create standards for the insurance industry. ANSI remains active, coming up with standards that are complementary to ACORD's work because ANSI concentrates on health, while ACORD covers other lines, says ISO's Marotta.
Wherever the standards come from, companies with good control of metadata can break down their data into ever-smaller cells that allow increasingly closer examination of customer data. Knowing customers better creates a competitive advantage, observers say.
Observers also stress the importance of data quality-having information that's correct and reflects the real world. Data quality and metadata, they say, are the two big issues in data.
Although approaches vary in handling data quality and metadata, the basic questions have remained unchanged for years, says Knoble. "What data should I be collecting," he asks, "and how should I be collecting it?"
The importance of the answers to those questions is difficult to overestimate, Knoble says, because those answers constitute the "nuts and bolts" of managing data.
ACORD ENJOYS WIDE SUPPORT, BUT ISSUES STILL STIR DEBATE
Insurance agents started ACORD several decades ago because they were tired of filling out a different kind of form for every insurance company and wanted a common application. Since then, the Pearl River, N.Y. trade group has matured into one of the most important standards-writing bodies for the industry.
Many in the insurance business applaud ACORD's work and even use the ACORD standards as models for handling data and metadata inside their own companies.
One carrier, however, offers constructive criticism to the standards body.
Take the example of Rich Maynard, an enterprise architect for The Hartford Financial Services Group, Hartford, Conn. The ACORD standards for exchanging information electronically among companies in the insurance business should become more "prescriptive" and "less flexible,' says Maynard.
He offers banking standards as an example of a tight framework exemplified by the lightning-quick approval of credit cards. By contrast, people in the insurance business have to hire third parties to scrub and categorize data to help interpret it.
A stricter set of standards would make the insurance industry "less mysterious and a lot better-positioned for the future," Maynard says, adding that the greater consistency of narrower standards would help the industry comply with state and federal regulations.
Meanwhile, consultant Gary Knoble, who helped form data-oriented groups and is active in ACORD and other trade groups, says ACORD may have been working too far ahead of insurers recently when the body began creating international standards.
American carriers do business in other countries, he concedes, but find their operations abroad so different from their work at home that they have little interest in setting standards to bring together those far-flung operations.
Pete Marotta, enterprise data administrator and principle in charge of data management consulting for ISO, the Insurance Services Office, Jersey City, N.J., suggests that taking standardization abroad in the insurance industry amounts to a continuation of work already accomplished in finding common ground among states in this country.
Meanwhile, ACORD is ascending to next level, working for the last year on an Industry Standards Framework that will mesh with service-oriented architecture, says Cindy Maike, ACORD's director of standards planning.
ACORD also is working on a business dictionary that will probably include 7,000 terms, Maike says.
DON'T REINVENT THE MATADATA WHEEL
Insurance Carriers don't have to labor in a vacuum when adopting external metadata standards for internal use.
At Frisco, Texas-based Skywire Software, for example, products are designed to take advantage of those standards, says Stephan Fields, the company's director of architecture and technology.
"Standards provide a good starting point," says Fields, and can become part of a drive to make internal communication a product differentiator. "Your ability to navigate your own data space provides you with a competitive advantage," he says.
That assessment jibes with the thoughts of Sheila Childs, director of software product marketing for EMC Corp., a Hopkinton, Mass., software provider "We see metadata as being useful to companies as they try to manage their data," she says.
EMC sells "intelligent" products that use metadata to classify data, based on its business value, says Childs. Once the data is classified, carriers can protect it and use it to comply with government regulations and e-compliance, she adds.
Vendors and consultants also can help carriers work out a data strategy that incorporates metadata, says Tracy Spadola, senior industry consultant at Teradata, a Dayton, Ohio, division of NCR, and a board member of the Insurance Data Management Association.
"There's a need to come up with an enterprise data strategy up front that says, 'This is where we want to be when we grow up,'" says Spadola. Such strategies have gained favor in the last four years or so and continue to gain momentum though they have not yet been widely adopted, she says.
Carriers need data strategies because different ways of naming and describing data have arisen in different departments, says Spadola. The term "written premium," for example, means something slightly different in each "silo" of responsibility, she notes. A strategy brings everyone to the "same table," she adds.
For Spadola the first step in developing a strategy lies in determining where the company wants to be in terms of data management. The carrier might establish the goal of a "single source of truth" for the company, meaning that data is defined consistently and appropriately and is available across the organization.
Next comes a metadata inventory to find what's there and how it is used, says Spadola. At that point, identify gaps-what isn't captured. Prioritize, based on business needs, and make data available to business users in a repository so that they don't need to ask IT for help, she advises.
WORKING HARD TO KEEP THE METADATA CLEAN
While some insurance companies contend with the silo effect of a dozen or so poorly maintained data dictionaries, Des Moines, Iowa-based EMC Insurance Group Inc. works with just two data systems-one for personal lines and the other for commercial business.
One of EMC's philosophies, says Mike Freel, bureau statistics manager for the insurer, is based on keeping metadata consistent even after acquiring another carrier.
"We always set an aggressive time line to integrate all of their processes into our existing systems," he says of the companies EMC has acquired. That means allowing 18 to 24 months for bringing consistency to policy, claims and billing systems.
Reconciling the systems may require that the smaller company change time-honored data definitions. Once everyone understands the underlying meanings of the terms, though, problems recede.
"Sometimes you have to say to them that the ultimate reality is that you were in financial trouble, and we bought you-you have to merge into our system," says Freel. "We say that jokingly, but there is a certain amount of truth. If you allow these companies to stay on their own systems, you've just perpetuated your lack of standards."
With that kind of determination, even large carriers can streamline their metadata. Take the example of The Hartford Financial Services Group, Hartford, Conn.
"We've started to successfully merge together what we think are the best of breeds within our organizations," says Rich Maynard, an enterprise architect at The Hartford.
"The individual segments have been maturing their process over the last five to 10 years," he says. ""From the enterprise perspective, we're 12 to 18 months into that."
Just how long it may take to standardize among segments remains difficult to predict, Maynard says. The timing depends on people on the business side of the company who have to make hard decisions and reach elusive compromise on definitions of business terms for the consolidation to continue, he notes.
Already, The Hartford can achieve the "householding" function that identifies a person with multiple policies, such as homeowners and auto, for example. That recognition of multiple policies allows The Hartford to estimate the lifetime value of a customer.
Householding represents "kind of a difficult challenge when you have multiple lines and you're trying to represent all that information in a connected way," observes Rod Travers, a senior vice president at Robert E. Nolan Co., a Simsbury, Conn., management consulting firm that specializes in the insurance business.
Register or login for access to this item and much more
All Digital Insurance content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access