April 20, 2005

Business intelligence, or BI, is a vague term that packs a lot of promise. It's right up there with "dream job," "in shape" and "perfect relationship" - few have it; most want it. Business intelligence comes with varying degrees of difficulty and always with much consideration, and no two BI solutions are exactly alike. BI can mean anything from the way data is delivered to the analytics tools that extract knowledge out of raw information. But what makes business intelligence "intelligent" is ensuring that it will generate better business through ease of use and competitive advantage.

Key technologies that are encompassed by BI basically are anything that has to do with moving data. The important part is the connectivity among technologies - the way data is transferred throughout an organization. Data must flow smoothly in a way that employees can understand; it has to give employees the ability to make intelligent decisions.

"BI is a question of putting different tools together and assuring that they work together and integrate properly," explains Michael Azoff, senior research analyst for the Butler Group in Hull, U.K. "An organization has to address the key issues of what it already has, where it is now and to where it wants to move. That should dictate what kind of solution to implement," Azoff continues.

Not only do insurers need to know what they already have, but they need to figure out just what it is they want to accomplish in order to guarantee the synergy among business units and employees that intelligent data can provide. According to Anthony Abbattista, vice president of enterprise technology strategy at Allstate (Northbrook, Ill.; $134 billion in assets), BI is a series of steps that comprise information delivery. "I think there are a lot of words in the marketplace hovering around business intelligence right now. Mostly, we think of BI vendors as being providers of the front-end or end-user tools that allow end users to do work, but I think the bigger view we should have is that BI is everything from sourcing data at the beginning and having it be rationalized and well understood among both the business and the technology clientele to a whole series of structured analytics and reporting," Abbattista explains. "At Allstate, we call BI 'information delivery' because the starting place is with good data and good information. Then you spend a whole lot less time questioning the results."

Although quality data - and lots of it - often is at the foundation of an insurance company, it sometimes can cause information overload, slowing workflow, frustrating employees and impeding a high-level picture of the business. "What we found was that although the data is always there - meaning it is always in the organization - it is the visibility and the ease of access that makes data work," reports Gary Coles, senior vice president of administration at National Life (Toronto; $5 billion in total assets for 2003). According to Coles, what BI vendors have to offer is clarity. "Everyone has huge databases, but you need to be able to get to the data quickly and easily and that is what makes some of these software packages valuable."

The main parameters for a successful BI technology solution, then, are where a company's data is located, how it is stored and how to get it out again. At Allstate, Abbattista is working with 13 data marts, or data warehouses, and what he describes as, "hundreds if not thousands of separate collections of data that people have at their desks - databases that people came up with themselves." Over time, many insurance companies have been harboring systems that grew in small colonies. "For insurance companies, the temptation has always been, 'How do I draw from the database just what I need for my job?'" Abbattista suggests. That way of thinking often leads to numerous legacy environments and the need for employees in different lines of business to cross-check each other's work.

Moving in Together: an IT Ecosystem

Besides wasting employees' time, uncoordinated, disparate systems have given rise to another looming problem over the years: tech support. National Life was running an entrenched executive information system that was implemented in the mid-1990s and written in the Oracle (Redwood Shores, Calif.) Express OLAP (Online Analytic Processing) environment with data stored in a proprietary cube. According to National Life's Coles, data was aggregated using proprietary import methods that were only accessible by the Express tool set. Over time, the limitations of the system became clear.

"The client/server installation limited rollout to the audience and was difficult to support and maintain," says Coles. "The system was not flexible enough to keep up with new requirements such as instant customized views of broker data in a variety of formats," he continues. To overcome this challenge, National Life installed a Web-based solution called WebFOCUS from New York-based Information Builders in January 2004. The solution establishes a relational database that can be accessed by a thin-client reporting tool, enabling complex query and analysis capabilities by users and flexible online reporting capabilities.

Similarly, Great American Insurance is working to transform a collection of siloed mainframe systems that have been developed over the past 25 years. Dan Daly, manager of the information management department at Cincinnati-based Great American, a subsidiary of the American Financial Group (2004 net earnings of $359.9 million), explains that in order to shut off production of large mainframe reports and employ an updated, easy-to-access data store that the company calls Claims Data Warehouse (CDW), the carrier has had to build a series of databases incorporating the original mainframe components that are connected to Cobalt (Sun Microsystems; Santa Clara, Calif.) programs, which gather the information from all of the mainframe systems and organize them in an IBM (Armonk, N.Y.) DB2 database that acts as a staging environment. The organized data then is transferred into an Oracle database. Finally, Great American connected to a Cognos (Ottawa, Ontario) database on top of the Oracle database to allow end users to generate reports on all of the data in a timely manner and in an effective format over a local Internet server. "Now that we have a way to integrate data from systems that were never meant to talk to each other, Cognos allows us to produce reports that were impossible or extremely difficult to construct previously," explains Daly.

"The question is, How do you take a technology solution and layer it into that mainframe system? And how do you architect your way out of that mess?" says Allstate's Abbattista. For Allstate, the answer was consolidating its extensive technical structure. Now, Allstate runs a dual environment that was deployed in mid-2003 by creating an Information Services Group - made up of IT and analytics employees who normally source data needed for reports to the employees that request it - that was put together by Abbattista for the purpose of identifying redundant products and systems. Abbattista then chose vendors to support two main databases: a financial data warehouse built on SAP (Waldorf, Germany) business warehouse technology for financial reporting data and what he calls the Allstate Data Warehouse, or ADW - a large Oracle-based custom data warehouse running on Unix technology for operational and performance metrics data.

"Pulling all of the information together helps us from a control aspect," relates Abbattista. "We can pull disjointed information together in one place and run reports across the entire organization."

Such enterprisewide visibility is the goal of creating a unified environment by tying together disparate systems. "Many times, a business customer will need information that is scattered over five different databases just to keep up with the speed of business today, which has to be 50 times what it was 20 years ago," says Great American Insurance's Daly. "So a BI system allows you to answer questions on a daily basis that your competitors are not able to answer - it is basically surviving."

For Great American Insurance to survive, it runs approximately 10,000 reports in its Cognos environment every month. In 2001, the insurer implemented Cognos release 6, which Web-enabled the insurer's CDW database. In April 2003, Great American upgraded to version 7 and Cognos functionality was extended beyond claims to a policy data warehouse and given a separate processing database. "We have got this great infrastructure that allows us to point to virtually any database source in the organization and present some of that data as an attractive report via the Web," says Daly.

Rules of the House: Business Culture

At Allstate, "We are making this massive shift from an IT shop that builds reports and gives people the data that they request to collaborating with all of our business users to say, 'Here are some tools in a toolbox - let's build something together,'" says Abbattista. Allstate uses the Crystal Reports platform from Business Objects (San Jose, Calif.) to run end-user tools such as regression, analytics and reporting capabilities on top of the ADW database.

Allstate also maintains a mirror of its data warehouse in a third database, dubbed Decision Research Analytics Environment, or DRAE. The analytics database allows Allstate to extract data either from external parties or on an ad hoc basis, according to Abbattista. "The Information Services Group and I recognized early on that unless we could provide a good 'what if' environment, we would never make the data system attractive enough to convince our workers that it was better than having individual large workstations, their own disk drives and analytic environments," he says. "By consolidating we are giving them something better than that - a way to use more information competitively."

What is becoming more visible as BI solutions are gaining maturity in the market is the soft benefits they can provide. "I must admit, it goes to the soft benefits when looking at ROI, but they are huge," relates National Life's Coles. "BI gives you some clarity as to what's happening across your organization as well as consistency of reporting, and it lets the entire organization increase accountability that eventually creates results in increased sales," he continues.

Accountability and governance are two of the biggest soft benefits insurers that implement BI can realize. "You simply cannot operate now without BI if you want to have effective governance," says the Butler Group's Azoff. "You need to know very rapidly what's happening with your organization, so at the board level you need to have information that is very timely and very accurate," he adds.

Happily Ever IT

The technology objective behind BI isn't really technological at all. The goal is to make things easier for employees from the bottom up. If it is easy for low-level employees to enter and report on data, it is easy for high-level employees to know what is happening and make business decisions accordingly.

"BI harbors no specific technology objective outside of the obvious: trying to make information more readily available and provide push-and-pull reporting services by consolidating data presentation into a single portal," asserts Great American Insurance's Daly. "But who hasn't been trying to do that for 10 years?" he asks.

Allstate's Abbattista finds ROI in the value of data interaction and governance as well. "I don't think the technologies themselves give you an ROI. It's how you put them together with other technologies and what kind of discipline you have in your environment," he says. According to Abbattista, what's really important for insurance companies and financial services firms are the building blocks a BI solution provides. "For those of us that can use value-added data to put really good building blocks together and really good products together, the real value is in the connective tissue," he asserts.

Another way of looking at ROI is through the users. According to Daly, when Great American Insurance began its CDW project with Cognos, the carrier had no more than five people using the technology. Now, the company has about 2,300 users, including about 800 agents and customers. "There is obvious value in that we are able to use our BI infrastructure as a marketing tool in working with agents and customers," says Daly.

Some ROI may even be unintended. Although the solution was purchased mainly to ease data import for employees, National Life's Coles has found that his company's marketing offices across the country are using the WebFOCUS tool as the base for sales and marketing discussions at all meetings. "And, although there were many contributing factors, National Life's sales were up 25 percent in 2004 from 2003," he relates.

Even though the objectives behind BI may be strictly business, according to Allstate's Abbattista, it is still important to choose the right tools. "When you pick good tools - if you pick top-tier tools - they should be a part of you getting to that equation faster and better, but I don't think one particular tool in the top-tier space gives an advantage over other vendors in that space," he says. "For the sake of competitive advantage, we have to do something differently and part of that is making sure we pick pieces and parts and combine them in a way that is better than something we can just buy from somebody off the shelf."

This attitude is reflected in the behavior of vendors as well, according to Butler Group's Azoff. The information technology market has become a harder sell because customers are becoming increasingly demanding and looking for tools that will interoperate. "Gone are the days of vendors just protecting their territory and having highly proprietary systems," says Azoff. "I think standards are common and those days are gone."