How about a short walk down memory lane? Remember in the early '90s, when client/server versions of decision-support tools moved data analysis from mainframes to PCs? Skilled analysts could use these tools to query databases and synthesise summarised information from multiple factors or dimensions.
Within the past two years, a familiar trend has developed: Almost all of these original data-analysis products became available in Web-enabled versions, and they've been joined by newer products built to be deployed on the Web. Products from Seagate Software, Hyperion Solutions and Cognos have been joined by newcomers like Broadbase Software to aid statistical analysts who select the dimensions and determine which reports and graphs will deliver the most meaningful information. But the evolution hasn't stopped there.
Now that Web-enabled versions of data-analysis products are available, there is also a growing contingent of do-it-yourself business users who prefer to point their browsers at data to turn it into the exact "slices" of information that they need in order to make business decisions. "We've found that any business user with a proficiency in Microsoft Excel can learn to formulate queries using Hyperion Essbase," says Jack Heilpern, director of cost and profitability analysis at cargo container shipper Sea-Land Service. "Regional managers can formulate their own queries to learn exactly how much it's costing them to move empty containers from one location to another."
Wider Distribution of Data
Another change that Web-enabled data-analysis brings: It's now easier to send the results of analysis to a broader audience. People with interest in the information generated from analytical applications are on the distribution lists to receive graphs and reports. Aetna US Healthcare's US Quality Algorithms affiliates, use a tool called Managed Care Monitor (MCM) that they built on Seagate's Crystal and Holos products to track 65 acute and chronic diseases. The HMO uses the information gleaned through analysis of data cubes generated in Holos to help the physicians deliver preventive medicine and thereby minimise the need for acute care of patients.
If, for example, you are a physician treating diabetes patients who are insured by an Aetna HMO, you might get a targeted monthly mailing reminding you to send your patients for an eye examination. Or you might receive a roster of your Aetna HMO patients over 65 who have not yet received a pneumonia vaccine. This information, summarised across all combinations of 23 business dimensions including clinical and demographic data linked to more than 400,000 health-care providers, comes from a multiterabyte data warehouse running on a cluster of massively parallel IBM SP2 mainframes that captures information about Aetna's 21 million HMO subscribers.
From Standalone to Web Access
The Web-enablement of analytical applications has made it easier to share the results of multidimensional analysis with everyone from the CEO to the marketing department. Ernex Marketing Technologies of Vancouver, Canada, uses Cognos PowerPlay to find trends in marketing data. The company shares the results of its analysis with its clients so that they can gauge the effectiveness of their marketing efforts. When Ernex began to use PowerPlay in 1997, it initially offered clients a licensed copy of the software and CD-ROMs containing monthly data updates. The marketer quickly switched to a Web implementation for client access, however, because it found that its clients' IS departments were reluctant to support any software, including PowerPlay, without it being internally certified and approved for use. "We encountered a 'not on our network' reaction when we offered a standalone application. We would have had to supply a standalone workstation in order for IS to allow them to run it," says Malcolm Fowler, Ernex's vice president of business development.
Now that Ernex can provide its clients with access to PowerPlay via Cognos Impromptu Web Reports (IWR), its clients get access to real-time data and their IS departments are no longer concerned about supporting the application. "One remaining concern is that some clients are dealing with firewall and bandwidth problems, both of which can slow Web access. We feel that by next year they will have installed higher-speed interfaces that should solve this problem," says Fowler.
Some companies that used an earlier generation of multidimensional analysis tools are now able to get better data after having updated their analytical applications. Sea-Land Service, which recently sold its international shipping segment to Danish company Maersk, is no newcomer to using analytical applications for costing. "We started slicing and dicing our business about five years ago [to keep track of costs]. Now we determine profit and loss for every load we move," says Heilpern. Sea-Land found that the costing models it had developed were accurate but hard to use. Now its financial analysts use ABC Technologies 's Oros software to do activity-based costing and analyze the results using Hyperion Essbase with a Microsoft Excel front end. Essbase has also become Sea-Land's primary interface to Oracle's Oracle General Ledger.
After deploying to its financial analysts, Sea-Land implemented Hyperion's Wired for its OLAP module to create a simplified interface of pull-down menus and buttons for business users. This Web-enabled interface has made it easier for business users at Sea-Land to choose their dimensions and then view their data in graphical form. Instead of ordering reports from the IS department, business decision-makers study graphical views of selected dimensions and use this information to assign accurate unit costs to each load on every trade route. "As we get better data, we are able to pinpoint costs much more accurately and then act upon that information. For instance, after the activity-based costing model showed us that 20 per cent of our revenue went to positioning empty loads, we took measures to compensate, including giving up some [dock] slots that were not profitable," says Heilpern.
Integration with Core Applications
Web-based analytic applications are also becoming more tightly integrated with core business applications than were the standalone client/server versions of analytic applications. Integration has extended beyond enterprise resource planning (ERP) applications to encompass customer relationship management applications. For example, Rockwell Automation of Cleveland integrated Clarify's ClearSupport and other tools with the E-Business Performance Management (EPM) application package from Broadbase Software in order to monitor customer satisfaction on a per-customer and per-region basis in its 14 North American call centres. Rockwell captures data from its call centres from various "touch points" including phone calls, e-mails from technical support people and customer-generated Clarify trouble tickets.
The integration of Clarify with EPM has allowed Rockwell Automation to make use of Clarify data in ways that would not have been possible with straight reporting tools. "We have been able to do management analysis of our call centres to see which businesses are generating the most calls. EPM also helps us do scheduling based upon the actual times of day when call centres are operating at their peak," says David Handley, manager of support systems.
Although an increasing number of analytic applications are integrated with core enterprise applications, Dan Sholler, a senior program director in the Application Delivery Strategies service at Meta Group, cautions IT managers not to "expect a single common architecture for analytical applications. Right now it's more realistic to expect to need multiple sets of infrastructure, especially if the enterprise needs specialised tools such as reservation pricing analysis," he says.
Though some companies like Aetna have analytical applications that use data warehouses with billions of rows of data, there is also a trend toward departmental solutions with small data marts. At Charles Schwab Institutional, a business unit of Charles Schwab, MIS Director Sergey Muza has implemented a comprehensive business intelligence application using Seagate's Holos. This application, called Service to Investment Managers Plus (SIM+), can look at summarised data from groups of advisers, drill down one level to see individual financial advisers or analyse all the way down to the customer level. SIM+ helps Schwab assess the profitability of each group of accounts and use the information to determine the optimal commission schedule for a specific institution by analysing the trading habits of many of its customers. The associated data mart for SIM+ has a terabyte of data about Charles Schwab Institutional's 800,000 accounts, grouped by business segment. "SIM+ provides business managers with a way to forecast how their relationship with particular clients will change if the commission structure changes," says Muza.
Even though you've used these kinds of analytical applications before, the latest evolution will reveal information that may manifest itself in unusual ways. This information can change the way a company views its suppliers, its competitors-and even its own sales force. Sometimes fear of change-in this case, in the increased availability of information-is a greater barrier than any technical challenge. With nearly 80 internal business users already accessing information via Wired for OLAP, it would be an easy matter for Sea-Land to let customers know why the company is now sending some of its business to other shippers rather than continuing unprofitable routes. "We had some intense discussions about sharing some of the data with customers. At this point we have not made this information available. Some managers are terrified that members of our sales force will overreact to our profitability findings and leave us for a competitor who now services the routes that used to be part of their territory," says Heilpern.
The key, then, is to use these latest data-analysis tools wisely-to reveal not only where business might be contracting but also where there's the potential for expansion. These are tools for making information available, not for hiding it. Better to use them to paint a complete picture than to hoard data and risk incorrect interpretations.