4 min read

SAP Dives Into A BI Market That's Far Short Of Its Potential

Companies have solved isolated problems with business intelligence, but higher-level predictive and operational analytics often take heroic effort.

Apart from the deal with SAP, Business Objects is working to tie its reports more closely into operations. At its annual user conference this week, the company will announce an upgrade to Crystal Reports, which produces reports from data. Areas of improvement include better support for XML and Adobe Flex, making it easier to create reports from operational data and act on that information. "Reporting becomes part of an operational process," says James Thomas, Business Objects' VP of marketing.

Startups such as SeeWhy Software are making a business of enabling "actionable BI." SeeWhy constantly compares incoming data with historical information and trends, flags anomalies--such as a regular customer not making a booking at the usual time--and sends alerts to customer service reps.

Oracle hasn't provided specific details on how it might integrate its ERP applications and data warehouse with its BI portfolio, which includes the performance management tools gained from the Hyperion acquisition and analytics software from Siebel. One technology to watch is Oracle Real-Time Decisions, based on software from Sigma Dynamics, a company it acquired last year. Oracle describes it as a transactional server that combines rules and predictive analytics to deliver real-time data into business processes and customer interactions.

Teradata's making its own play in operational analytics, which it markets as Active Enterprise Intelligence. CTO Stephen Brobst says the pressure's particularly high in retail, where companies are constantly tracking inventory levels and assessing short-term promotions to drive sales. While the trend toward real-time analysis gets a lot of attention, companies still comb slowly through very large data sets, such as customer segmentation analysis, to choose the right prospects for marketing campaigns. "Real time is just a small piece of it," SAS CEO Jim Goodnight said during an interview at Teradata's user conference. "Businesses not involved in heavy use of modeling and analytics will fall behind," he warns.


Medco's Halloran is typical of the high ambitions some CIOs have for their business analytics. For BI in health care, the goal is to "go very quickly from cost containment to prevention," he says.

Medco uses BI for such things as tracking pharmaceutical transactions for signs of abuse and fraud. Next it will try to develop analytics to determine, and even forecast and predict, which patients are most at risk of getting sicker. To do that, it will have to integrate prescription, lab, medical history, and demographic data to develop a "longitudinal" view of individuals' therapies, plus use clustering models to look at patients across different types of therapy. By analyzing a cluster of people with complex diabetes, Medco hopes to identify trends that can predict who among a population of patients without those complications are at risk enough to suggest an intervention. "This is how we'll practice pharmacy going forward," he says.

How would BI be brought to bear? It might go something like this: A patient previously treated for high cholesterol comes to the pharmacy for insulin. At the point of transaction, Medco could instantly run the patient's data through SAS analytical models and determine how to reclassify that patient and whether any type of intervention might be recommended. The increased use of Web services, Halloran believes, will make such links between operational systems and analytics models more feasible.

Real-time BI could have huge implications in customer service. uses a Teradata data warehouse, GoldenGate's transactional data management software, Oracle's data integrator, and Business Objects' reports and dashboards for a customer e-mail initiative that has increased sales significantly, according to Paul Longhurst, the company's director for data warehousing, who made a presentation at the Teradata conference. Overstock, which distributes more than 33 million e-mails a week in marketing campaigns, segments customers in 55 different ways in the data warehouse based on their purchase histories. The online retailer scores each customer after every purchase based on how profitable he or she is to the company, and that information is used for its e-mail blasts and other customer interactions. If a valued customer clicks on iPod accessories, for example, Overstock might send a coupon the next day.

Medco and Overstock are examples of where business intelligence is headed. They show the potential for masterfully applying BI for greater business impact, in part by embedding analytical capabilities right into everyday operations for here-and-now decision making. Given the costs and complexity, however, few companies are that far along. The BI vendors--and you can now count SAP among them--are scrambling to get it right.