A Bigger Bang

Hard work often brings rewards, and those insurance carriers implementing business intelligence (BI) projects are finding they're in a good position to affirm this statement. Business intelligence isn't easy because of the sheer amount of data carriers have collected over time and because that data never has been placed in a single repository. Claims data has had its own silo as did marketing data and customer service data. Integrating that information has been a challenge. The benefits, though, make the work worthwhile. Both decision makers and business users can leverage their company's data to make more intelligent business decisions.

United Heritage Life (UHL) discovered early on in its BI project there were problems with its data. The carrier needed to improve the integrity of its data as well as enable business users to work with the data by adding codes that could be extracted from the data warehouse. “Once we went through that first project, we understood what it was going to take to have the right data and the integrity we needed within the cube,” says Mick Ware, vice president of IT. From that point on, whenever the carrier made system changes or anticipated building new cubes, UHL made sure the data was cleansed and had the appropriate information in it.

“Getting data in line is part of why you don't roll out a new cube each week,” indicates Ware. “It takes some design work upfront on the cube and then back to your data to make sure [the data] has the integrity you need.”

Carriers need to understand the specific architecture of their business, explains Marilyn Martin, a director with Collaborative Consulting, and they must answer several questions such as: How do carriers use data in their operations? What do their operations look like? Where is data being produced, consumed, and adjusted throughout the process? Why do users need it during that process? Where does the business have redundancy around data that is causing disconnects?

Insurers must be aware of how they use business information on a day-to-day basis, Martin adds. This includes knowing the operational components of the processes of the organization itself as well as the entities involved in collecting and using the information–including external entities that could be producers and consumers of the same information. “In that context, [carriers] need to understand how information flows and then step back and understand how they need to analyze it or do predictive analysis on it,” she says.

With that grasp of intelligence requirements, carriers can put together an enterprise data strategy, Martin asserts. “That's the missing piece for a lot of companies,” she says. “They don't have any basis for putting together an overall strategy for their information and how to manage it. There is a clear lack of accountability for quality of information. Very rarely are there well-understood governance processes identified. If you don't have accountability for quality, how can you ensure quality is going to occur?”

At this point, QualChoice of Arkansas seems to have the quality issue under control. QualChoice is not having any major issues with data integration, according to Mike Stock, the carrier's COO and CFO. “The vast majority of our data comes from our primary transaction system we use to pay and manage all of our claims, and we are very familiar with this system,” he says. “Other data not from this system still comes to us in industry standard formats we work with often.” Stock reports most data transmissions in the health insurance industry comply with various industry standard file formats, so QualChoice has not run into problems. “As we add more of our operation data and begin to pull in information from our telecommunication system, we may encounter issues, but that is further down the road,” he says.

Like many companies, Aetna has been performing data mining and analysis for years but without the benefit of a BI technology solution. In February 2005, the company selected Enkata and has been using the vendor's BI tool since the middle of last year.

The tool is used to support claims handling and customer service. “Within that, our mission is to provide differentiated, superior-quality service,” says Frank Abetti, head of the business intelligence area for national customer operations at Aetna. “What my area does in particular is to make sure we have the right customer-focused metrics in place.”

Aetna was fortunate to have a file with literally hundreds of claim attributes, some of which were from “upstream” systems, according to Abetti, which meant some of the silos already had been broken down. As an example, Abetti explains, claim elements would contain data from current procedural terminology (CPT) code, a modifier, DX code, and place of service. He adds an example of elements from upstream systems would be provider type, provider specialty, and provider number, all from Aetna's enterprise provider database.

“Not all attributes we were interested in were present on the master claim file used by the Enkata application,” says Abetti. Examples include provider contract information, specific plan information, and member eligibility information. “In order to supplement the claim attributes, we are in the process of appending the desired elements to the master file used by Enkata, which requires IT support,” he says. “The issues we have had to deal with include securing funding for IT support and getting subject matter expertise to map non-claim-specific attributes (e.g., contract and plan attributes) to specific claims.”

The issues surrounding data become more complex with time. Carriers are seeking new distribution and support channels through affinity relationships with agents and third-party providers, which creates more opportunities to acquire data, according to Martin. “[Carriers] are looking for ways to differentiate themselves, so they are focused more on how they can put those [relationships] into play quickly, integrate those perspectives into the day-to-day process, and not step back and say, 'What's wrong with my infrastructure?'” she says.

Martin doesn't believe the problems with data involve a company's technical infrastructure. “I'm talking about business infrastructure in terms of whether [the company] is structured correctly to be able to change quickly and to be able to access the information that will tell [carriers] whether they are performing the way they thought they were going to,” she continues.

Insurers have no way to assess effectively these new channels beforehand. The channels also expect more intelligence, which puts a lot of pressure on insurers. “Some companies really are scrambling to try to satisfy those needs,” says Martin. “I've seen veneers put on where [the company] might be able to present a good story from an information standpoint, but under the covers, companies spend a lot of cycles of manpower cobbling together information and hoping it is accurate enough.” She points out this activity is not peculiar to the insurance industry but found in other industries, as well.

Jim Lazarz, director of sale compensation for CUNA Mutual, believes insurers need to be careful any data entered into the system is valid and comes from solid sources. “We get feeds from about 20 different business systems around our company that ultimately go into the comp processing,” he says. “On the one hand, we would like to think it's all clean, reliable data. However, the real world is we in compensation always think we're at the bottom of the river, and whoever throws garbage in upstream, we have to clean it up. Overall I'd say our [Callidus] systems here at CUNA Mutual are pretty reliable”

There are a number of vendors that market out-of-the-box tools they sell as a data warehouse-type product for companies in the healthcare payer industry, according to Stock. Such tools capture claims, membership, and provider database information and put it in a relational database. “They have some standard reporting packages they put together and some drill-down capabilities to analyze the claims experience and why your costs are doing what they are doing,” he notes.

One such tool examines what Stock calls episodic treatment groups. Claims data for an insured individual is studied from the first claim that would trigger an episode of illness, and the tool tracks and adds all the claims together until a period of time after the last claim is incurred that would relate to that illness. “It groups all that cost data together to look at what was the true cost of that episode of illness and how efficiently it might have been managed by whoever the providers were–the physicians or the facilities involved in the rendering of the care,” he says.

QualChoice selected a tool to use in that arena called MedAI. “We'll use it for our episodic treatment grouping and analysis and for physician profiling so we can compare how efficient Physician A is with Physician B for the same type of illness,” Stock explains. One of the best features of the tool is the predictive modeling capability, he asserts, claiming the tool can take multiple years' worth of claims data about an insured population and, based upon historical claims, project what future claims cost will be. “That way you can target those people who are projected to generate the most medical costs and work with them to help manage their care and to hold those costs down as low as possible,” he says.

While the MedAI tool is great for the medical side, Stock believes it didn't address QualChoice's business needs, such as the reporting the company does for employers, internal financial reporting, operational reports, and the day-to-day activities of managing the business. QualChoice ended up selecting a tool from Cognos for the business issues. “We'll be loading a lot of the same data in there, but we'll be using the claims data to do our internal financial reporting, our financial analysis, projections, and budgeting,” says Stock. “We'll be using [Cognos] for underwriting purposes to pull data and project what future revenue streams need to be for groups. We'll also use it to monitor operational statistics.”

The first business intelligence endeavor at United Heritage Life Insurance was a marketing production reporting cube, according to Ware. “We analyzed data we were able to extract from our midrange system, mostly for reporting line of business and production from different geographic areas as well as product lines,” he says. “Beyond that, we went to some mortality studies; we've got a budget cube for all our midrange managers to use for analyzing their budgets; we've got an extranet for our regional managers to use in the field that ties into our production cubes; and we've got agent financial cubes with financial balances for each of our agents for advances and leads. We started off with just marketing, and we've branched out. We've got plans this year to branch out even more into our financial systems for our accounting people.”

Carriers need to provide business intelligence to more than just the executives. “It's intelligence for the guy in the trenches, too,” Martin says. Everyone in the organization has to make intelligent decisions on a day-to-day basis. The sales rep out in the field needs intelligence to be able to qualify leads effectively. Intelligence about customer relationships is going to be a key to driving new business in the door, she adds. “Where insurers will realize the benefits is in the day-to-day operations and strengthening their third-party relationships by treating [intelligence] as an asset,” she says.

Two of Aetna's main areas for BI are what the carrier calls first claim and first contact resolution. Aetna decided to focus on these areas to provide superior service because they are two of the carrier's main outputs–a claim and handling an inquiry that is sometimes the result of a claim, according to Abetti. “My [BI] area in particular has a mission to provide the organization with the analytics and the business knowledge to drive improvements–specifically, to reduce claim rework and improve first claim and first call resolution,” he says.

Incorrect claims payments might come in the form of a reworked claim or a call to client/sponsor services, provider service teams, or one of the member segments. “We did not have a holistic way of looking at this,” says Abetti. “What we liked about Enkata was it could take our whole claim data set and differentiate between good and bad outcomes.” Bad outcomes, essentially, are any claims that came back for reprocessing.

Haley Wilson, QualChoice's CIO, be-lieves the effect the Cognos product is going to have will allow users–whether they are management staff or data analysts–to be able to spend more of their time analyzing the data as opposed to trying to define what data they need. “That's the transformation that will happen in this company as the tools give us the ability to be more analytical and strategic in our approach,” he says.

Training is important if the business side wants to get everything it needs from an intelligence project, Ware advises. Some business users are ambitious, he claims, and enjoy digging into the data and using new tools. Such power users are easy to convince to take advantage of this tool. “Then you have your other users whom you have to drag kicking and screaming and give lots of training to and lots of handholding,” he says.

Ware indicates his staff has to build reports for these users before getting them to see the value of the tool. These types of users are not going to progress quite as quickly as power users. “Even though the interface is easy to use, if they don't get the results they expected the first few times they try it, they tend to give up,” he says. “They are frustrated, and they don't know why they are not getting the right results. You really have to be a business analyst to get a lot of value out of a business intelligence tool. Some of the users who are not quite experienced and don't understand the data have a tough time figuring that out. You really have to build them some reports and some views that are correct to begin with so they have a starting point.”

The Enkata system takes the Aetna data, distinguishes between the good and the bad outcomes, and then allows business users to drill into those claims to conduct an attribute analysis, says Abetti. The system pulls data from throughout the Aetna enterprise, including claims systems, provider data, plan sponsor data, and Aetna's own plan data, he notes. “If you look at the bad claims, you can pinpoint combinations of attributes that are driving higher-than-average rework rates,” he says. Aetna is able to detect common causes for the claims mistakes, and this gives the carrier what Abetti calls a targeted sample. “This really helps us target our analysis, and that's more actionable because you often come up with one specific cause driving the claim rework,” he adds.

For QualChoice, the tool is going to have two broad uses–one area is there will be a defined set of parameters, reports, and measurements to create report cards that always will be kept current with the most current data, Stock explains. “Management will be able to review and monitor to make sure all the different aspects are functioning the way they should be,” he says. “If aberrations appear, the tool will give us the ability to drill in quickly, figure out what is causing the aberration and what we need to do to fix it, and get back on track.”

The second use will allow QualChoice to examine situations that arise where there are one-off projects in which the carrier has to handle an issue in the marketplace or a regulatory concern and must get into the data to do a special analysis to assess what is needed to address the business requirement. “Right now, whenever we have those situations, we have to pull in the IT people, explain the situation to them, and turn them loose on the data to try to pull it together for us,” says Stock. “This will put that data in the hands of the business people, so we can get quicker resolutions by doing some of that analysis on our own without relying on the IT people to get to the data.”

Aetna began using its system with one state–albeit its biggest one, Texas–and right now there is 15 months of Texas data loaded into the system. “We have engaged Enkata to bring the whole country into the system, and we are going to rely on it for the proper hardware to have the queries run quickly,” says Abetti. “It's primarily a server issue, so it's something that can be addressed. I'm not anticipating a problem because it has worked very well with the Texas data.”

The IT department has many initiatives to deal with other than just business intelligence, Wilson states. Being able to have the resources available to implement those initiatives on a timely basis and within the expectations set was a major challenge when the IT department received business intelligence requests from the business side. “Having those resources free to do [other] initiatives is going to be very helpful,” he says.

The biggest and easiest benefit to quantify from UHL's BI project is the reduced time IT has spent working on reports and analyzing the data, according to Ware. Users can take the ProClarity tool and don't have to involve IT at all once the cubes are built and the data is put in place, he remarks. “Prior to having this tool, [IT] was running queries, building programs, and trying to extract information for [business users],” he says. “[Users] didn't know how to tell us what they really wanted. It was an iterative process, and IT spent a lot of time trying to solve those problems.”

From Martin's perspective, the insurance industry tends to lag behind other financial organizations in terms of the adoption of technology. “That's not surprising being a risk-averse industry,” she says. “[Insurers] are going to be behind the curve.” She adds large carriers have a real challenge. “While we are seeing their technical architecture is maturing, they haven't taken the next step of understanding how technology is an integrated functional area, just like any other functional area of their business,” she says.

This brings about difficult issues, particularly for property/casualty insurers, Martin continues. With a soft market today, she feels carriers are struggling with where to make their investments internally. Do they invest in a data strategy or a more aggressive marketing strategy? Where do investment dollars go? Rather than spending the dollars on strategic initiatives, carriers are pressured to put in a tactical fix. “There is this pressure to fix things quickly and to focus the investments there,” says Martin. “Technology departments seem to be evolving from the technical dimension.” She believes IT organizations need to think from a strategic business standpoint–having sophisticated business analysts, not just systems analysts–that can understand the business architecture and how that leads to better systems. “[IT departments] don't have in-house the capacity to understand business architecture effectively,” says Martin.

Insurers are clamoring for business intelligence, but in Martin's opinion, carriers don't have the infrastructure established to be able to deliver the intelligence they know they need. Carriers still are using antiquated and manual mechanisms to get their intelligence, she contends, because of the state of their data and the availability of their raw data. “Insurers have built up silos over time, and data was collected from a silo perspective,” she says. “Depending on the organization, its data strategy may be nonexistent, or it may be in progress but not completely realized yet.”

BI will become a driver for data proj-ects when insurance companies realize the value information provides not only to themselves but to their partners, Martin asserts. Information is what's going to be the key to relationships because carriers need to look for ways to differentiate themselves. “Intelligence is going to be the key that will allow them to differentiate themselves,” she says. “To be able not only to provide information as an asset to their partners but also to demonstrate to their partners the value of the business relationship.”